MLPerf v6.0 covers 11 benchmark families: DeepSeek-R1, GPT-OSS-120B, LLaMA2-70B, LLaMA3.1-8B, LLaMA3.1-405B, Mixtral, DLRM-v3, Qwen3-VL, RGAT, WAN-2.2, and Whisper. These are the workloads running in production AI systems right now.

NVIDIA (NVDA) chips topped all 11. Not NVIDIA the company directly in every case. But every benchmark that ran was won on an NVIDIA chip. The wins that look like they belong to HPE, Nebius, ASUSTeK, Krai, and Supermicro? All NVIDIA silicon underneath.

The generational story inside NVIDIA's results is just as important. The GB300 NVL72 at 72 GPUs delivers 647,076 tokens per second on DeepSeek-R1. The GB200 NVL72 at identical scale delivers 486,141. That is a 33% lift, validated by an independent benchmark.

Investor read: NVIDIA’s chip dominance in this round is not a close race. It is every benchmark, every workload type, every scale. The upgrade cycle from GB200 to GB300 is confirmed by independent data at a 33% improvement. That is the recurring revenue argument for NVDA’s data center segment.

Full Analysis: 26 Charts, All 4 Tickers, Interactive Dashboard Available To Members (Get 33% OFF At Check Out)

#2 AMD's MI355X Is a 189% Per-Chip Leap. But AMD Only Ran 2 of 11 Benchmarks.

AMD (AMD) submitted results on the MI355X across 3 configurations. They covered 2 of 11 benchmark families: GPT-OSS-120B and LLaMA2-70B. AMD silicon did not run DeepSeek-R1, any LLaMA3 variant, Mixtral, Qwen3, DLRM, RGAT, WAN-2.2, or Whisper.

On both families AMD did run, NVIDIA silicon won. At cluster scale, AMD needs 21 to 31 percent more chips than NVIDIA to reach comparable throughput. That chip count premium is the core bear case for AMD in AI inference.

The bull case is the generational leap. The MI355X posts 12,935 tokens per second per chip on LLaMA2-70B. The prior generation MI325X posts 4,235. That is a 189% improvement in one generation.

Investor read: The MI355X per-chip improvement is real and significant. But 2 of 11 benchmark families tells you the software stack is not yet enterprise-broad. That is the gap AMD needs to close before hyperscalers run mixed workloads on AMD silicon.

Subscribe now

#3 Nebius Beat NVIDIA on NVIDIA's Own Chip. In Production Mode, by 44%.

Nebius (NBIS) ran the same GB300 NVL72 hardware as NVIDIA’s own submission. On GPT-OSS-120B, Nebius’s 72-GPU system posts 1,046,150 tokens per second offline. NVIDIA’s own system posts 1,042,980. Nebius wins.

The offline gap is 0.3%. The server mode gap is 44%.

Server mode measures how a system handles real concurrent requests under latency constraints. That is actual production traffic, not batch processing. On DeepSeek-R1 server mode with the same hardware, Nebius posts 575,580 tokens per second. NVIDIA posts 400,437.

Investor read: A cloud company that out-optimizes NVIDIA's own engineers on NVIDIA's own chip has a software moat. The server mode gap is not a rounding error. It means Nebius has specifically tuned their inference stack for live production traffic in a way that beats the reference implementation by 44%. For NBIS investors, this is the clearest public evidence that Nebius is building something more than a GPU rental business.

Full Analysis: 26 Charts, All 4 Tickers, Interactive Dashboard Available To Members (Get 33% OFF At Check Out)

#4 CoreWeave's Server Mode Beat Its Own Offline Score. That Does Not Happen.

CoreWeave (CRWV) submitted a 64-GPU GB300 NVL72 configuration on GPT-OSS-120B. Offline mode: 910,468 tokens per second. Server mode: 956,523.

Server mode beat offline by 5%. This does not normally happen. Offline processes requests in large batches with no latency constraints. It should always be faster. When server mode exceeds offline, it means the inference stack has been specifically tuned to generate higher sustained throughput under live concurrent load.

No other organization in this benchmark round shows this behavior. CoreWeave also posts the highest per-chip efficiency of any 8-GPU system in the round on DeepSeek-R1, beating NVIDIA's own DGX B300 submission.

Investor read: CoreWeave topped 0 benchmark families on the leaderboard. That framing misses the signal entirely. The server-beats-offline anomaly is unique in this dataset. It indicates a production serving stack engineered specifically for the workloads enterprise AI customers are actually running, not benchmark conditions.

#5 What the AMD Heterogeneous Cluster Tells You About Upgrade Cycles

Dell and MangoBoost submitted a 24-GPU cluster mixing 8x MI300X, 8x MI325X, and 8x MI355X chips in the same system. This is the only heterogeneous GPU cluster submission in the entire round.

Total throughput on LLaMA2-70B: 151,843 tokens per second. The per-chip average across the 24 mixed GPUs: 6,327 tokens per second. Pure MI355X nodes post 12,935 per chip. The MI300X chips pulled the cluster average down nearly in half.

Investor read: Hyperscalers will not mix GPU generations in production clusters. The efficiency penalty is too severe. This data confirms that AI infrastructure upgrade cycles are not optional. They are structural. Every generation of MI355X deployed today becomes a future replacement cycle for AMD. That is a constructive long-term data point for AMD investors even if today's numbers are mixed.

Full Analysis: 26 Charts, All 4 Tickers, Interactive Dashboard Available To Members (Get 33% OFF At Check Out)

The full What The Chip Happened dashboard covers all 91 systems, 23 organizations, and 11 benchmark families with 26 interactive charts.

MEMBERS NVDA vs AMD chip count efficiency analysis — how many GPUs does it take each platform to cross 1M tokens/second?

MEMBERS CoreWeave vs Nebius direct head-to-head across every comparable configuration

MEMBERS Full 8-GPU market ranking — every B300 and B200 system ranked by per-chip efficiency

MEMBERS Benchmark scorecard for each company — which models ran, which they led

MEMBERS AMD generation ladder: MI300X → MI325X → MI355X per-chip evolution chart

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

Google’s TurboQuant paper, presented at ICLR 2026, describes a method to compress something called KV cache, the working memory AI models use during inference, down to 3 bits. That yields a 6x reduction in KV cache memory. Financial media ran with it, and some even framed it as a 50% reduction in total memory costs. Memory stocks sold off across the board. Micron, Western Digital, SanDisk, even Seagate, a hard drive company with zero connection to KV cache.

The problem is that the paper never made that claim. KV cache is one slice of the total GPU memory. It does not touch model weights. It does not touch training. It does not touch activations. The paper tested only on small models, 8 billion parameters or fewer. No results exist for the 70B+ frontier models or mixture-of-experts architectures that drive the bulk of hyperscaler spending. [The headline wrote a check that the paper cannot cash.]

Here is where the math breaks the narrative. Micron CEO Sanjay Mehrotra, speaking on the Q2 2026 earnings call on March 18, said:

“Without more memory, without faster memory, AI just cannot scale up. Just look at from last year to this year, the DRAM requirement in the advanced AI accelerators has now doubled.”

DRAM content per accelerator is doubling every year. A one-time 6x compression of one memory subcategory does not offset 2x annual content growth that compounds year after year.

Think about that for a second. Even if you compress the KV cache by 6x today, the total DRAM each accelerator needs will be larger next year than it is right now. The compression buys you headroom. It does not reverse the trajectory.

Micron VP Satya Kumar added that systems shipping this year will have

“triple the amount of LPDDR compared to what we had last year.”

Cutting time to first token by 98%, he said, requires adding more memory, not less. Meanwhile, all three major HBM suppliers, Micron, SK Hynix, and Samsung, independently confirmed that HBM is sold out through 2026 with demand exceeding supply. Samsung told investors:

“Major customer demand for HBM in 2026 still exceeds available supply from us.”

That is three companies controlling nearly 100% of the market, all saying the same thing.

Lam Research CFO Douglas Bettinger put a dollar figure on it at the Cantor Fitzgerald Conference in March:

“WFE this year, $135 billion, up from $110 billion last year. It’s KV cache driving the need for DRAM.”

The equipment supplier building the factories sees KV cache as the primary demand driver. Not a risk. A driver. We read 24 management quotes from 11 companies over the past 90 days. Zero described memory compression as a risk to demand. Not one. The market invented a narrative that no industry participant supports. [That should tell you something.] Our AI/Semiconductor Research Community (Get 33% Off With This Link)

For investors watching the memory sector, the question is not whether KV cache compression works. It probably does, on a small scale. The question is whether it changes the supply-demand picture for 2026 and 2027. On a 2026 basis, it does not. HBM capacity is committed under contract. Equipment orders totaling $135 billion are locked in.

Even Jensen Huang at GTC 2026 said this about KV cache

"The large language models are going to get larger and larger and larger. It's going to generate more and more tokens more quickly, so it could think more quickly, but it also has to access memory. It's going to pound on memory really hard. KV Cache, structured data, cuDF, unstructured data, cuVS. It's going to be pounding on the storage system really, really hard, which is the reason why we reinvented the storage system."

A company reinventing the storage system is still placing massive orders of HBM and other memory solutions

SK Hynix called KV cache offloading

“essential to ensure smooth inference services.”

Astera Labs reported customers exploring CXL-based memory expansion specifically for KV cache applications. KV cache is creating new memory markets, not destroying existing ones. No hyperscaler, not Google, not Meta, not Microsoft, has told investors that KV cache compression reduces their infrastructure spend. Google created TurboQuant and has not mentioned it in a single investor forum.

The real signal for Micron investors is not TurboQuant. It is the capex raise to $25 billion or more, up from roughly $20 billion prior. That is real money funding HBM expansion. And the Summit Insights downgrade from Buy to Hold, one analyst against a consensus of 38 buys versus 2 sells. [The capex question deserves more attention than it got.]

Here is what could make this wrong. If independent researchers replicate TurboQuant’s zero accuracy loss on frontier-scale models, 70B parameters and above, the timeline to production adoption accelerates. What the synthesis estimates as 6 to 12 months could compress to 3 to 6 months. If hyperscalers deploy rapidly and cut HBM orders for 2027, the “sold out through 2026” defense has a built-in expiration date. Layer that on top of aggressive supply expansion from all three memory makers, and 2027 could look very different. The capex cycle turning is a real risk. TurboQuant on its own is not.

The market reacted to a headline about memory compression without checking what the paper actually compresses. The arithmetic says DRAM content growth absorbs it. Every management team in the supply chain says demand exceeds supply. That disconnect between narrative and data is the kind of gap this analysis exists to close.

---

If you found this useful, this is the surface.

The full breakdown, including the numbers management hoped you would miss, every product timeline with dates and dollar figures, and what analysts kept pushing on, is inside the What The Chip Happened community.

Institutional-quality semiconductor analysis for serious investors. Deep dives on earnings within 24 hours, conference coverage, and a network of investors who take this seriously.

[Join the community, 33% off at checkout](whatthechiphappened.com)

Get 15% OFF FISCAL.AI — WHERE CHARTS LIKE THIS CAN BE FOUND —

This will be a comprehensive analysis of Q4 FY2026 earnings expectations and why the hyperscaler CapEx supercycle makes the bear case untenable. You can find more research like this at our AI Community - WhatTheChipHappened.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

With Nvidia trading at ~26x forward earnings and AI bubble fears reaching fever pitch, bears argue the stock is priced for perfection ahead of Q4 FY2026 earnings. Dot-com comparisons are proliferating, custom silicon competition is accelerating, and skeptics question whether hyperscaler ROI can justify the spending.

Our View: EXTREMELY BULLISH

The bear case fundamentally misreads the supply-demand picture. Four hyperscalers simultaneously confirmed they are supply-constrained for AI compute. Over $660 billion in CapEx is committed for 2026, up dramatically from 2025 levels. TSMC’s CEO personally validated demand with every major cloud customer before committing to a $52-56 billion capex cycle to fuel semiconductor manufacturing. And $150 billion in AI lab funding has been raised in just the first seven weeks of 2026, with capital explicitly earmarked for Nvidia GPUs. At ~26x forward P/E with 50%+ revenue growth, Nvidia may be the cheapest it has been relative to its growth in this entire AI cycle.

Key Evidence:

• Every major hyperscaler stated demand exceeds supply — they will buy every GPU Nvidia ships

• TSMC wafer/advanced packaging supply is the bottleneck through 2028, preserving Nvidia’s pricing power indefinitely

• At ~26x NTM P/E with 50%+ growth, one of the cheapest profitable growth stocks.

• China sales represent pure upside optionality — AMD already reported Q4 China chip sales, confirming the regulatory pathway

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Note: This does not include capex from Neocloud, AI labs, and enterprise players, both public and private.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE BEAR CASE

Before making the bullish argument, intellectual honesty demands we examine what critics are saying, and why their concerns, while understandable, are ultimately not supported by the evidence.

The AI bubble narrative has intensified in recent months. The Case-Shiller price-to-earnings ratio for the U.S. market exceeded 40 for the first time since the dot-com crash. Michael Burry, the investor who famously shorted subprime mortgages—warned that the AI buildout is following similar “patterns” to the late 1990s. Speculation has emerged that leading AI tech firms are involved in circular investment flows artificially inflating stock values: Nvidia invests in OpenAI, OpenAI buys Nvidia chips, Nvidia revenue grows, repeat.

The custom silicon competition narrative has also gained traction. All four hyperscalers are developing proprietary chips: Google’s TPUs, AWS Trainium, Microsoft Maia, and Meta’s MTIA. Amazon disclosed that Trainium is now a $10 billion-plus annualized run-rate business, growing triple digits. The concern: Nvidia’s moat is eroding as customers vertically integrate.

ROI skepticism persists despite the spending. Reports indicate that 95% of organizations are seeing zero return on GenAI investments. If hyperscaler CapEx fails to generate adequate returns, the spending could prove unsustainable.

Finally, valuation sensitivity creates asymmetric risk, or so the bears argue. The stock is priced for perfection. Any guidance miss could trigger a significant selloff.

Why these concerns are gaining traction:

• Historical pattern recognition—dot-com parallels are intellectually seductive

• Custom silicon announcements create credible-sounding competitive threats

• The magnitude of spending ($600B+) seems unsustainable on its face

• Nvidia’s stock has already appreciated significantly—”how much higher can it go?”

We take these concerns seriously. Each deserves evidence-based responses. But when we examine what hyperscaler executives, TSMC’s CEO, and the broader supply chain are actually saying, in their own words, a very different picture emerges.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE HYPERSCALER SUPPLY CONSTRAINT

The single most important data point for Nvidia investors is this: all four hyperscalers, Meta, Alphabet, Microsoft, and Amazon, independently confirmed on their Q4 2025 earnings calls that AI compute demand currently exceeds their ability to supply it.

A supply-constrained customer will take every unit Nvidia can ship. This is not theoretical demand or speculative ordering. These are four of the largest, most sophisticated technology infrastructure buyers in the world, representing approximately 75% of global cloud compute, all saying the same thing simultaneously.

Meta’s CFO Susan Li was explicit during their January 28 earnings call:

“We do continue to be capacity constrained. Our teams have done a great job ramping up our infrastructure through the course of 2025. But demands for compute resources across the company have increased even faster than our supply. So we expect over the course of 2026 to have significantly more capacity this year as we add cloud. But we’ll likely still be constrained through much of 2026 until additional capacity from our own facilities comes online later in the year.”

Meta guided to $115–135 billion in 2026 CapEx—up 59–87% year-over-year—and they still expect constraints.

Alphabet’s Sundar Pichai echoed this on February 4. When asked directly about supply dynamics, he did not hedge:

“You are right, and we’ve been supply constrained even as we’ve been ramping up our capacity... I do expect to go through the year in a supply-constrained way.”

He described compute capacity as the single biggest thing that keeps him up at night:

“The top question is definitely around compute capacity, all the constraints, be it power, land, supply chain constraints, how do you ramp up to meet this extraordinary demand for this moment.”

Alphabet guided to $175–185 billion in 2026 CapEx—nearly double 2025 levels.

Microsoft provided the most granular GPU spending disclosure of the group. CFO Amy Hood stated on January 28:

“Capital expenditures were $37.5 billion, and this quarter, roughly 2/3 of our CapEx was on short-lived assets, primarily GPUs and CPUs. Our customer demand continues to exceed our supply.”

Two-thirds of $37.5 billion, approximately $25 billion, is spent on GPUs and CPUs in a single quarter. Microsoft then revealed what Azure growth would have been with unconstrained supply:

“If I had taken the GPUs that just came online in Q1 and Q2 and allocated them all to Azure, the KPI would have been over 40.”

Azure grew 39%. GPU supply is literally constraining Microsoft’s revenue growth.

Amazon’s Andy Jassy committed to approximately $200 billion in 2026 CapEx during the February 5 call:

“We expect to invest about $200 billion in capital expenditures across Amazon, but predominantly in AWS because we have very high demand, customers really want AWS for core and AI workloads, and we’re monetizing capacity as fast as we can install it.”

The phrase “monetizing capacity as fast as we can install it” is the most bullish demand signal a supplier can hear.

The takeaway: When demand exceeds supply across every major customer simultaneously, pricing power accrues to the supplier. The question is not “will demand hold?” but “how fast can TSMC manufacture these AI Chips?”

AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE SUPPLY CHAIN SAYS DEMAND IS REAL

The most compelling independent validation of Nvidia’s demand thesis comes not from Nvidia itself, but from TSMC, the sole manufacturer of every advanced Nvidia GPU. CEO C.C. Wei’s January 15 earnings call contained what may be the single most important passage for understanding the durability of AI infrastructure demand.

Wei addressed the bubble question head-on, acknowledging his own nervousness before revealing what convinced him:

“You essentially try to ask us, say, whether the AI demand is real or not. I’m also very nervous about it. You bet because we have to invest about $52 billion to $56 billion for the CapEx, right? If we didn’t do it carefully, that would be big disaster to TSMC for sure. So of course, I spend a lot of time in the last 3, 4 months talking to my customer and end customers’ customer. I want to make sure that my customers demand are real. So I talked to those cloud service providers, all of them. The answer is that I’m quite satisfied with the answer. Actually, they show me the evidence that the AI really help their business.”

This is not a chip company repeating what customers tell them. This is TSMC’s CEO personally validating demand by speaking directly with the hyperscalers’ customers, the enterprises, and AI labs actually deploying the compute. He found the evidence satisfactory enough to commit over $50 billion in capital.

On the longevity of the cycle, Wei was remarkably candid:

“All in all, I believe in my point of view, the AI is real, not only real, it’s starting to grow into our daily life... you—another question is can the semiconductor industry to be good for 3, 4, 5 years in a row, I’d tell you the truth, I don’t know. But I look at the AI, it looks like it’s going to be like an endless, I mean, that for many years to come.”

Critically, Wei identified where the bottleneck lies, and it is not demand:

“Today, from my point of view, still the bottleneck is TSMC’s wafer supply. Not the power consumption, not yet... TSMC’s wafer can support how much of the gigawatt, still not enough.”

He then provided the timeline for relief:

“If you build a new fab, it takes 2 to 3 years to build a new fab. So even we start to spend the $52 billion to $56 billion, the contribution to this year almost none and to 2027, a little bit. So we actually are looking for 2028, 2029 supply.”

The memory supply chain tells the same story. Micron CEO Sanjay Mehrotra stated on December 17:

“We are more than sold out. I mean, we have a significant amount of unmet demand in our models... the demand on us is so much higher than the supply, that even small increases in supply are not going to be able to make a dent in that demand.”

Micron’s entire HBM supply for 2026 is already contracted with volume and price locked.

Applied Materials CEO Gary Dickerson added the capstone on February 12:

“With the accelerated growth of AI end markets, we believe that global semiconductor industry revenues can potentially reach $1 trillion in 2026, several years earlier than prior predictions.”

The takeaway: When TSMC commits $52–56 billion in CapEx after personally validating demand with every hyperscaler, when Micron is “more than sold out” on HBM with “significant unmet demand,” and when Applied Materials sees the $1 trillion semiconductor milestone arriving four years early, this is not speculative froth. This is industrial-scale conviction backed by capital commitments. The supply constraint will persist through 2028, meaning Nvidia retains pricing power for at least the next 2 years.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE NEOCLOUD AND AI LAB DEMAND LAYER

Beyond the hyperscalers, a massive secondary demand wave is emerging from neocloud providers and AI labs, representing over $150 billion in fresh capital raised in early 2026 alone, with the explicit destination being Nvidia GPU purchases.

Nebius CEO Arkady Volozh described the demand picture on his recent earnings call:

“Demand remains robust, and our pipeline continues to grow substantially. We sold out of capacity in Q3 and Q4 last year, and we’re already now in Q1 of 2026 also sold out. Even before we bring capacity online, it’s often sold out.”

He added:

“Everything we build, we sell. We are in the very early days of one of the biggest industrial and technological revolutions in history.”

A critical data point counters the concern that older GPU generations will become stranded assets. IREN COO Kent Draper provided definitive evidence:

“If you look more broadly across the industry, if you think of A100, H100s, those are more than 5 years old and more than 3 years old, respectively now. Those chips are still effectively 100% utilized across the industry and still earning very good rates of return against their original capital costs.”

AWS CEO Matt Garman confirmed the same at Cisco 2026 AI Summit:

“We actually are completely sold out of and have never retired an A100 server.”

Contract durations are lengthening, signaling customer confidence in multi-year demand. Nebius CRO Marc Boroditsky noted:

“In Q4, we saw nearly twice as many transactions completed over 12 months in duration over what we succeeded with in Q3, while average selling prices increased by more than 50%.”

When customers sign longer contracts at higher prices, they are voting with capital on demand durability.

The AI lab funding picture is staggering. OpenAI is nearing a deal to raise more than $100 billion. Anthropic closed a $30 billion Series G on February 12 at a $380 billion valuation, the second-largest venture round ever. Elon Musk’s xAI started the year with a $20 billion Series E. Combined early 2026 funding from just these three labs exceeds $150 billion.

Where does this capital go? The reporting is explicit: “Much of OpenAI’s new capital is likely to be used to purchase Nvidia chips to power OpenAI’s AI models.” These startups don’t build their own data centers, they rent from AWS, Azure, and GCP.

The takeaway: The “circular AI financing” concern is overstated. The hyperscalers are funding their CapEx from operating cash flows and balance sheets, profitable companies deploying existing capital, not leveraged speculation.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE VALUATION DISCONNECT

At approximately 26x forward earnings with 50%+ revenue growth expected in FY2027, Nvidia trades in-line with the Magnificent 7 average despite dramatically superior growth. The market is pricing in excessive risk that the data does not support.

The earnings trajectory tells the story clearly. Wall Street expects Q4 fiscal 2026 earnings of approximately $4.69 per share. If Nvidia hits that number, the stock at $180 trades at roughly 38x trailing earnings, which sounds elevated. But analysts expect fiscal 2027 earnings to reach $7.66 per share, placing the stock at just mid 20s forward earnings on next year’s numbers.

Why is the market mispricing this? Several factors contribute. AI bubble fears create an excessive risk premium that depresses multiples. China uncertainty keeps estimates conservative, any revenue is excluded from guidance. The custom silicon narrative creates perceived competitive threat that the data does not support. And Nvidia has historically beaten and raised guidance, meaning consensus estimates likely understate actual results.

The critical timing observation: the hyperscaler CapEx commitments that support the bull case were made after Nvidia’s Q3 earnings. Meta, Alphabet, and Amazon all reported in late January and early February, after Nvidia guided. The demand environment is strengthening, not weakening. If the market fully incorporated the incremental demand signals, the stock would trade higher.

The takeaway: If Nvidia delivers Q4 in-line and guides Q1 FY27 above $74 billion (versus ~$71.6 billion consensus), the stock could re-rate over 10-15% simply on multiple expansion as AI bubble fears dissipate. At current prices, the risk-reward is asymmetric in favor of longs.

AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE CHINA OPTIONALITY

AMD’s Q4 earnings demonstrated that chip sales to China are flowing again with MI308, including licenses for MI325 accelerators. This provides a critical proof point: if AMD is selling chips to China, Nvidia’s H20/H200 approvals should translate into material revenue, revenue currently excluded from Nvidia’s guidance.

Nvidia CEO Jensen Huang’s first visit to China in 2026 represented a significant shift. Beijing has reportedly approved the sale of 400,000 H200 chips to China’s top AI firms. Reuters indicated that Nvidia has already received more than 2 million orders for H200 chips for 2026.

At an estimated ASP of $25,000–30,000 per H200, 400,000 approved chips represent $10–12 billion in potential revenue. If Nvidia has 2 million orders (per Reuters), the total addressable opportunity is $50–60 billion, against zero in current guidance.

The AMD proof point de-risks the thesis. If one U.S. AI chip company is successfully navigating China sales, there is no regulatory reason Nvidia cannot do the same.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

HISTORICAL CONTEXT

The dot-com comparison deserves direct address. Yes, the Case-Shiller P/E ratio exceeded 40 for the first time since the dot-com crash. Yes, AI spending growth rates echo 1990s infrastructure buildouts. But the structural differences matter more than surface similarities.

First, real earnings versus projections. During the dot-com bubble, companies were valued on eyeballs and revenue projections that never materialized. Today’s AI leaders are already profitable. Nvidia’s earnings grew 67% year-over-year in Q3 while revenue grew 62%. Meta, Alphabet, Microsoft, and Amazon are deploying operating cash flow into AI infrastructure. Today’s AI giants have lower debt-to-earnings ratios than WorldCom or the other poster children of 1999.

Second, CapEx as percentage of GDP remains below historical peaks. AI capex has recently equated to approximately 0.8% of GDP. During other technology booms, peak levels reached 1.5% or greater.

Third, physical assets are being built. Unlike the 1990s, today’s investment is heavily capex-driven: data centers, GPUs, networking equipment, power systems. AWS disclosed nearly 4 gigawatts of power capacity added in 2025 alone, real infrastructure with 20–30 year useful lives.

Fourth, post-Sarbanes-Oxley standards prevent the accounting fraud that inflated dot-com demand.

The lesson from history: Bubbles are characterized by leverage, fraud, and disconnection from fundamentals. The current AI infrastructure buildout is characterized by operating cash flow, audited financials, and supply-constrained demand. The comparison does not hold.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SCENARIO ANALYSIS

We weight the bull and base scenarios at 95% combined probability. The hyperscaler CapEx announcements since Q3 suggest guidance will surprise to the upside, not the downside. The 5% bear scenario probability reflects execution risk and potential macro disruption, not demand concerns.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

KEY MANAGEMENT COMMENTARY

Microsoft (MSFT) — Cisco AI Summit 2026

“I just don’t see how the demand for inference is gonna go down. And it’s hard to imagine, given what the silicon situation is, the hardware situation, how difficult it is to build data centers and deploy power and all the things that you wanna do, how you get ahead of that anytime soon.” — Kevin Scott, CTO

This quote addresses the “efficiency will reduce demand” concern directly — Microsoft’s CTO sees no scenario where inference demand declines.

Alphabet (GOOGL) — Cisco AI Summit 2026

“Deep partnership with NVIDIA — a lot of our success at Google has been as a result of that partnership with NVIDIA and GPUs... Every efficiency we deliver — and it’s the rate of improvement on the efficiency side, I’ve never seen anything like it — but it gets consumed instantaneously.” — Amin Vahdat, Chief Technologist

This is the Jevons Paradox in real-time: efficiency gains get immediately absorbed by expanding workloads, not by reduced purchasing.

Amazon (AMZN) — Q4 FY2025 Earnings Call

“Data centers are a 20-30-year amortization timeline. The power, you have to make commits for long periods of time. I’m gonna have this asset for 20 years or 30 years.” — Matt Garman, CEO AWS

The multi-decade commitment horizon signals this is not cyclical spending that can be quickly reversed.

TSMC (TSM) — Q4 2025 Earnings Call

“The AI is real, not only real, it’s starting to grow into our daily life... it looks like it’s going to be like an endless, I mean, that for many years to come.” — C.C. Wei, CEO

Independent foundry validation after direct conversations with every hyperscaler and their customers.

Super Micro (SMCI) — Q2 FY2026 Earnings Call

“AI GPU platforms, which represent over 90% of Q2 revenue, continue to be the key growth driver... we are also preparing for upcoming NVIDIA Vera Rubin.” — Charles Liang, CEO

Vera Rubin systems already have committed customer orders for H2 2026 delivery.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

RISKS TO OUR VIEW

Our thesis is not without risks. Here’s what could prove us wrong:

1. Guidance Miss Risk The stock is priced for strong guidance. Any guide below $70 billion for Q1 FY27 would likely trigger a 10-15% selloff regardless of Q4 results. The market has elevated expectations, and disappointment would be punished. Our mitigant: hyperscaler CapEx announcements since Q3 suggest upside, not downside, to guidance.

2. Gross Margin Compression Blackwell ramp costs and rack-scale system complexity could pressure margins below the 75% guide. The transition from Hopper HGX to Blackwell full-scale data center solutions involves new component costs and manufacturing learning curves. Our mitigant: management has flagged this as temporary, with margins expected to improve as Blackwell scales.

3. Custom Silicon Acceleration If hyperscalers accelerate Trainium, TPU, and Maia adoption faster than expected, Nvidia’s share of incremental AI compute spend could decline. Amazon’s Trainium is already a $10B+ business. Our mitigant: every hyperscaler explicitly reaffirmed Nvidia partnership alongside custom silicon. Alphabet confirmed being “among the first to offer” Nvidia’s Vera Rubin platform. These are complements, not substitutes.

4. China Regulatory Reversal Geopolitical tensions could re-freeze H200 exports despite recent approvals. The regulatory environment remains fluid. Our mitigant: China revenue is excluded from guidance entirely — any sales are pure upside, not base case.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

BOTTOM LINE

The bear case rests on abstract fears: bubble comparisons, custom silicon threats, ROI skepticism. The bull case rests on concrete evidence: $660–690 billion in committed hyperscaler CapEx, four CEOs simultaneously declaring supply constraints, TSMC’s CEO personally validating demand after speaking with every major customer, Micron’s HBM sold out with “significant unmet demand,” and $150 billion-plus in AI lab funding raised in early 2026, capital mainly earmarked for Nvidia GPUs.

The supply chain is unanimous. The hyperscalers are unanimous. The neoclouds are unanimous. The only dissent comes from observers who are not writing purchase orders.

Nvidia is the most mispriced mega-cap in technology. At ~26x forward earnings with 50%+ growth, implying the market assigns a massive risk premium that the evidence does not support.

For investors evaluating the setup into Q4 earnings, the asymmetry favors being long. A guidance beat above $74 billion for Q1 FY27 could trigger meaningful re-rating. A guidance miss, which would require demand to have softened despite every hyperscaler announcing increased CapEx, is unlikely given the timing of those announcements. China represents a free call option with substantial upside.

The +$600 billion in committed 2026 hyperscaler capital expenditure is not a ceiling—it is a floor. The AI infrastructure supercycle has not peaked. It is accelerating. And Nvidia remains the indispensable supplier at the center of it all.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The AI infrastructure buildout just shifted into overdrive. Astera Labs is staring at a 10x TAM explosion to $25 billion as connectivity becomes the new bottleneck. Nebius is deploying $20 billion in CapEx after selling out every GPU through Q1. Applied Materials just called a $1 trillion semiconductor industry, years ahead of schedule, while guiding 20%+ equipment growth. Three different layers of the stack, one message: AI demand isn’t slowing down. — Let’s Chip In.

What The Chip Happened?

🚀 Astera Labs Unveils $25B TAM as PCIe 6 Dominance Pays Off
⚡ Nebius Sold Out Through Q1, Plans $20B CapEx Blitz to 7x Revenue
🏭 Applied Materials Calls $1T Chip Industry as AI Accelerates Everything

Read time: 7 minutes

AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Astera Labs (NASDAQ: ALAB)
🚀 Astera Labs Sees 10x TAM Explosion to $25B, Revenue Rips Higher

What The Chip: Astera Labs just dropped the connectivity mic. Q4 revenue hit $270.6 million (+92% YoY), full-year 2025 closed at $852.5 million (+115% YoY), and management casually unveiled a 10x TAM expansion to $25 billion over five years. The driver? Merchant scale-up switching opportunities they believe could reach $20 billion by 2030—and they’re positioning to capture “at least half” near-term. With PCIe 6 leadership, UALink production starting 2027, and custom NVIDIA solutions shipping, Astera is cementing its position as the critical middleware layer for AI infrastructure.

Details:

💰 Revenue Acceleration: Q4 delivered $270.6M, up 17% sequentially and 92% YoY, while full-year 2025 revenue of $852.5M (+115% YoY) crushed expectations across all product lines. The sequential growth demonstrates momentum heading into 2026, not just easy comps.

🔮 TAM Bombshell: Management raised their 5-year served addressable market estimate to $25 billion—10x their original projections. The game-changer: merchant scale-up switching alone expected to reach ~$20B by 2030, with Astera targeting “at least half” of the near-term opportunity. This isn’t incremental—it’s a complete reframing of the business’s scale potential.

🔥 Scorpio Outperformance: The P-Series PCIe 6 fabric exceeded 15% of 2025 revenue versus the 10% target, and management emphasized it’s “the only PCIe 6 fabric shipping in volume.” Two new major hyperscalers join in 2026 (bringing total to 3), with production starting late 2026 and meaningful revenue contribution expected in 2027. Customer diversification is accelerating.

🚀 X-Series Scale-Up: The scale-up switching product shipped preproduction quantities in Q4 and expects “material step up in back half” of 2026, with volume ramp in 2027. Management says X-Series will become “our biggest product line”—10+ customers are engaged, and the TAM here dwarfs existing products.

⚡ Taurus Surge: The strongest Q4 performer with 4x YoY growth in 2025. The 400G designs provide a baseline for 2026 growth, and the 800G switching transition represents the next major catalyst. Taurus is proving that AI networking demand extends well beyond the headline PCIe story.

🎯 Aries Leadership: The retimer portfolio grew ~70% YoY, with Astera claiming “industry’s only PCIe 6 DSP retimer solutions shipping in high volume.” Development is already extending to PCIe Gen 7, ensuring multi-generation staying power in a category the company pioneered.

🤖 NVIDIA Partnership Deepens: Custom solutions announced to enable NVLink Fusion scale-up architecture for hybrid racks, with opportunities expanding to additional hyperscalers for “interconnect flexibility and optionality.” This positions Astera as a key enabler of NVIDIA’s most advanced architectures.

💎 Balance Sheet Fortress: $1.19B in cash/equivalents with Q4 operating cash flow of $95.3M and full-year 2025 operating cash flow of $319.3M. Astera is funding growth from operations, not dilution.

Why AI/Semiconductor Investors Should Care: Astera is establishing itself as the critical connectivity layer that prevents AI infrastructure from choking on its own bandwidth requirements. The 10x TAM expansion to $25B suggests the AI connectivity opportunity is far larger than anyone modeled, and the company’s PCIe 6 leadership, UALink positioning, and custom NVIDIA solutions create multiple vectors for sustained hypergrowth. The addition of two new hyperscaler customers for Scorpio de-risks concentration concerns while X-Series represents a product cycle that could dwarf existing revenue. Watch gross margin trajectory carefully as hardware mix increases—the Q1 guide of 74% versus 75.7% actual is manageable, but continued compression would pressure the bull case. The key catalysts: X-Series production ramp in H2 2026, UALink ecosystem adoption timing, and whether those two new hyperscaler Scorpio wins convert to material 2027 revenue as management projects.

AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Nebius Group (NASDAQ: NBIS)
⚡ Nebius Posts 547% Revenue Surge, Sold Out Through Q1 2026

What The Chip: Nebius just proved that AI cloud infrastructure is the most explosive subsector in tech. Q4 revenue exploded 547% year-over-year to $227.7 million, with core AI cloud revenue surging 800%+ and the business flipping to profitability for the first time. The company is completely sold out of GPU capacity through Q1 2026, has already secured over 2 GW of contracted power (ahead of schedule), and is reiterating guidance of $7-9 billion ARR by year-end 2026—representing 7x growth from today’s $1.25 billion run rate. With major hyperscaler contracts from Microsoft and Meta now delivering, Nebius is positioning itself as the premier independent AI cloud infrastructure play.

Details:

🚀 Revenue Explosion: Q4 revenue of $227.7M (+547% YoY, +56% QoQ) crushed expectations, with core AI cloud revenue of $214.2M (+800% YoY) now representing 94% of total revenue. The sequential acceleration demonstrates this isn’t a one-time comp benefit—demand is compounding.

💰 ARR Blowout: Ended December at $1.25B ARR, crushing the high-end guidance of $1.1B and more than doubling from $551M in September. The company added $700M in ARR in a single quarter while operating at 100% utilization—imagine what happens when capacity expands.

🔥 Profitability Inflection: First-ever positive group adjusted EBITDA of $15M (7% margin), with core AI cloud business hitting 24% adjusted EBITDA margin, up from 19% in Q3. The business model scales beautifully—management is guiding ~40% EBITDA margins for 2026.

⚡ Sold Out Status: Operating at 100% utilization with capacity sold out in Q3, Q4, and now Q1 2026. GPU pricing is holding firm even on previous generations, indicating demand far exceeds supply. This isn’t a race to the bottom—it’s a land grab in a supply-constrained market.

🔌 Power Secured: Already at 2+ GW of contracted power in February, ahead of the original year-end 2026 target of 2.5 GW. Management raised the new year-end target to 3+ GW. Power is the ultimate constraint in AI infrastructure, and Nebius is securing it faster than competitors.

💎 Massive 2026 CapEx: Planning $16-20B in capital expenditure this year—~60% funded from balance sheet and existing operations, with the balance from debt, asset-backed financing, and ATM equity program. This is audacious and aggressive, but the demand supports it.

🤖 Hyperscaler Delivery: Meta capacity fully deployed as of early February (both tranches); Microsoft first tranche delivered November 2025, with remainder throughout 2026 and full run-rate contribution in 2027. These aren’t aspirational deals—they’re producing revenue now.

📈 Customer Economics Strengthening: Average contract duration up 50%; ASPs increased 50%+ in Q4; pipeline on track to exceed $4B in Q1 2026 alone. Customers are committing longer and paying more—a sign of product-market fit at the highest level.

Why AI/Semiconductor Investors Should Care: Nebius is capturing an outsized share of the AI infrastructure buildout with hyperscaler validation (Microsoft, Meta), sold-out capacity, and a clear path to 7x revenue growth in a single year.

The 40% EBITDA margin target for 2026 suggests the business model scales beautifully—profitability isn’t sacrificed for growth. The bear case centers entirely on execution risk: deploying $16-20B in CapEx while maintaining quality and on-time delivery is extraordinarily challenging, and any delays would crater the stock given current expectations.

Investors should watch Q2/Q3 capacity deployment milestones closely, monitor Microsoft tranche delivery timelines, and assess whether pricing holds as industry capacity expands in H2 2026. The company’s claim that “$7-9B ARR is not dependent on any new mega deals” is bold—it suggests the existing pipeline and planned capacity additions alone support the target. If true, Nebius is one of the few purest AI infrastructure plays available. If execution falters, the stock will be unforgiving.

AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Applied Materials (NASDAQ: AMAT)
🏭 Applied Materials Sees 20%+ Equipment Growth as AI Drives $1T Chip Industry

What The Chip: Applied Materials just called the semiconductor industry’s future—and it’s arriving years ahead of schedule. The company delivered Q1 revenue of $7.01B with non-GAAP EPS of $2.38, then guided Q2 to $7.65B revenue (+9% sequential) and $2.64 EPS (+11% sequential). But the real story is management’s boldest forecast yet: the global semiconductor industry could hit $1 trillion in revenue in 2026—years ahead of prior predictions—driven by AI infrastructure buildout. Applied expects its equipment business to grow over 20% this calendar year, with record DRAM revenue, dominant positions in GAA and HBM, and extended customer visibility into 2027.

Details:

💰 Revenue Beat: Q1 revenue of $7.01B came in at the upper end of guidance, down 2% YoY but up 3% sequentially as demand shifts toward AI-enabling technologies like leading-edge logic, HBM, and advanced packaging. The sequential growth signals inflection.

📈 Margin Expansion Continues: Non-GAAP gross margin hit 49.1%—the highest in 25 years—up 70bps above expectations. Gross margins have expanded 700bps since Gary Dickerson became CEO, driven by value-based pricing in AI-critical technologies.

🚀 Q2 Acceleration: Guiding Q2 revenue to $7.65B (midpoint) with non-GAAP EPS of $2.64, representing 9% and 11% sequential growth respectively. The guidance signals demand inflection, not plateau.

🔮 Calendar 2026 Growth: Equipment business expected to grow >20% in CY26, with demand weighted toward H2. Customer cleanroom availability is pacing investment rate—they want to spend faster but need facilities ready.

🌍 $1 Trillion Industry Call: Management believes global semiconductor revenue could reach $1T in 2026, “several years earlier than prior predictions.” This isn’t a long-term aspiration—it’s a near-term forecast that reframes the entire sector’s growth trajectory.

🔥 Record DRAM Revenue: Semiconductor Systems delivered record DRAM revenue, driven by HBM demand where wafer starts are running 3-4x higher to deliver equivalent bit output. HBM isn’t just additive—it’s multiplicative for equipment intensity.

🎯 Market Leadership: #1 position in leading-edge foundry/logic, DRAM, HBM, and advanced packaging. The company is targeting >50% share in gate-all-around and advanced wiring—the two most critical technology transitions happening now.

💡 Product Pipeline: Launching 12+ new products in 2026 including three major systems for GAA and advanced DRAM. eBeam revenue expected to double to >$1B, driven by GAA adoption. Applied is winning share at the most critical technology nodes.

Why AI/Semiconductor Investors Should Care: The $1 trillion semiconductor TAM call for 2026—years ahead of prior forecasts—isn’t promotional hyperbole; it’s backed by extended customer visibility into 2027 and the massive cleanroom buildout currently underway. The company’s ability to expand gross margins to 25-year highs while growing 20%+ demonstrates pricing power in mission-critical technologies.

Watch customer cleanroom completion timelines closely—management noted demand is H2-weighted based on fab readiness, not chip demand. The 49.1% gross margin is impressive but creates difficult comps; any mix shift toward China (currently 27-30% of sales, expected flat) could pressure margins. Key catalysts: HBM capacity expansion pace, GAA transistor adoption at Intel and Samsung, and whether the record AGS services momentum (15% YoY growth, record spares) can sustain double-digit growth as guided.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
AI & CHIP STOCK RESEARCH COMMUNITY — CLICK HERE FOR 33% OFF
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The earnings confessionals this week revealed an industry grappling with a problem most would kill for: demand so strong it's breaking supply chains. Intel's data center business grew 15% QoQ but warned Q1 revenue will crater 11% because they literally can't make enough chips. TSMC's CEO spent his earnings call validating hyperscaler bank accounts and just committed $56B in CapEx because silicon is the bottleneck. Meanwhile, Intuitive Surgical proved it can print $2.5B in free cash flow while nervously eyeing six different macro threats. — Let's Chip In.

What The Chip Happened?
🏭 Intel’s Data Center Surge Hits a Brick Wall Made of Wafers
🚀 TSMC CEO Does Due Diligence on Hyperscaler Wealth, Likes What He Sees
🤖 Intuitive’s 20 Million Patient Victory Lap Gets Interrupted by Reality

Read time: 7 minutes

Join WhatTheChipHappened Community — 33% OFF Annual Use Code “2026”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Intel Corporation (NASDAQ: INTC)
🏭 Intel’s Data Center Boom Hits Supply Wall — Q1 Revenue Dropping 11%

What The Chip: Intel just posted its fifth consecutive quarter of beating guidance with Q4 revenue hitting $13.7B, but the celebration lasted about as long as their buffer inventory. The company warned that severe supply constraints will hammer Q1 2026 results to just $12.2B at midpoint—an 11% sequential nosedive. The cruel irony: their Data Center and AI business is growing at the fastest pace this decade with 15% QoQ growth, and management flat-out admitted revenue “would have been meaningfully higher” if they could actually manufacture enough chips. Oh, and they just became the only company on Earth shipping gate-all-around transistors with backside power at volume with their Intel 18A process.

Details:

⚡ Data Center Inflection is Real: DCAI revenue hit $4.7B, up 15% QoQ and 9% YoY—the fastest sequential growth Intel has seen in over ten years. Management confirmed they’re supply-constrained in this segment, which validates the bull thesis that Intel’s AI CPUs actually matter in the data center buildout.

🚩 Q1 Guide is Brutal: Revenue guided to $11.7B-$12.7B ($12.2B midpoint) as internal supply constraints hit peak pain. Buffer inventory has been depleted, and the wafer mix shift toward servers that started in Q3 won’t emerge from fabs until late Q1. Gross margin collapses to 34.5% with EPS at breakeven ($0.00 non-GAAP).

🎯 18A Milestone Delivers: Intel shipped its first Intel 18A products with Core Ultra Series 3 (Panther Lake), launching 3 SKUs versus the 1 SKU commitment. This makes Intel the only manufacturer globally shipping GAA transistors with backside power delivery at volume—a genuine technology leadership moment that’s being overshadowed by near-term supply pain.

💰 ASIC Business Crosses $1B: The custom ASIC business grew 50%+ in 2025 and hit a $1B+ annualized run rate in Q4. Intel is positioning this against a $100B TAM opportunity, suggesting Foundry is gaining traction beyond the headline-grabbing hyperscaler deals.

📉 Margin Reality Check: That 34.5% Q1 gross margin isn’t just supply mix—Panther Lake is actively dilutive to margins right now. CFO David Zinsner admitted this level is “by no means acceptable” and set 40% as the first recovery milestone before they’ll set a higher target. Translation: yield issues are real.

💡 Cost Cuts are Working: Non-GAAP OpEx fell 15% YoY to $16.5B as headcount dropped from 108.9K to 85.1K employees. Intel is targeting $16B OpEx for full year 2026, showing restructuring discipline while navigating the supply crisis.

🛡️ Balance Sheet Fortified: Intel exited 2025 with $37.4B in cash and investments after closing NVIDIA’s $5B equity investment, completing the Altera sale, and securing government funding. The liquidity crisis narrative is dead—they have runway to execute through this rough patch.

⚠️ Industry Component Shortage: Intel warned that DRAM, NAND, and substrate shortages are hitting the entire industry, which could pressure client margins and “limit our revenue opportunity this year.” The supply chain crunch extends beyond Intel’s internal fab constraints.

📈 Client Business Struggles: CCG revenue declined 7% YoY despite AI PC units growing 16%, pointing to significant ASP pressure in the consumer segment. Management is targeting 45% client market share “over the next several years”—an admission they’ve lost share and need multiple years to recover.

🔮 The Coordinating CPU Thesis: CFO Zinsner laid out Intel’s long-term bull case: “The world is shifting from human prompted requests to persistent and recursive commands driven by computer-to-computer interactions.” Intel believes the CPU’s role coordinating AI agent traffic will drive both traditional refresh and entirely new demand vectors.

Why AI/Semiconductor Investors Should Care: This is what genuine AI-driven demand inflection looks like—growing so fast you can’t manufacture enough to capture it. The Q1 guide is ugly, but it’s a supply problem not a demand problem, and those tend to be fixable with capital and time. The 18A milestone validates Intel’s process technology comeback story, but the 34.5% gross margin reveals the cost of ramping new nodes. Bulls should watch for evidence of Q2 supply improvement and Panther Lake yield trajectory—if Intel executes, there’s significant pent-up revenue waiting to be captured in DCAI. Bears will rightfully focus on the path from 34.5% to 40%+ margins, which remains dependent on yield improvements that haven’t been quantified yet. Intel 14A customer announcements expected H2 2026 to H1 2027 will be the next validation point for Foundry credibility.

Join WhatTheChipHappened Community — 33% OFF Annual Use Code “2026”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Taiwan Semiconductor Manufacturing (NYSE: TSM)
🚀 TSMC CEO Checks Hyperscaler Bank Accounts, Then Commits $56B

What The Chip: TSMC’s CEO C.C. Wei did something remarkable on the Q4 earnings call: he admitted he was “nervous” about committing $52-56B in 2026 CapEx, so he personally validated AI demand with every major hyperscaler. His conclusion after reviewing their financials? “They are very rich.” The company delivered 62.3% gross margins (beating guidance by 130bps), raised its 5-year AI accelerator revenue CAGR forecast to the mid-to-high 50s%, and warned that supply will remain constrained through 2027 despite the massive capital deployment. Translation: the AI infrastructure buildout is real, it’s massive, and TSMC has line-of-sight that few others possess.

Details:

💰 Q4 Crushed Expectations: Revenue hit $33.7B (+5.7% QoQ in NT$ terms) with gross margins at 62.3%—130bps above the high end of guidance. Operating margin reached 54% driven by cost improvements and high utilization rates. Full year 2025 revenue hit $122B, up 35.9% YoY.

🤖 AI Revenue Trajectory: AI accelerators reached “high teens %” of total 2025 revenue, with the entire HPC platform now representing 58% of revenue (+48% YoY). TSMC raised its 2024-2029 AI accelerator CAGR forecast to mid-to-high 50s%—a breathtaking growth rate sustained over five years.

⚡ Advanced Nodes Dominate: 77% of Q4 wafer revenue came from 7nm and below, up from 69% in 2024. The 3nm node alone contributed 28% of wafer revenue as N2 entered high-volume manufacturing in Q4 with “good yield” at both Hsinchu and Kaohsiung sites.

🏭 Unprecedented CapEx Commitment: 2026 budget of $52-56B (up from $40.9B in 2025), with the next 3 years “significantly higher” than the prior 3 years’ $101B combined total. TSMC is betting $150B+ that AI demand sustains, making this one of the largest capital commitments in semiconductor history.

🔥 Supply Stays Tight Through 2027: Despite the massive CapEx, CEO Wei confirmed capacity constraints will persist through 2026-2027. His message to investors: “We have to work extremely hard to narrow the gap.” Hyperscalers told him silicon from TSMC is the bottleneck, not power or infrastructure.

🌍 Arizona Expansion Accelerates: Fab 1 is in high-volume production, Fab 2 has been pulled forward to H2 2027 (from 2028), Fab 3 construction has started, Fab 4 permits are underway, and TSMC just purchased a second land parcel in Arizona. This is what conviction looks like in physical form.

📈 Monster 2026 Guidance: Revenue growth guided to ~30% YoY with Q1 revenue at $35.2B midpoint (+38% YoY). Gross margins guided to 63-65% for Q1, with long-term framework targeting 56%+ gross margins through the cycle despite margin headwinds from new nodes and overseas fabs.

🎯 CEO Did His Homework: Wei spent “a lot of time” talking to customers and end customers. He checked hyperscaler financials personally and confirmed they’re investing in power infrastructure 5-6 years ahead of silicon needs. His takeaway: “AI really helps their business. They grow their business successfully and healthy in their financial return.”

💡 Dividend Boost Signals Confidence: 2026 dividend guided to at least TWD 23/share versus TWD 18 in 2025—a 28% increase that shows TSMC can fund massive CapEx while returning more cash to shareholders.

⚠️ Margin Headwinds Acknowledged: N2 ramp creates 2-3% gross margin dilution while overseas fabs add 2-4% dilution. However, N3 is expected to cross above corporate average margins “sometime in 2026,” showing the path to absorbing new node costs.

Why AI/Semiconductor Investors Should Care: TSMC just gave investors unprecedented demand visibility backed by CEO-level validation of hyperscaler financials and infrastructure planning. When the world’s leading foundry commits $150B+ over three years and still expects supply constraints through 2027, that’s a signal the AI infrastructure buildout is structurally larger than most models anticipate. The combination of 25% overall revenue CAGR, 50%+ CAGR for AI accelerators, 56%+ target gross margins, and technology leadership with N2/A16 creates a multi-year compounding story. The risk is execution—$150B is enormous capital at risk if AI demand disappoints, and Wei’s admission of being “nervous” shows management understands the stakes. But his due diligence on customer financials and power infrastructure planning suggests downside risk is limited. Watch for N3 margin crossover timing in 2026 and whether Q1 gross margins hit the 64% midpoint—those will validate whether TSMC can maintain margin discipline while ramping capacity aggressively. This is the purest AI infrastructure play available to public market investors.

Join WhatTheChipHappened Community — 33% OFF Annual Use Code “2026”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Intuitive Surgical (NASDAQ: ISRG)
🤖 Intuitive Hits 20M Patients, Then Lists Six Reasons to Worry

What The Chip: Intuitive Surgical closed 2025 with a victory lap—$2.87B in Q4 revenue (+19% YoY), 18% procedure growth, and $2.5B in free cash flow that nearly doubled the prior year. The company crossed 20 million patients treated since 1997, with the da Vinci 5 upgrade cycle accelerating as 303 systems shipped in Q4 (+74% YoY). But the champagne was flat when management guided 2026 procedure growth to just 13-15%, down sharply from Q4’s 18% exit rate. The culprit: a laundry list of macro concerns from ACA subsidy uncertainty to “intensified” Chinese competition. Investors now face a classic setup—does the da Vinci 5 cycle and new ASC strategy overcome external headwinds, or is the deceleration structural?

Details:

🏥 Milestone Achievement: Intuitive has now treated over 20 million patients with 3.1 million procedures performed in 2025 alone. This validates robotic surgery’s transition from niche innovation to mainstream standard of care across multiple procedure types.

📈 da Vinci 5 Momentum: 303 systems placed in Q4 (+74% YoY) with 146 trade-ins (+135% YoY) as hospitals aggressively upgrade their fleets. The installed base hit 1,232 systems being used by 10,000+ surgeons. The upgrade cycle is just beginning at 11% penetration of eligible systems.

💰 Cash Generation Inflection: Free cash flow hit $2.5B for 2025 (+92% YoY), funding $2.3B in share buybacks at an average price of $478/share while growing cash to $9B. The business model is inflecting from growth investment to cash return mode.

🛡️ Margin Resilience: Pro forma operating margin expanded 70bps to 37% for full year despite absorbing 95bps of tariff headwinds. This demonstrates pricing power and operating leverage—management guided to 120bps of tariff impact in 2026 yet margins should still hold.

🚩 China Competition Intensifies: Management explicitly stated “robotic competition in China intensified in Q4” with provincial tenders showing preference for local suppliers and lower pricing impacting win rates. This is the first time Intuitive has acknowledged material competitive pressure in China.

🔮 ASC Strategy Unveiled: The XiR system is positioned as a “sizable long-term opportunity” for ambulatory surgery centers, with 70% of the ASC opportunity coming from facilities already affiliated with Intuitive’s IDN customers. This leverages existing relationships to unlock a new market segment.

⚡ After-Hours Surge: Emergency procedures (cholecystectomy, appendectomy) grew 35% YoY—the fastest growth in any procedure category. This indicates robotic surgery is penetrating acute care settings, not just scheduled elective procedures, expanding the addressable market.

📦 Ion Platform Accelerating: Ion lung biopsy procedures grew 44% YoY with the installed base approaching 1,000 systems. Management enhanced disclosure around Ion metrics, signaling confidence this becomes a meaningful growth driver alongside

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 33% OFF Annual Use Code “2026”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

Meta just wrote a check to fast-track its enterprise ambitions with an agent startup that’s already processed 147 trillion tokens. NVIDIA’s getting independent receipts proving its “expensive” chips actually cost 1/15th per token versus AMD. And Applied Digital is carving out its cloud business into a reverse-merger special—complete with a medtech shell company and a 97% ownership stake. — Let’s Chip In.

What The Chip Happened?
🤖 Meta Acquires 147T-Token Agent Beast to Chase Enterprise Dollars
⚡ NVIDIA’s GB200 Delivers 28x Performance—And 15x Better Economics
🔄 Applied Digital Spins Cloud Biz Into ChronoScale via Reverse Merger
Read time: 7 minutes

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Meta Platforms Inc. (NASDAQ: META)
🤖 Meta Buys Manus AI: 147 Trillion Tokens and an Enterprise Appetite

What The Chip: Meta scooped up Manus, an autonomous AI agent startup that’s already processed over 147 trillion tokens and spun up 80 million virtual computers since launching its General AI Agent earlier this year. Unlike typical acqui-hires where the product gets sunset, Manus will keep operating as a standalone service while its tech gets baked into Meta AI and the company’s broader consumer and business products. This is Meta planting a flag in the enterprise AI agent wars, a market Microsoft and OpenAI currently own.

Details:

🚀 Scale Before Exit: Manus hit 147T+ tokens served and 80M+ virtual computers created in less than a year, demonstrating product-market fit that justified Meta writing the check rather than building from scratch.

🤖 Autonomous Everything: The platform handles market research, coding, data analysis, and complex multi-step workflows independently, not just chat completion, but actual task execution that enterprise customers pay real money for.

💼 Enterprise Focus Explicit: Meta’s announcement specifically highlights “unlocking opportunities for businesses”. This isn’t about making Llama chat better, it’s about chasing business market tools that complement its ad business.

📈 Daily Usage at Scale: Already serving “millions of users and businesses worldwide” daily, meaning Meta acquired a revenue-generating business, not a prototype in someone’s garage.

💰 Standalone Revenue Preserved: Unlike most Meta acquisitions, Manus continues operating and selling its service independently, meaning Meta sees immediate monetization potential worth preserving during integration.

🏗️ Talent Acquisition Multiplier: The Entire Manus team is joining Meta to accelerate general-purpose agent development, acquiring both technology and the team that knows how to ship it at scale.

Why AI/Semiconductor Investors Should Care: This acquisition signals Meta’s serious about diversifying revenue beyond advertising into enterprise AI services. For semiconductor investors, enterprise agent adoption drives sustained high-margin inference compute demand that runs 24/7, unlike chatbot queries. The 147T token milestone provides a reference point for scaling economics as more enterprises deploy autonomous agents. Watch whether Meta leverages its infrastructure cost advantages and Llama ecosystem while maintaining margin, which would accelerate enterprise AI adoption and drive incremental chip demand across the market.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

NVIDIA Corporation (NASDAQ: NVDA)
⚡ Independent Research Proves NVIDIA’s “Expensive” Chips Cost 1/15th Per Token

What The Chip: Signal65, the research firm run by former Intel and AMD executive Ryan Shrout, published a comprehensive third-party analysis demolishing the “NVIDIA is expensive” narrative. Their testing shows NVIDIA’s GB200 NVL72 rack-scale system delivers up to 28x the performance of AMD’s MI355X on frontier reasoning workloads like DeepSeek-R1. Despite nearly 2x higher per-GPU pricing ($16 vs $8.60/hour), NVIDIA’s architecture delivers as low as 1/15th the cost-per-token on high-interactivity AI reasoning, the workloads that now represent over 50% of all tokens served according to OpenRouter data.

Details:

🔬 Performance Crush: GB200 NVL72 delivers 5.9x to 28x better per-GPU performance than AMD MI355X on DeepSeek-R1, depending on required tokens-per-second-per-user—advantage widens dramatically as interactivity demands increase.

💰 TCO Inversion: Despite 1.86x higher per-GPU hourly cost, NVIDIA achieves up to 15x better performance-per-dollar due to throughput advantages. Optimizing for GPU price instead of token economics is the strategic mistake buyers make.

🏗️ Rack-Scale Moat: All 8-GPU systems, both NVIDIA B200 and AMD MI355X, hit communication bottlenecks when scaling MoE reasoning models; only NVL72’s 72-GPU NVLink domain at 130 TB/s solves the expert-routing problem these architectures create.

🤖 MoE Gap Explodes: On dense models like Llama 3.3 (70B), NVIDIA leads by 1.8-6x; on mixture-of-experts reasoning models, that advantage expands to 6-28x, and 12 of the top 16 open-weight models are now MoE architectures.

📈 Reasoning Is The Market: OpenRouter data shows 50%+ of all tokens now route through reasoning models. This isn’t a niche workload; it’s becoming the primary revenue driver for AI infrastructure.

⚡ Interactivity Ceiling: GB200 NVL72 achieves 275+ tokens/sec/user while MI355X peaks at 75 tokens/sec/user. NVIDIA can deliver user experiences that AMD architecturally cannot match today on frontier models.

🚀 Generational Economics: GB200 NVL72 delivers ~20x performance improvement over H100 while costing only 1.67x more per GPU-hour, translating to ~12x better performance-per-dollar versus the prior generation—Hopper customers have a massive upgrade incentive.

🔮 AMD’s Helios Gambit: Report notes AMD’s upcoming Helios rack-scale platform may close gaps “over the next 12 months,” but NVIDIA’s Vera Rubin ships on a similar timeline.

Why AI/Semiconductor Investors Should Care: This independent analysis directly challenges the core bull thesis for AMD and other NVIDIA competitors, that price-per-GPU advantage translates to customer wins. For frontier MoE reasoning workloads that now dominate inference economics, NVIDIA’s rack-scale architecture delivers better unit economics despite higher sticker prices. This validates NVIDIA’s pricing power sustainability and explains why hyperscalers keep choosing NVIDIA despite vocal cost concerns. The catch: This focuses on bleeding-edge reasoning models; AMD may still win cost-sensitive dense model deployments. Watch whether AMD’s Helios platform can narrow the MoE gap before Vera Rubin extends NVIDIA’s lead and whether enterprises prioritize lowest TCO or lowest upfront cost in buying decisions.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Applied Digital Corporation (NASDAQ: APLD)
🔄 Applied Digital Spins GPU Cloud Into ChronoScale via Reverse Merger Special

What The Chip: Applied Digital is carving out its cloud computing division and merging it with Nasdaq-listed EKSO Bionics, a medical exoskeleton company, in a reverse merger that creates ChronoScale Corporation, a pure-play GPU cloud platform targeting AI workloads. Post-closing, APLD shareholders would own roughly 97% of the new entity, while Applied Digital would retain its data center development business. The deal structure is classic SPAC-adjacent financial engineering: separate the high-multiple cloud story from the capital-intensive real estate story, give each its own public currency, and let the market re-rate both on a sum-of-the-parts basis. EKSO gets to explore selling its legacy medtech business while providing the shell.

Details:

📊 Meaningful Scale: Applied Digital Cloud generated $75.2 million in trailing twelve-month revenue as of August 31, 2025. This isn’t a startup spin; it’s a business with real revenue exiting the parent.

🚀 H100 Early Mover: Applied Digital Cloud was “among the first platforms to deploy NVIDIA’s H100 GPUs at scale in 2023”. Securing allocation early matters in capacity-constrained markets, a legacy advantage for ChronoScale.

🔄 Deal Structure: Non-binding term sheet for business combination; Applied Digital Cloud merges with EKSO, creating ChronoScale with APLD owning ~97%, essentially a dividend of new equity to existing shareholders.

🏗️ Strategic Separation Logic: Data center ownership (capital-intensive, long-cycle development) and cloud operations (software-forward, faster growth) get to “scale independently with distinct growth trajectories and capital flexibility.”

🔌 Synergy Preservation: ChronoScale expected to maintain “advantaged access” to Applied Digital’s expanding AI factory campus portfolio for faster capacity deployment—keeps strategic tie while separating equity structures.

⏰ Timeline and Contingencies: Expected H1 2026 close, subject to due diligence, binding docs, regulatory approval, shareholder votes—deal is non-binding, terms could shift or crater entirely before closing.

💼 EKSO’s Exit Path: EKSO Bionics continues to operate post-merger while exploring the sale of its legacy medtech business, a classic shell company that provides a public listing vehicle for private assets.

Why AI/Semiconductor Investors Should Care: This spin-merge creates a pure-play AI infrastructure public equity, potentially unlocking valuation multiple expansion if the market assigns ChronoScale a CoreWeave-style premium versus Applied Digital’s current blended multiple. The bull case: GPU cloud businesses trade at higher multiples than data center REITs, separation captures that arbitrage. The bear case: At $75M TTM revenue, ChronoScale would be small-cap in capital-intensive infrastructure competing against hyperscalers with infinite balance sheets. Watch final deal terms, any committed financing, customer concentration disclosure, and whether APLD announces additional GPU allocation commitments. If executed well, this could be legitimate value unlock; if botched, it’s financial engineering distracting from operating fundamentals.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

Google just dropped $4.75B on its own power company, Alibaba’s eyeing a half-billion-dollar AMD chip order, and Nvidia’s racing to ship 80,000 H200s to China before Lunar New Year. The AI infrastructure race is no longer just about silicon, it’s about securing the power to run it and the geopolitical access to sell it. — Let’s Chip In.

What The Chip Happened?

🔌 Google Buys the Power Company: $4.75B Intersect Deal
🇨🇳 AMD’s China Comeback: Alibaba Eyes 50K MI308 Chips
🔓 Nvidia’s H200 China Gambit: 80K Chips by February
Read time: 7 minutes

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Alphabet Inc. (NASDAQ: GOOGL)
🔌 Google Drops $4.75B on Its Own Power Company to Fuel AI Dominance

What The Chip: On December 22, Alphabet announced it’s acquiring Intersect Power for $4.75 billion cash plus debt assumption, one of its largest acquisitions ever. This isn’t about search or cloud or quantum computers. It’s about raw electricity. Google is tired of waiting on utilities that can’t keep pace with AI’s hunger for power, so it bought multiple gigawatts of generation capacity and the team that knows how to build more. The deal transforms Google from a power customer into a power producer.

Details:

⚡ The Asset Haul: Intersect brings multiple gigawatts of energy projects in development or under construction, including Google’s existing partnership sites. By 2028, Intersect projects totaling 10.8 gigawatts are expected online, which is more than 20 times the Hoover Dam’s output. That’s serious infrastructure.

🏗️ Texas Anchors the Strategy: The crown jewel is Project Quantum in Haskell County, Texas, a 840 MW of solar PV and 1.3 GWh of battery storage, co-located with data centers. Google previously committed $40 billion to Texas through 2027, and this acquisition accelerates that buildout with dedicated power generation on-site.

🔋 Tesla Megapacks at Scale: Intersect secured 17.7 gigawatt-hours of Tesla Megapacks, making it one of the world’s largest buyers. Pair that with a multi-billion dollar First Solar partnership for American-made panels, and you’ve got a vertically integrated clean energy supply chain purpose-built for AI.

📦 What Stays, What Goes: The deal excludes Intersect’s existing Texas and California operating assets, which remain independent under TPG Rise Climate and other investors. Google gets the development pipeline and team, essentially buying future capacity, not legacy revenue streams.

💰 Google Already Had Skin in the Game: Alphabet held a minority stake after leading an $800M funding round in December 2024, targeting $20B in total investment by 2030. This acquisition fast-tracks that timeline and gives Google full control over execution.

🌍 Competitive Context: Amazon, Microsoft, and Meta have each announced multi-billion dollar data center investments requiring dedicated energy. Google’s moving first to own the entire stack, generation, storage, and data center infrastructure, while competitors still negotiate with utilities and independent power producers.

🔮 Beyond Solar and Batteries: Alphabet isn’t stopping at renewables. The company’s exploring advanced geothermal, long-duration energy storage, and gas with carbon capture. Translation: Google’s hedging its energy bets across every viable technology to ensure no bottleneck stalls AI scaling.

🚩 Integration Risk is Real: Google’s core competency is software and chips, not managing gigawatt-scale energy projects. Construction delays, permitting issues, and grid interconnection challenges could push timelines and balloon costs. Energy infrastructure is notoriously complex.

🎯 Timing and Regulatory Hurdles: Deal close is expected in H1 2026, subject to regulatory approval. Any delays extend Google’s dependence on third-party power while competitors potentially lock up their own capacity deals.

⚠️ The Bear Case is Simple: If energy buildout lags AI compute deployment, Google ends up with expensive power projects generating electricity for data centers that aren’t ready, or worse, stranded assets if AI demand doesn’t materialize as expected.

Why AI/Semiconductor Investors Should Care: Energy is the new chokepoint in AI. Google’s betting $4.75B that owning the power supply unlocks faster scaling than waiting on utilities. If Intersect delivers gigawatts on schedule, Google gains a structural advantage for training bigger models, deploying more inference capacity, and offering cloud customers guaranteed power availability. That’s a moat competitors can’t easily replicate. Watch for project timelines post-close and whether Microsoft or Amazon respond with their own energy M&A. The AI race just became an energy race, and Google fired the starting gun.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Advanced Micro Devices (NASDAQ: AMD)
🇨🇳 AMD’s Lisa Su Lands Alibaba’s 50K Chip Order After China Crackdown Thaw

What The Chip: AMD’s China-compliant MI308 accelerator is moving from vaporware to volume orders. Alibaba is reportedly considering purchasing 40,000 to 50,000 MI308 chips, a potential $500M+ deal that would mark one of AMD’s largest single Chinese orders since export controls began. The timing is perfect: President Trump just approved limited GPU sales with a 25% fee, and AMD CEO Lisa Su personally met with China’s Commerce Minister last week. If confirmed, this validates AMD’s strategy of building export-compliant chips while Nvidia dominated headlines.

Details:

💰 The Numbers Tell the Story: 40,000 to 50,000 MI308 accelerators would generate hundreds of millions in revenue, likely $500M+ depending on pricing. That’s meaningful scale for AMD’s data center segment, especially in a market where bulls questioned China relevance after export bans.

🛡️ MI308: The Compliance Play: The MI308 isn’t AMD’s flagship. It’s specifically designed to fall within U.S. export control performance limits, meaning it’s deliberately downgraded for the Chinese market. AMD’s playing within the rules, not pushing boundaries—a calculated move to avoid regulatory blowback.

📉 MI325: While the article reports that the chip will be the MI308, it is possible that the chip could be the MI325, as it's the equivalent of Nvidia’s H200, which just recently got approved by the current administration.

🤝 Lisa Su’s Diplomatic Tour: Su met with China’s Commerce Minister Wang Wentao last week to discuss operations and cooperation. No specifics were disclosed, but the timing, immediately before this order leaked, suggests the meeting wasn’t ceremonial. Su’s building relationships while competitors wait on policy clarity.

🚀 Guidance Upside Potential: AMD hasn’t included meaningful China revenue in recent forecasts, meaning a confirmed Alibaba deal would likely trigger upward revisions. That’s a catalyst the stock desperately needs after getting hammered on AI bubble fears.

🔵 MLex is the Sole Source: Currently, only MLex is reporting specific Alibaba-MI308 figures. Neither AMD nor Alibaba has commented publicly. Until confirmed, this remains a well-sourced rumor, not a guaranteed order. Investors should temper excitement until official announcement.

🎯 Alibaba’s AI Ambitions: Alibaba’s rapidly expanding cloud and large language model capabilities, which demand high-performance accelerators. The MI308 order aligns with their broader AI infrastructure buildout, signaling they’re prioritizing compute capacity for 2025-2026 deployments.

⚠️ Export Policy Remains Fluid: Trump’s approval for limited sales is reversible. Congressional scrutiny is intensifying, and any policy shift could freeze shipments mid-order. AMD’s exposure to Chinese policy whiplash is real.

Why AI/Semiconductor Investors Should Care: This order validates AMD’s China strategy at a moment when bears questioned its relevance. Even limited access to Chinese hyperscalers represents a multi-billion-dollar TAM, and AMD’s winning share despite being the underdog to Nvidia. If Alibaba confirms, expect analyst upgrades and momentum in AMD’s data center narrative. The stock needs positive catalysts after recent weakness, and a $500M order delivers exactly that. Watch for official confirmation, additional Chinese hyperscaler orders, and any MI308 follow-on capacity announcements.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

NVIDIA Corporation (NASDAQ: NVDA)
🔓 Nvidia Races to Ship 80,000 H200 Chips to China Before Lunar New Year

What The Chip: Nvidia’s targeting mid-February shipments of H200 chips to China, according to a report today by Reuters. The first deliveries of its second-most-powerful AI accelerators since Trump greenlit exports with a 25% fee. Initial shipments could total 40,000 to 80,000 H200 chips from existing inventory, with new production capacity opening Q2 2026. For Alibaba, ByteDance, and other Chinese AI giants, this is a game-changer: H200s are six times more powerful than the downgraded H20s they’ve been stuck with. But dual approval from Washington and Beijing is far from guaranteed.

Details:

📦 Initial Inventory Drop: Nvidia’s fulfilling first orders from existing stock, 5,000 to 10,000 chips, translating to 40,000 to 80,000 H200 chips. That’s immediate revenue recognition if shipments clear customs, no waiting on new production ramps.

🔮 Q2 2026 Capacity Expansion: Beyond the initial batch, Nvidia’s adding new H200 production capacity with orders opening second quarter 2026. That signals confidence in sustained Chinese demand despite political uncertainty, Nvidia’s betting this isn’t a one-time opening.

💰 25% Government Fee: The U.S. takes a quarter of every sale. That’s a significant margin hit, but access to China’s AI market justifies the toll. Nvidia’s effectively paying for permission to compete against AMD and domestic Chinese alternatives.

🏛️ Interagency Review in Progress: Commerce Department sent license applications to State, Energy, and Defense for review, standard 30-day process. If agencies disagree, Trump makes the final call. That’s a political wild card that could swing either direction based on headlines and lobbying.

⚡ H200 vs. H20 Performance Gap: Chinese customers have been limping along with H20s, deliberately crippled chips designed to meet export thresholds. H200s deliver six times the performance, fundamentally changing what AI models Chinese firms can train and deploy.

🇨🇳 Beijing’s Hesitation: Chinese officials held emergency meetings this month, debating whether to allow imports. One proposal requires bundling each H200 purchase with domestic chips at a set ratio, forcing customers to buy inferior local silicon alongside Nvidia’s premium hardware.

🚩 Domestic Chip Industry Pressure: China’s invested billions in homegrown AI accelerators, and allowing H200 imports risks undercutting that progress. Beijing’s caught between satisfying Alibaba and ByteDance’s immediate needs and protecting long-term strategic independence. No easy answers.

📈 Market Reaction: Nvidia jumped 3.93% Monday on export approval progress. Analysts maintain Strong Buy consensus (39 Buys, 1 Hold, 1 Sell) with an average $263.58 target—45.6% upside from current levels.

⚠️ Congressional Opposition Brewing: Critics argue large-scale H200 exports represent a strategic error, handing China capabilities that narrow the AI gap. The proposed SAFE Chips Act could reverse course, and vocal opposition from defense hawks could pressure Trump to tighten restrictions.

Why AI/Semiconductor Investors Should Care: Reopening China, even partially, expands Nvidia’s addressable market by billions and validates the bull thesis that U.S. chips remain essential despite Chinese domestic efforts. If shipments proceed, Nvidia locks in revenue from customers who’ve been sidelined for quarters, and competitors face an even steeper uphill battle. But dual approval from Washington and Beijing is uncertain, and Congressional scrutiny is intensifying. This is high-reward, high-risk, watch for license approvals, Beijing’s bundling requirements, and any SAFE Chips Act momentum. If 80,000 H200s ship by February, Nvidia’s China comeback is real. If approvals stall, it’s back to export control limbo.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

Palantir just weaponized Accenture’s enterprise rolodex with a 2,000-person deployment force, OpenAI claimed the image generation crown with GPT Image 1.5 while quadrupling speed, and the robotaxi wars exploded as Waymo chases a $100B valuation while Tesla removes safety drivers in Austin. — Let’s Chip In.

What The Chip Happened?

🤝 Palantir Weaponizes Accenture’s Rolodex With 2,000-Person AI Blitz
🎨 OpenAI Drops GPT Image 1.5: 4x Faster and #1 Ranked
🚗 Waymo Eyes $100B While Tesla Goes Full Driverless in Texas
Read time: 7 minutes

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Palantir Technologies (NASDAQ: PLTR)
🤝 Palantir and Accenture Launch 2,000-Person AI Army for Enterprise Domination

What The Chip: Palantir just solved its biggest growth problem, limited implementation capacity and painfully long enterprise sales cycles. The company announced a massively expanded partnership with Accenture, creating the dedicated “Accenture Palantir Business Group” staffed with over 2,000 Palantir-trained professionals plus embedded Forward Deployed Engineers. Accenture is now Palantir’s “preferred global partner” for enterprise transformation, giving PLTR instant access to Accenture’s massive client base across healthcare, telecom, manufacturing, consumer goods, and financial services.

Details:

🚀 Partnership Structure: The new “Accenture Palantir Business Group” operates as a dedicated unit with joint FDEs working side-by-side with enterprise clients. This isn’t a loose referral arrangement—it’s a fully integrated deployment engine that Palantir could never build alone.

💪 Scale of Firepower: Over 2,000 Accenture professionals are now Palantir-skilled, creating a scalable implementation force that addresses PLTR’s historical constraint. These aren’t just salespeople, they’re technical resources who can actually deploy Foundry and AIP at enterprise scale.

🎯 Strategic Designation: Accenture named as Palantir’s “preferred global partner” for enterprise transformation. In Palantir’s typically exclusive ecosystem, this designation signals serious strategic alignment and likely preferential economics.

📈 Early Traction: The partnership is already showing momentum in government, energy, and oil & gas sectors. Now expanding deliberately into healthcare, telecommunications, manufacturing, consumer goods, and financial services, all massive TAMs.

💡 Value Proposition: Tackles the enterprise pain point of “siloed data” by combining Accenture’s implementation muscle with Palantir’s integration platform. The pitch is compelling: unified AI-powered decision making across previously disconnected systems.

⚡ Speed to Value: Partnership explicitly designed to help enterprises “transform at speed and scale”, addressing feedback that Palantir implementations take too long. With 2,000+ trained implementers, time-to-value should compress meaningfully.

🛡️ Platform Lock-In: Deployments leverage Palantir Foundry and AIP as the core platform. Once enterprises standardize on these for AI infrastructure management, switching costs become prohibitive, exactly the moat Palantir wants.

🔮 Go-to-Market Acceleration: Accenture already has relationships with most Fortune 500 companies. Instead of Palantir cold-calling for 18 months, Accenture can embed PLTR into existing transformation contracts. Sales cycle compression is the real prize here.

Why AI/Semiconductor Investors Should Care: This partnership could be the critical point for Palantir’s commercial business. The company has demonstrated product-market fit but struggled with implementation scalability, 2,000 trained Accenture professionals solves that overnight. Watch commercial customer count and average contract values over the next 2-3 quarters. If this partnership actually delivers bookings acceleration, PLTR’s commercial growth rate (already strong) could surprise to the upside. The bear case: revenue-sharing with Accenture will pressure margins, and Palantir’s premium positioning risks commoditization when bundled into broader consulting deals. But if Karp is willing to sacrifice some margin for distribution scale, that’s a sign of confidence in the TAM.

OpenAI
🎨 OpenAI Drops GPT Image 1.5: 4x Faster, #1 Ranked, Game Over for Banana?

What The Chip: OpenAI just dropped GPT Image 1.5, and it’s not subtle, 4x faster image generation while claiming the #1 spot in both text-to-image AND image editing on Artificial Analysis’s Image Arena. The model excels at surgical photo editing, changing only what you request while preserving lighting, composition, and fine details. Available now to all ChatGPT users and via API, this positions ChatGPT as a “creative studio in your pocket” and puts serious pressure on Google, Midjourney, Adobe Firefly, and Stability AI.

Details:

⚡ Speed Blitz: Image generation is now up to 4x faster than previous ChatGPT capabilities. In a market where latency kills user experience, this matters, especially for iterative creative workflows where users generate dozens of variations.

🏆 Benchmark Dominance: GPT Image 1.5 ranks #1 in BOTH Text-to-Image AND Image Editing categories on Artificial Analysis Image Arena, surpassing Nano Banana Pro and other leading models. OpenAI isn’t just competitive, they’re setting the benchmark.

💼 Enterprise Use Cases: Practical applications unlock immediately: clothing/hairstyle try-ons for e-commerce, stylistic filters for marketing, conceptual transformations for product design. These aren’t hobbyist features, they’re monetizable B2B capabilities.

🔌 API Availability: Developers can access “GPT Image 1.5” via API immediately. This enables third-party applications to integrate best-in-class image generation without building competing infrastructure.

💰 Aggressive Pricing: Token-based pricing runs approximately $133 per 1,000 images at high quality (1MP), down to $9 per 1,000 images at low quality. That high-quality price point could pressure margins if usage scales, but it’s competitive with Adobe and significantly cheaper than human designers.

🚀 Competitive Kill Shot: This directly targets Midjourney (best image quality), Adobe Firefly (professional editing), and Stability AI (open-source alternative). OpenAI is bundling best-in-class generation AND editing into the world’s most-used AI interface.

Why AI/Semiconductor Investors Should Care: OpenAI is methodically closing the moat around creative AI tools. Achieving #1 rankings across categories while embedded in ChatGPT creates a distribution advantage competitors can’t match. Watch for enterprise adoption metrics and API usage growth. If this accelerates OpenAI’s path to profitability ahead of rumored 2025 funding rounds, it strengthens the bull case for the entire AI application layer. The risk: aggressive pricing ($133/1k images) and delayed Business/Enterprise rollout hint at capacity constraints or margin pressure.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Alphabet (NASDAQ: GOOGL) / Tesla (NASDAQ: TSLA)
🚗 Waymo Eyes $100B Valuation as Tesla Goes Fully Driverless in Texas

What The Chip: The autonomous vehicle wars just went from simmer to boil. Waymo is in discussions to raise more than $15 billion at a valuation approaching $100 billion, more than double its $45 billion October 2024 valuation. Simultaneously, Tesla is testing driverless vehicles in Austin without safety drivers, six months after launching limited Robotaxi service with humans on board. These twin developments mark the commercialization inflection point for self-driving technology.

Details:

💰 Waymo Valuation Explosion: From $45B in October 2024 to potentially $100-110B in early 2026. That’s $55B+ in value creation in roughly one year, driven by operational scale, safety data, and commercial traction. For GOOGL shareholders, this could be a massive value unlock via spinoff or IPO.

🚀 Waymo Ride Volume Surge: Now delivering 450,000 weekly driverless rides across Austin, Phoenix, San Francisco, Los Angeles, and Atlanta, an 80% increase from 250,000 rides six months ago. This isn’t pilot program scale anymore; it’s a real transportation network.

🛡️ Safety Moat: Waymo’s robotaxis were involved in 90% fewer serious injury or worse crashes compared to human drivers. That safety record is the regulatory and insurance moat that justifies a $100B valuation. Data wins policy arguments.

🌍 Aggressive Expansion: Waymo is launching in St. Louis, Pittsburgh, Baltimore, Philadelphia, New Orleans, Tampa, and Minneapolis. Plus international expansion to London (rides starting 2026) and Tokyo (vehicle training underway). This is a global scaling play.

📏 Operational Scale: 127 million miles driven without a human driver through September, including 56.5M in Phoenix and 38.8M in San Francisco. Fleet size exceeds 2,500 vehicles. Waymo has more autonomous miles than any competitor, and it’s not close.

⚡ Tesla Goes Driverless: A Tesla vehicle was spotted driving Austin public roads without anyone in the driver’s seat or safety monitor. Musk confirmed testing began “within the next three weeks” after his earlier announcement. This is the make-or-break moment for Tesla’s camera-only approach.

💵 Cost Advantage: Morgan Stanley estimates Tesla’s robotaxi cost at $0.81 per mile versus Waymo’s $1.36-$1.43 per mile. If Tesla can validate safety at that cost structure, they win on unit economics. Morgan Stanley expects Waymo to narrow the gap once next-gen hardware scales in 2027.

🚩 Tesla Safety Questions: Seven collisions reported in Tesla’s Austin fleet as of mid-October. NHTSA launched an investigation into 2.88M Tesla vehicles in October for traffic violations and dangerous FSD maneuvers. Safety validation is the bottleneck.

⚠️ Regulatory Window Closing: Texas requirements change in May 2026 with Senate Bill 2807, autonomous-vehicle operators will require DMV authorization for commercial use. Tesla has a six-month window to prove driverless capability before regulations tighten.

Why AI/Semiconductor Investors Should Care: Waymo’s $100B valuation would unlock massive value for GOOGL shareholders, potentially $70B+ in equity value if marked-to-market. An IPO or spinoff in 2026 could be one of the decade’s largest tech offerings. For TSLA, the stakes are existential: Musk has bet the company’s narrative on FSD and robotaxis delivering a $5 trillion valuation. The next six months determine if that’s vision or vaporware. Watch Waymo’s funding close (expected early 2026), Tesla’s timeline to paying passengers in truly driverless vehicles, and NHTSA investigation outcomes.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

NVIDIA just absorbed the orchestration software running half the world’s supercomputers and dropped open models designed to power AI agent armies, pretty much vertical integration on steroids. Meanwhile, Micron’s about to report earnings into a memory market so tight that analysts are falling over themselves to slap $300 targets on the stock. The AI infrastructure buildout isn’t slowing down; it’s getting more concentrated. — Let’s Chip In.

What The Chip Happened?

🔧 NVIDIA Swallows the Supercomputer Scheduler and Promises to Play Nice
🤖 NVIDIA Goes Open-Source (But Keeps the GPU Moat Intact)
💾 Micron Earnings: Memory Shortage Meets Sky-High Expectations
Read time: 7 minutes

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

NVIDIA Corporation (NASDAQ: NVDA)
🔧 NVIDIA Acquires SchedMD, the Software Powering Half the World’s Top Supercomputers

What The Chip: NVIDIA announced the acquisition of SchedMD, creator of Slurm, the open-source workload management system that orchestrates over half of the TOP500 supercomputers globally. This isn’t just another tuck-in acquisition; NVIDIA now controls the scheduling layer that determines how AI training and inference workloads run across massive compute clusters. The company committed to keeping Slurm open-source and vendor-neutral, but make no mistake: NVIDIA just grabbed the traffic controller for the world’s most powerful AI infrastructure.

Details:

🎯 Dominant Market Position: Slurm manages workloads on more than half of both the top 10 and top 100 supercomputers in the TOP500 list, making it the de facto standard for HPC and AI cluster orchestration. Making it a mission-critical infrastructure.

🤝 Decade in the Making: NVIDIA and SchedMD have collaborated for over 10 years, meaning this acquisition formalizes a relationship that already shaped how AI clusters operate. NVIDIA knows the codebase, the team, and the strategic leverage points.

🛡️ Open-Source Commitment (For Now): NVIDIA explicitly pledged to continue Slurm development as open-source, vendor-neutral software supporting heterogeneous hardware environments, not just NVIDIA GPUs.

🚀 AI Training Bottleneck Solved: As clusters scale to 100,000+ GPUs, efficient resource allocation becomes the difference between economical AI training and burning cash. Slurm excels at scalability, throughput, and complex policy management, the exact capabilities needed for foundation model development.

💼 Enterprise Customer Goldmine: SchedMD serves hundreds of customers spanning cloud providers, AI labs, autonomous driving companies, healthcare, energy, financial services, and government agencies. NVIDIA just inherited those relationships.

🔌 Vertical Integration Play: NVIDIA now controls critical layers across the AI stack: GPUs (hardware), CUDA (software framework), NeMo (training tools), and now Slurm (workload orchestration). Each layer reinforces the others, deepening customer lock-in.

🌍 Heterogeneous Support Maintains Ecosystem Buy-In: By continuing to support non-NVIDIA hardware, NVIDIA avoids immediate customer backlash and regulatory scrutiny. But over time, one might expect NVIDIA-specific optimizations that make Slurm work best on NVIDIA silicon.

Why AI/Semiconductor Investors Should Care: This is vertical integration disguised as an open-source play. NVIDIA now owns the orchestration layer that sits between AI workloads and compute hardware, giving it unparalleled leverage as clusters scale. The open-source commitment buys goodwill and avoids antitrust heat, but the strategic position is clear: NVIDIA controls more of the stack than ever. Watch whether competitors respond by backing alternative schedulers, and whether NVIDIA introduces “enterprise” features that fragment the open-source offering. If Slurm remains truly neutral, NVIDIA gains influence; if it tilts toward NVIDIA hardware, the company gains pricing power. Either way, this acquisition tightens NVIDIA’s grip on AI infrastructure at exactly the moment when orchestration complexity is exploding.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

NVIDIA Corporation (NASDAQ: NVDA)
🤖 NVIDIA Drops Nemotron 3: Open Models Built to Power the Agentic AI Era

What The Chip: NVIDIA launched the Nemotron 3 model family, Nano, Super, and Ultra, designed specifically for multi-agent AI systems. These are open-source models, NVIDIA bundled training datasets (3 trillion tokens), reinforcement learning libraries (NeMo Gym, NeMo RL), safety tools, and enterprise deployment infrastructure. Early adopters, including Perplexity, ServiceNow, Palantir, CrowdStrike, and Accenture, are already integrating Nemotron as they develop specialized AI agents to automate complex enterprise workflows.

Details:

⚡ Architecture Breakthrough: Nemotron 3 uses a hybrid latent mixture-of-experts (MoE) design that activates only a fraction of total parameters per token: Nano (30B total/3B active), Super (~100B/10B active), Ultra (~500B/50B active). This delivers frontier-model performance at a fraction of the compute cost.

💰 4x Efficiency Gains: Nemotron 3 Nano delivers 4x higher token throughput versus Nemotron 2 Nano and reduces reasoning-token generation by up to 60%, directly attacking the inference cost problem that has plagued large-scale AI deployments.

🔮 1-Million-Token Context Window: The massive context window enables long-horizon, multistep agent workflows without context drift, critical for complex enterprise automation like multi-day project management, legal document analysis, or supply chain optimization.

⚛️ Training Innovation on Blackwell: Super and Ultra models leverage NVIDIA’s NVFP4 (4-bit floating-point) training format on Blackwell architecture, slashing memory requirements while maintaining accuracy parity with higher-precision formats. This is NVIDIA showing off Blackwell’s capabilities while making a case for hardware upgrades.

🌍 Open Ecosystem Play: NVIDIA released 3 trillion tokens of training data, open-source RL libraries, safety evaluation tools, and the Nemotron Agentic Safety Dataset. Nvidia is building dependency. Developers train on NVIDIA tools, deploy on NVIDIA infrastructure, and run inference on NVIDIA GPUs.

📈 Enterprise Adoption Wave: Early adopters span cybersecurity (CrowdStrike), EDA (Cadence, Synopsys), consulting (Accenture, Deloitte, EY), and workflow automation (ServiceNow, Zoom, Palantir). These aren’t startup companies, they’re production deployments signaling real enterprise demand for agentic AI.

🔌 Cloud Distribution Strategy: Nemotron models are coming to Amazon Bedrock, Google Cloud, Microsoft Azure Foundry, CoreWeave, and others, plus NIM microservices for on-prem deployment. NVIDIA is ensuring Nemotron runs everywhere customers want to deploy AI.

🛡️ Sovereign AI Alignment: Nemotron explicitly supports NVIDIA’s sovereign AI push, with adoption from Europe to South Korea for locally-controlled, regulation-compliant AI development. As governments demand domestic AI capabilities, NVIDIA provides the model layer.

🟢 Availability Timeline: Nano is available now on Hugging Face and via inference providers like Baseten, DeepInfra, Fireworks, and Together AI. Super and Ultra models expected H1 2026, conveniently timed with Blackwell volume production ramp.

Why AI/Semiconductor Investors Should Care: NVIDIA is building a moat around agentic AI before the category fully materializes. Open-sourcing models might seem like giving away the farm, but it’s actually a brilliant strategy: NVIDIA captures value at the hardware layer (GPUs for training and inference), the software layer (NeMo tools), and the services layer (NIM deployment). Every enterprise deploying Nemotron agents is a customer for NVIDIA compute. The real test is whether Super and Ultra deliver on time in H1 2026 and whether Nemotron becomes the default “efficiency model” in hybrid routing architectures, where agents switch between open models (Nemotron) for routine tasks and proprietary models (GPT, Claude) for complex reasoning. Watch enterprise adoption metrics, model performance benchmarks versus Meta’s Llama and Mistral, and whether hyperscalers launch competing open models to avoid NVIDIA dependency.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Micron Technology (NASDAQ: MU)
💾 Micron Earnings: HBM Sold Out, Analysts Bullish, Expectations Sky-High

What The Chip: Micron reports fiscal Q1 2026 earnings after the bell on December 17, walking into a memory market so tight that TrendForce says supplier inventory has collapsed to two-to-four weeks and SK Hynix warns shortages could last through late 2027. The company’s own guidance calls for ~$12.5B revenue and $3.75 EPS, but Street consensus is clustering higher at $12.6B-$12.86B. Shares are up nearly 200% year-to-date as analysts pile into $300 price targets, framing Micron as the ultimate AI memory play. The setup is perfect, unless the company disappoints or guides cautiously.

Details:

💰 Revenue Beat Setup: Street consensus sits at $12.6B-$12.86B versus Micron’s ~$12.5B guidance, with Citi analyst seeing potential upside to $14B. Expectations are elevated but achievable given the supply environment and pricing power.

📈 EPS Expectations High: Consensus EPS forecast is $3.93-$3.94, implying strong margin expansion. Micron guided gross margin to ~51.5%, up from prior levels, reflecting the shift toward high-value HBM and AI-optimized DRAM.

🔥 HBM Revenue Surge: Fiscal Q4 HBM revenue hit nearly $2B, implying an annualized run rate near $8B. Micron expects to sell out its calendar 2026 HBM supply within months, with HBM3E pricing locked and HBM4 pricing expected to be “significantly higher.” This is pricing power in action.

🚩 Memory Shortage Intensifies: DRAM supplier inventory has collapsed to two-to-four weeks per TrendForce, down sharply from historical norms. SK Hynix commentary suggests shortages persist through late 2027, supporting sustained pricing power.

⚡ Pricing Environment Inflects: Counterpoint Research expects advanced and legacy memory prices to rise ~30% through Q4 2025 and potentially another ~20% in early 2026. Micron is riding a pricing wave that’s just getting started.

🎯 DRAM Market Share Gains: Micron’s DRAM market share increased 3.7 percentage points year-over-year in Q3 2025 to 25.7% per TrendForce. The company is gaining share while the market expands, a rare combination.

🛡️ Strategic Exit from Consumer: Micron announced it will exit the Crucial consumer memory business to improve supply allocation for higher-margin AI-driven data center customers. This signals management is prioritizing profitability over volume.

💡 Capex Ramp for HBM: Capex (net of incentives) is expected to exceed $18B in fiscal 2026, up from $13.8B in fiscal 2025. Micron plans to invest ~$9.6B in a Hiroshima, Japan facility for advanced HBM, with construction starting in 2026. The company is betting big on HBM capacity.

🟢 Analyst Upgrade Wave: Wedbush and Mizuho both lifted price targets to $300 on December 15. UBS moved to $295 from $275, citing stronger DRAM and NAND pricing. Wall Street is converging on the bullish case.

📊 Fiscal 2025 Context: Fiscal 2025 revenue grew nearly 50% to $37.4B, with gross margins expanding 17 percentage points to 41%. Fiscal 2026 consensus calls for $17.27 EPS—if Micron achieves that and trades at 25x earnings, stock price could hit $432 (71% upside potential).

Why AI/Semiconductor Investors Should Care: Micron sits at the intersection of genuine supply shortage, rapid AI-driven product mix improvement, and elevated market expectations. The bull case is straightforward: memory shortage + pricing power + HBM ramp = sustained earnings growth. At 15x forward earnings, the stock looks cheap if the memory supercycle plays out through 2026-2027. The bear case is equally simple: expectations are high after a near-tripling YTD, and any cautious guidance or competitive pressure (especially from SK Hynix in HBM) could trigger a sharp reversal.

Watch three things: (1) HBM revenue trajectory and whether management reconfirms selling out 2026 capacity; (2) DRAM/NAND pricing commentary and whether pricing momentum sustains into H1 2026; (3) competitive dynamics, especially Samsung’s HBM qualification progress with NVIDIA. If guidance disappoints, the stock’s stretched valuation multiples offer limited downside protection. This earnings print will either validate or deflate the AI memory supercycle thesis, there’s no middle ground.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 15% OFF Annual Use Code CHIPS
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

🎁 This week’s FREE premium report is here. This is the same deep-dive analysis our community members get, yours at no cost. Want the numerous reports per week, plus live Q&A, and the full community? Less than $0.70/day with code CHIPS at CLICK HERE

Now let’s get into it. — Let’s Chip In.

What The Chip Happened?

Join WhatTheChipHappened Community — 15% OFF Annual Use Code “CHIPS”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Executive Summary

Oracle delivered a quarter that spooked investors despite extraordinary demand signals. Revenue of $16.1B (+13% YoY constant currency) came in slightly light of expectations, while non-GAAP EPS of $2.26 (+51% YoY) was boosted by a $2.7B one-time gain from selling Oracle’s Ampere stake. The story I’m following isn’t the revenue print; it’s the $68 billion in new RPO signed in a single quarter from Meta, NVIDIA, and others, pushing total backlog to $523B (+433% YoY). Management responded by hiking FY26 CapEx guidance by $15B, triggering a double-digit stock drop as investors digested the cash burn implications. The market wanted AI revenue now; Oracle delivered AI contracts for later (luckily for those that listen to investor analyst events, this isn’t a surprise, the AI revenue is coming in steps).

Why This Quarter Mattered

Investors came in focused on three questions: Can Oracle convert its massive backlog to recognized revenue? How bad is customer concentration risk beyond OpenAI? And how much capital will this AI buildout actually require? This quarter partially answered all three. The Meta and NVIDIA contract announcements diversified the customer narrative. Still, management’s admission that funding needs will be “substantially less than $100B” while raising near-term CapEx by $15B didn’t fully calm financing concerns. The modest revenue result reinforced concerns that investors continually expect incremental CapEx to drive incremental revenue faster than current reality allows.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code “CHIPS”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Key Takeaways

• RPO Explosion Validates AI Positioning: Oracle added more contracted backlog in one quarter than many cloud companies generate annually, driven by contracts with Meta, NVIDIA, and others. Total backlog growth of 433% YoY is unprecedented at Oracle’s scale. [This is the de-risking disclosure bulls needed; the concentration narrative shifts from “OpenAI dependency” to “hyperscaler magnet.”]

• Revenue-Backlog Disconnect Concerns Investors: Despite record bookings, revenue grew only 13% and came in slightly light. Management expects incremental FY27 revenue from Q2 bookings, but the lag between signing and recognizing is frustrating a market that wants AI revenue now. [The conversion timeline, not demand, is the issue.]

• CapEx Guidance Jumps $15B: Oracle raised FY26 CapEx expectations significantly above Q1 forecasts to convert near-term capacity opportunities. Combined with a negative $10B quarterly free cash flow, this is one of the factors that triggered the post-earnings selloff. [The market heard “we need to spend more” not “we have more profitable deals to fund.”]

• Alternative Financing Models Underappreciated: Management emphasized customer-owned chips (no Oracle CapEx) and vendor rental models (payments sync with receipts) as reasons they’ll need substantially less external funding than analyst estimates. But they didn’t quantify the mix, leaving investors guessing. [This is either a legitimate capital efficiency story or hand-waving, we need more disclosure to know which.]

• GPU Capacity Delivery Accelerating: Oracle delivered 50% more GPU capacity vs. Q1, with the Abilene Supercluster on track and AMD MI355 deliveries begun. Capacity transfers between customers in hours due to bare metal architecture. [The fungibility point addresses counterparty risk concerns directly.]

• Applications Acceleration Defies SaaS Gravity: Cloud apps deferred revenue is growing faster than recognized revenue, a leading indicator of acceleration. Oracle’s unified sales organization combining industry and Fusion teams is driving larger, multi-component deals. [Every large SaaS peer is decelerating; Oracle claims to be accelerating. The deferred revenue spread supports the claim.]

• AI Data Platform Is the Long Game: Larry Ellison spent his entire prepared remarks on Oracle’s vision for AI reasoning on private enterprise data, not quarterly results. He positioned Oracle databases as containing “most of the world’s high-value private data” and called this opportunity larger than training AI on public data. [Classic Larry, ignore the quarter, sell the decade.]

• Margins Compressed During Buildout: Operating income grew only 8% while revenue grew 13%. Management acknowledged that margins are temporarily lower because most capacity isn’t online yet. Path to 30-40% AI data center gross margins depends on accelerating capacity delivery, but no specific timeline was provided. [This remains the key investor concern alongside capital requirements.]

Key Quotes

“Remaining performance obligations, or RPO, ended the quarter at $523.3 billion, up 433% from last year and up $68 billion since the end of August, driven by contracts signed with Meta, NVIDIA and others as we continue to diversify our customer backlog.” — Doug Kehring, EVP & Principal Financial Officer

[The headline number that captures the entire demand story in one sentence.]

“Training AI models on public data is the largest, fastest-growing business in history. AI models reasoning on private data will be an even larger and more valuable business. Oracle databases contain most of the world’s high-value private data.” — Larry Ellison, Chairman & CTO

[The strategic thesis in three sentences—Oracle as the platform for enterprise AI, not just GPU rental.]

“We’ve been reading a lot of analyst reports, and we’ve read quite a few that show an expectation of upwards of $100 billion for Oracle to go out and kind of complete these buildouts. And based on what we see right now, we expect we will need less, if not substantially less money raised than that amount.” — Clay Magouyrk, CEO (Infrastructure)

[Directly addressing the biggest bear concern with a specific benchmark, but “substantially less” needs quantification.]

“Just moving a customer to the cloud results in a 3 to 5x annual revenue lift compared to support revenue.” — Mike Sicilia, CEO

[Hidden gem many people aren’t talking about, it quantifies the embedded growth opportunity in Oracle’s massive on-prem install base.]

“Anyone right now with a credit card can show up... and you can spin up a bare metal computer as quickly as a few minutes. And at the end of that you can turn it off, and I will recycle that, and I can hand it to another customer in less than an hour.” — Clay Magouyrk, CEO (Infrastructure)

[Addresses counterparty risk directly, that capacity isn’t stranded if a customer doesn’t pay.]

Key Metrics

Financial Performance

• Total Revenue: $16.1B (+13% YoY constant currency)

• Non-GAAP EPS: $2.26 (+51% YoY)

• GAAP EPS: $2.10 (+86% YoY) (due to one-time Ampere Sale)

• Operating Income: $6.7B (+8% YoY)

• Ampere Sale Gain: $2.7B pretax (one-time)

Cloud Segment

• Total Cloud Revenue: $8B (+33% YoY), now 50% of total revenue

• Cloud Infrastructure (OCI): $4.1B (+66% YoY)

• GPU-Related Revenue: +177% YoY

• Cloud Applications: $3.9B (+11% YoY), $16B annualized run rate

• Cloud Apps Deferred Revenue: +14% YoY

Infrastructure Detail

• Cloud Database Services: +30% YoY

• Autonomous Database: +43% YoY

• Multicloud Consumption: +817% YoY

• Dedicated Region & Alloy: +69% YoY

• Marketplace Consumption: +89% YoY

Applications Detail

• Strategic Back-Office: $2.4B (+16% YoY)

• Fusion ERP: +17% | Fusion SCM: +18% | Fusion HCM: +14%

• NetSuite: +13% | Fusion CX: +12%

• Industry Clouds: +21% YoY

Balance Sheet & Cash Flow

• RPO (Total Backlog): $523.3B (+433% YoY)

• RPO Added in Q2: $68B

• Short-Term RPO Growth: +40% YoY (vs. 25% prior quarter, 21% year-ago)

• Operating Cash Flow: $2.1B

• Free Cash Flow: -$10B

• CapEx: $12B

  

Operational

• OCI Regions Live: 147 (64 more planned)

• Multicloud Regions: 45 live (27 more planned)

• GPU Capacity vs. Q1: +50% QoQ

• Abilene Supercluster: 96,000+ GB200 GPUs delivered

• Clinical AI Agent Customers: 274 live in production

• Cloud App Go-Lives: 330 in quarter

• Capacity Transfer Time: Hours (customer to customer)

• Time to Revenue After Expenses: Couple of months (AI data centers)

• On-Prem to Cloud Revenue Lift: 3-5x annual revenue vs. support

Guidance

• Q3 FY26 Revenue (USD): +19-21% YoY

• Q3 FY26 Cloud Revenue (USD): +40-44% YoY

• Q3 FY26 Non-GAAP EPS (USD): $1.70-$1.74 (+16-18% YoY)

• Q3 FY26 Non-GAAP EPS (CC): $1.64-$1.68 (+12-14% YoY)

• Currency Impact: +2-3% on revenue, +$0.06 on EPS

• FY26 Revenue: $67B (unchanged)

• FY26 CapEx: ~$15B higher than Q1 forecast

• FY27 Incremental Revenue from Q2 RPO: $4B

• Target AI Data Center Gross Margin: 30-40% over contract life

Investment Implications

For Bulls:

This quarter delivered exactly what bulls needed on the demand and diversification fronts, which was record new contracts from Meta, NVIDIA, and others, proving Oracle isn’t just an OpenAI story. The 433% YoY RPO growth and 40% short-term RPO acceleration (vs. 21% last year) provide unprecedented visibility for a company of Oracle’s scale. Applications momentum: deferred revenue growth outpacing recognized revenue, validates the unified selling strategy and AI halo effect. The infrastructure fungibility explanation that capacity can be transferred in hours directly addresses counterparty risk concerns. Management’s emphasis on alternative financing models (customer-owned chips, vendor rentals) suggests capital intensity may be lower than models assume. If you believe the funding needs will genuinely be “substantially less than $100B,” and are bullish in the AI infrastructure race, then the stock reaction creates an opportunity to own a company converting massive AI demand to contracted revenue faster than any enterprise software peer.

For Bears:

The modest revenue print validates concerns that backlog conversion takes longer than investors want. More troubling: operating income grew only 8% while revenue grew 13%, and management offered no timeline for margin normalization beyond “deliver capacity faster.” Negative $10B quarterly free cash flow and a $15B CapEx guidance increase suggest the funding story isn’t as clean as prepared remarks implied. Management’s “substantially less than $100B” claim lacks quantification, they didn’t disclose what percentage of buildout uses capital-light financing models. The sell-side pressed hard on margin timing and capital requirements, and management’s answers, while detailed, were strategically vague on specifics. With Oracle betting the balance sheet on AI infrastructure economics unproven at this scale, the post-earnings selloff suggests the market shares this skepticism until management provides more transparency on the funding mix and margin timeline.

What to Watch

• Short-Term RPO Conversion: Accelerated from 21% → 25% → 40% YoY over three quarters. Continued acceleration = backlog converting to revenue. Deceleration = extended monetization timeline and bear case validation.

• Cloud Apps Deferred vs. Recognized Revenue: The current spread (deferred growing faster than recognized) is the leading indicator for applications acceleration. Gap widening = acceleration thesis confirmed. Gap narrowing = sales reorg didn’t deliver.

• Free Cash Flow Trajectory: -$10B in Q2 with $12B CapEx. As capacity comes online (couple of months from expense to revenue), this should improve. Continued cash burn at this rate = funding concerns intensify. Inflection to positive = narrative shifts.

• Alternative Financing Mix Disclosure: Management needs to quantify what percentage of buildout uses customer-owned chips and vendor rental models vs. Oracle-funded CapEx. Without this, the “<$100B funding” claim remains unverifiable.

My Thoughts

I own Oracle and I’m bullish on it, but that doesn’t mean I think the stock goes up in a straight line. The optics right now look rough: heavy capex, negative free cash flow, low-teens revenue growth, margin compression, more funding needed, and perceived OpenAI dependency. These optics create real fear, and fear creates selling pressure.

The pullback has brought forward P/E down to around 26x, which feels more reasonable than where it was trading a few months ago. I’m planning to add to my position during this weakness over time.

Here’s how I’m thinking about the risk/reward: This is somewhat of a CoreWeave-type bet for me. High growth potential if the AI wave plays out the way I expect, but they’ve gone all-in on the buildout. If AI demand disappoints or the ramp is slower than expected, this stock will have an extremely rough time. I’m comfortable with that risk in my portfolio, but it’s real.

What gives me conviction: As data centers come online, revenue should start catching up to capex, which would completely shift the narrative. On the OpenAI concentration concern, even if OpenAI struggles, I believe whoever wins the AI race will need all the compute they can get. OCI capacity gets used regardless of which customers scale fastest. Finally, Oracle sits on massive enterprise data relationships that create natural upsell opportunities for AI training and inference.

Join WhatTheChipHappened Community — 15% OFF Annual Use Code “CHIPS”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclosures

This analysis is based on publicly available information from Oracle’s Q2 FY2026 earnings call, press release, and related SEC filings. The author and “What The Chip Happened” may hold positions in securities discussed. This is not investment advice. Do your own work.

Some investors might want a bit more data, and the portion below is an optional deep-dive for readers who want more detail. The main report stands alone without this. As I write these reports, certain things might just be too much, and I decide to cut off. Instead, I’ll add them here so the added research doesn’t go to waste.

A. Segment Deep Dive

Cloud Infrastructure ($4.1B, +66% YoY)

The star of the show, and it’s not close. OCI grew faster than the prior quarter despite an already massive base, driven by AI training and inference demand.

Key drivers:

• GPU revenue up 177% YoY—the fastest-growing line item

• Delivered 50% more GPU capacity than Q1

• 96,000+ NVIDIA GB200 GPUs at Abilene Supercluster on track

• AMD MI355 deliveries began this quarter

Mega-customer traction:

• Meta and NVIDIA contracts drove the $68B RPO addition

• Uber surpassed 3 million cores (”highest traffic ever this Halloween”)

• Temu scaled to ~1 million cores for Black Friday/Cyber Monday

Management commentary reveals strategic ambition beyond GPU rental:

“Our diversity of capabilities within infrastructure differentiates us from AI infrastructure neoclouds. Our unique combination of infrastructure and applications differentiates us from other hyperscalers.”

Dedicated Regions & Alloy (+69%):

• 39 live regions, 25 more planned

• New launches: ITHCA Group (Oman), NTT Data, SoftBank

• Dedicated Region 25: full OCI capability in 3-rack footprint

Multicloud Database (consumption +817%):

• 45 regions live across AWS, Azure, GCP; 27 more planned

• New Universal Credits program enables single commitment across clouds

• Channel reseller program launched

• “Billions in identified pipeline”

Assessment: The 66% growth at this scale is remarkable, but the real story is diversification. Oracle is no longer just GPU rental for AI labs—it’s becoming infrastructure for hyperscalers (Meta), AI companies (NVIDIA, xAI, OpenAI), and enterprises running databases across multiple clouds.

Cloud Applications ($3.9B, +11% YoY)

The “boring” business that’s quietly accelerating while every SaaS peer decelerates.

Back-office strength:

• Strategic back-office apps: $2.4B (+16%)

• Fusion ERP +17%, SCM +18%, HCM +14%

Industry vertical momentum:

• Combined industry clouds +21%

• Healthcare: 274 clinical AI agent customers live in production

• AI ambulatory EHR received U.S. regulatory approval

• Q3 expectation: both bookings and revenue to “accelerate materially”

Leading indicator:

• Deferred revenue +14% vs. in-quarter revenue +11%

• 330 cloud app go-lives in quarter (multiple per day)

Mike Sicilia explained the differentiation:

“We are the only applications company in the world that’s selling complete application suites... All of our competitors are largely in the best-of-breed business because they’re not in the applications business in totality. They’re not in the back-office business. They’re not in the industry business and they’re not in everything in between.”

The “1 Oracle” sales reorg—combining industry and Fusion sales teams—is driving larger, multi-component deals:

“This is something we’ve been talking about for many years, that is the synergies between our back-office applications and our industry applications. We’re seeing more and more deals where our industry apps are pulling Fusion, or the Fusion apps are pulling the industry apps. And as a result of seeing more and more deals, we’re also seeing larger deals with more components.”

Assessment: The 14% deferred revenue growth vs. 11% recognized is the number to watch. If Oracle can sustain applications acceleration while peers decelerate, it validates both the unified selling strategy and the AI feature investment (400+ features live in Fusion). The healthcare vertical is the proving ground for AI agents that implement in weeks without professional services.

B. Extended Quotes with Context

“Remaining performance obligations, or RPO, ended the quarter at $523.3 billion, up 433% from last year and up $68 billion since the end of August, driven by contracts signed with Meta, NVIDIA and others as we continue to diversify our customer backlog.” — Doug Kehring, EVP & Principal Financial Officer

Context: This extraordinary backlog growth ($68B added in one quarter) demonstrates unprecedented demand and provides multi-year revenue visibility. The 433% YoY growth is exceptional even for Oracle’s scale.

“Uber has now surpassed 3 million cores on OCI, powering their highest traffic ever this Halloween. Temu scaled to nearly 1 million cores for Black Friday and Cyber Monday.” — Clay Magouyrk, CEO (Infrastructure)

Context: Demonstrates OCI’s ability to handle massive scale and burst workloads for non-AI customers—Oracle isn’t just GPU rental.

“We also delivered 50% more GPU capacity this quarter than Q1... Our pace of capacity delivery continues to accelerate.” — Clay Magouyrk, CEO (Infrastructure)

Context: Directly addresses the core investor concern about converting demand to revenue. Faster delivery = faster revenue recognition = faster margin improvement.

“The good thing is that as I mentioned earlier, we don’t actually incur any expenses for the data centers until they’re actually built up and running...the period of time where we’re incurring expenses without that kind of revenue and the gross margin profile that we talked about is really on the order of a couple of months.” — Clay Magouyrk, CEO (Infrastructure)

Context: Explains the short lag between expense and revenue recognition for AI data centers, supporting the margin improvement thesis.

“As we go through this buildout phase, right now, we’re in a phase of very rapid buildout without the majority of the capacity online, obviously, the aggregate mix is going to be lower. But as we actually get the majority of this capacity online...the best way to improve margins quickly is to actually go out and deliver capacity faster.” — Clay Magouyrk, CEO (Infrastructure)

Context: Explains why margins are temporarily compressed (8% operating income growth vs. 13% revenue growth) and lays out the path to recovery: accelerate capacity delivery.

“Customers can actually bring their own chips. And in those models, Oracle obviously doesn’t have to incur any capital expenditures upfront for that model. Similarly, we have different models that we’re working on with different vendors where some vendors are actually very interested in the model where they rent their capacity rather than selling their capacity.” — Clay Magouyrk, CEO (Infrastructure)

Context: Key to the “substantially less than $100B” funding claim. These alternative financing models dramatically reduce Oracle’s capital requirements but were not quantified.

“In health care, in Q3, we expect both our bookings and our revenue to accelerate materially.” — Mike Sicilia, CEO

Context: Explicit forward commitment that will be scrutinized next quarter. Healthcare AI agents implementing in weeks (vs. months for traditional SaaS) could transform delivery economics.

“The cloud application deferred revenue is up 14%. That is higher than the cloud apps revenue growth of 11%, just to reinforce my earlier statements that we expect continued apps growth acceleration.” — Mike Sicilia, CEO

Context: Deferred revenue is a leading indicator. This growth spread provides concrete evidence for the applications acceleration thesis.

“The result is we now expect $4 billion of additional revenue in FY ‘27. Our full year FY ‘26 revenue expectation of $67 billion remains unchanged. However, given the added RPO this quarter that can be monetized quickly starting next year, we now expect fiscal 2026 CapEx will be about $15 billion higher than we forecasted after Q1.” — Doug Kehring, EVP & Principal Financial Officer

Context: Quantifies the FY27 revenue benefit from record Q2 bookings while signaling increased near-term CapEx investment. Shows management converting demand to revenue quickly.

Subscribe now

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 15% OFF Annual Use Code “CHIPS”
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

Oracle’s AI buildout hits labor and material walls. Broadcom prints a $73 billion backlog that won’t stop growing. And Nvidia gets the green light to sell H200s to China, except Beijing might not pick up the phone. Three stories that reveal where AI infrastructure is overheating, where it’s printing money, and where geopolitics just made everything messier. — Let’s Chip In.

What The Chip Happened?

🚨 Oracle’s $300B OpenAI Dream Hits 2028 Delays
🚀 Broadcom’s AI Revenue Doubles as $73B Backlog Explodes
🇨🇳 Trump Greenlights H200 China Sales—But Beijing Says No Thanks

Read time: 7 minutes

Join WhatTheChipHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Oracle Corporation (NASDAQ: ORCL)

🚨 Oracle’s $300B OpenAI Data Centers Pushed to 2028 Amid Labor Crunch

What The Chip: Reports showing that Oracle just admitted what many feared: building AI infrastructure at hyperscale is harder than signing the contracts. The company pushed back completion dates for several OpenAI data centers from 2027 to 2028, citing shortages of skilled labor and materials. This delay compounds a brutal earnings aftermath where shares crashed 15% post-Q2 and now sit 45% below September highs. The Abilene facility remains on track, but the cracks in Oracle’s $300 billion AI infrastructure buildout are showing.

Details:

⚠️ Delay Timeline: Multiple large-scale AI data centers Oracle is constructing for OpenAI have been pushed from 2027 to 2028 delivery. The facilities are part of the Stargate initiative to deploy two million AI accelerators and 5 GW of power capacity.

🔨 Root Cause Ambiguity: Bloomberg’s sources blame “labor and materials” shortages, but Oracle hasn’t clarified whether this means construction workers, specialized data center equipment, or both. That vagueness is concerning—it suggests Oracle doesn’t control the bottleneck.

🛡️ Oracle’s Denial: The company issued a Friday denial of the delay report, but shares still fell 5% that day following a 10% drop the day before. When your stock is down 45% in three months, denials don’t reassure anyone.

💰 Debt Market Panic: The cost of insuring Oracle’s debt against default hit a five-year high Thursday and climbed again Friday. Credit markets are pricing in real execution risk on a company carrying $111.6 billion in debt.

📈 Capex Surge: Oracle now projects $50 billion in full-year capital expenditures, up from $35 billion as of September. That’s a 43% increase in capex guidance in just one quarter—a sign of either ambition or desperation.

📉 Q2 Revenue Miss: Oracle reported $16.06 billion in revenue versus $16.21 billion expected, though adjusted EPS of $2.26 crushed the $1.64 estimate. The revenue miss is what tanked the stock, AI investors care about top-line growth, not bottom-line engineering.

🚩 Analyst Exodus: At least 13 brokerages slashed price targets post-earnings. Evercore cut its target from $385 to $275 while maintaining Buy, citing “long-term opportunity” but acknowledging near-term capex and leverage friction, Wall Street speak for “this is going to hurt.”

🎯 Abilene Still On Track: The first Stargate data center in Abilene, Texas remains on schedule, with 96,000 Nvidia chips already delivered. One facility doesn’t solve the narrative, but it’s the only execution win Oracle can point to right now.

⚛️ Infrastructure Bottleneck Signal: Analyst Bob O’Donnell nailed it: “Concerns about the ability to build data centers due to construction delays, power availability and other practical factors are becoming a much bigger factor than expected demands for AI capabilities.” Translation: demand isn’t the problem—physics is.

🔮 Concentration Risk: OpenAI itself is unprofitable and projected to spend over $1 trillion through 2030. Oracle’s backlog is impressive until you realize it’s concentrated on a customer burning cash and dependent on Microsoft’s capital patience. Rating agencies Moody’s and S&P have shifted to negative outlooks, warning net debt-to-EBITDA could exceed 4× by 2027-2028.

Why AI/Semiconductor Investors Should Care: Oracle’s stumble exposes the inconvenient truth of the AI buildout: chips aren’t the constraint anymore, construction crews, electrical transformers, and cooling systems are. The company’s $523 billion backlog (up 438% YoY) proves demand is real, but execution risk is mounting. Oracle is borrowing tens of billions to build infrastructure for an unprofitable customer in a market where power, labor, and materials are all bottlenecks.

Bulls will point to the backlog and Meta/NVIDIA commitments. Bears will point to the debt load and delivery delays. Watch infrastructure capex trends across hyperscalers, if Oracle’s problems are isolated, the stock is oversold. If they’re systemic, the entire AI infrastructure thesis needs repricing.

Join WhatTheChipHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Broadcom Inc. (NASDAQ: AVGO)

🚀 Broadcom’s AI Revenue Doubles as $73B Backlog Signals Insatiable Demand

What The Chip: While Oracle struggles to build, Broadcom prints money designing what goes inside. The company crushed Q4 with AI semiconductor revenue surging 74% YoY to $6.5 billion, capping a fiscal year where AI revenue hit $20 billion—a 10x increase over just 11 quarters. The real story is acceleration: Q1 guidance calls for AI revenue to double year-over-year to $8.2 billion, with CEO Hock Tan confirming this pace will continue throughout FY26. A record $73 billion AI backlog stretching 18 months out proves hyperscaler demand shows no signs of slowing.

Details:

⚡ AI Revenue Acceleration: Q1 FY26 AI semiconductor guidance of $8.2B represents 100% YoY growth, up from initial 60-70% FY26 expectations just six months ago. Management buried the lede—they’re not slowing down, they’re speeding up.

💰 $73B Backlog Monster: Broadcom’s AI backlog alone comprises roughly half of its $162B total consolidated backlog, to be delivered over 18 months. Tan emphasized the backlog “keeps growing”—meaning the $73B is a floor, not a ceiling.

🤖 XPU Customer Expansion: Broadcom now serves five custom AI chip customers. The fourth customer (believed to be Anthropic) placed a $10B order in Q3 followed by another $11B in Q4. A fifth customer added a $1B order—Broadcom is becoming the custom silicon design house for every hyperscaler building alternatives to Nvidia.

🔥 Tomahawk 6 Demand: The 102 Tbps switch is seeing “bookings of a nature we have never seen,” per Tan. It’s the fastest deployment ramp for any switch product in company history—a sign that networking, not just compute, is the new gold rush.

🛡️ VMware Cash Machine: Infrastructure software delivered $6.9B revenue (beating $6.7B guidance) at 78% operating margin, up from 72% YoY. The VMware integration is complete, and it’s pure profit now—providing a margin cushion for AI semiconductor investments.

📈 Free Cash Flow Dominance: FY25 free cash flow hit $26.9B (42% of revenue), up 39% YoY. Broadcom returned $17.5B to shareholders via dividends and buybacks while funding $11B in R&D—this is what capital efficiency looks like.

💡 Dividend Aristocrat: 15th consecutive annual dividend increase, raising the quarterly payout 10% to $0.65/share ($2.60 annualized). Even as Broadcom invests billions in AI, it’s rewarding shareholders—confidence in sustainable cash generation.

⚠️ Margin Compression Coming: Gross margin expected to compress ~100bps in Q1 as AI mix rises and systems shipments begin in H2. This is the bull case headwind—more revenue from lower-margin products means profitability metrics will lag revenue growth.

🌍 Supply Response: Broadcom is building advanced packaging capacity in Singapore to address multi-chip custom accelerator demand.

🔵 Non-AI Stabilization: Non-AI semis flat YoY at ~$4.1B guidance. Broadband is recovering, but enterprise spending remains muted as “AI is sucking the oxygen” from other budgets. The company is 65% AI now, and that concentration is both strength and risk.

Why AI/Semiconductor Investors Should Care: Broadcom is the ultimate picks-and-shovels winner of the AI buildout, with diversified exposure across custom XPUs (Google TPU, Meta, ByteDance, Anthropic), networking (Tomahawk, Memory Link, optical), and high-margin VMware software. The $73B backlog provides visibility most semis can only dream of, and management’s confidence in accelerating growth through FY26 suggests hyperscaler capex isn’t peaking—it’s compounding.

The bear case is customer concentration (3-5 hyperscalers drive everything) and margin compression from systems revenue. But here’s the reality: Oracle is struggling to build the buildings; Broadcom is selling the brains that go inside them. One business model is asset-heavy and execution-risky. The other is design-focused and capital-efficient. Watch backlog trajectory and Tomahawk 6 shipment timing—this is the cleanest AI infrastructure play in the market right now.

Join WhatTheChipHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Subscribe now

Nvidia Corporation (NASDAQ: NVDA)

🇨🇳 Trump Greenlights Nvidia H200 China Sales—But Beijing May Say No Thanks

What The Chip: President Trump handed Nvidia a gift on December 8: permission to sell H200 AI chips to “approved customers” in China, with 25% of sales paid to the U.S. government. The H200 is roughly six times more capable than the H20 chips previously available to Chinese buyers, and CEO Jensen Huang has pegged the China AI chip market at $50 billion this year, with Bloomberg estimating H200 annual revenue at $10 billion. But there’s a problem: White House AI czar David Sacks says China has “figured out the U.S. strategy” and is rejecting the chip in favor of domestic alternatives.

Details:

🔄 Deal Structure: H200 chips manufactured by TSMC in Taiwan would be imported into the U.S. for security reviews, taxed at 25%, then re-exported to approved Chinese buyers. This convoluted routing adds cost, complexity, and delays—not exactly competitive with domestic options.

⚡ Performance Gap: The H200 features HBM3e memory with 4.8 TB/s bandwidth (up from H100’s 3.35 TB/s) and 141GB capacity versus 80GB. It’s Nvidia’s third-most powerful processor(Blackwell Ultra #1, Blackwell #2) and roughly six times more capable than the H20—a genuine technological leap.

🎯 Strategic Rationale: U.S. officials told The Information this was a middle ground: Blackwell remains banned, but a complete H200 ban would accelerate China’s domestic chip development. The logic: better to let them buy somewhat-restricted hardware than force self-sufficiency.

🚩 China’s Cold Shoulder: Beijing convened emergency meetings with its largest tech companies to assess demand. Proposals include capping Nvidia purchases relative to domestic accelerator buys, or barring H200 use in strategic sectors like finance and energy. Translation: we’ll grudgingly allow minimal imports, not embrace them.

💰 $70B Self-Sufficiency Push: China is weighing incentives worth as much as $70 billion to support local chipmaking. That’s more than Nvidia’s entire quarterly revenue—Beijing is dead serious about chip independence.

🔌 CUDA Advantage: Unlike most Chinese accelerators, the H200 supports Nvidia’s CUDA software ecosystem, simplifying model porting and cluster integration. This is Nvidia’s moat—even if Chinese chips match performance, CUDA lock-in is real.

⚠️ Smuggling Precedent: A recent case revealed individuals exported or attempted to export at least $160 million worth of H100 and H200 GPUs between October 2024 and May 2025. The gray market exists because demand exists—but it also proves enforcement is porous.

📉 Muted Market Reaction: Nvidia shares climbed on initial news, then pared gains and rose only ~2% after hours. The market isn’t pricing in a China windfall—it’s pricing in uncertainty.

🔮 Analyst Skepticism: Swissquote Bank notes the approval may have a limited impact unless Nvidia can export other chip lines like Blackwell or Rubin. One product line approval doesn’t reopen the Chinese market—it just creates a narrow, heavily monitored channel.

Why AI/Semiconductor Investors Should Care: Trump’s H200 approval is optionality, not certainty. Beijing’s rejection signals a strategic shift: China would rather accept near-term AI performance disadvantages than remain dependent on U.S. silicon. For Nvidia, this means the Chinese market, historically a major revenue driver, is functionally closed for the foreseeable future.

The bull case hinges on Chinese tech giants ultimately being unable to resist H200 performance for training workloads, especially given CUDA ecosystem advantages. The bear case is that Beijing doubles down on Huawei and Cambricon, permanently shutting out Nvidia while subsidizing domestic alternatives. Watch whether any major Chinese AI labs (ByteDance, Alibaba, Tencent) publicly commit to H200 orders—if they stay silent, the market is closed. Either way, Nvidia’s growth story is now 100% dependent on U.S., European, and allied demand.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChipHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

What The Chip Happened?
🔥 NVIDIA’s CFO Spills the Tea: $500 Billion Is Just the Warm-Up
⚡ AWS Flexes Its Chip Muscle: Trainium Goes Multi-Billion, AI Factories Come Home
🔮 Marvell Goes All-In on Light: $5.5B Bet on Photonics Signals AI’s Next Infrastructure Wave

Read time: 7 minutes
Join WhatTheChippHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Nvidia (NASDAQ: NVDA)
🔥 NVIDIA’s CFO Spills the Tea: $500 Billion Is Just the Warm-Up

What The Chip: At the UBS Global Technology Conference on December 2nd, 2025, NVIDIA CFO Colette Kress sat down for a fireside chat that turned into a masterclass in managing investor expectations, firmly rejecting the “AI bubble” narrative while revealing that the company’s massive $500 billion revenue visibility through 2026 doesn’t even include a potential blockbuster direct deal with OpenAI.

Details:

💰 The $500 Billion Floor, Not Ceiling: Kress confirmed that NVIDIA’s cumulative $500 billion Blackwell and Vera Rubin visibility through 2026 represents only what’s coming through cloud service providers. The analyst suggested this translates to roughly $350-400 billion in calendar 2026 revenue, and Kress didn’t push back. More importantly, she revealed: “Right now, that $500 billion does not include any of the work that we’re doing right now on the next part of the agreement with OpenAI.” That’s a potentially massive deal sitting on the sidelines.

📦 Inventory Surge Signals Locked-In Demand: One standout data point: NVIDIA’s combined inventory and purchase commitments jumped $25 billion in a single quarter, up from the usual ~$2 billion quarterly increase. Kress indicated most of that inventory has already shipped to customers as of early December, hard evidence of near-term revenue visibility rather than speculative stockpiling.

⚠️ The OpenAI Elephant in the Room: The analyst pressed hard on the 10-gigawatt OpenAI letter of intent, which UBS estimates could be worth $400 billion over the life of the deal. Kress acknowledged: “Yes, we still have not completed a definitive agreement, but we’re working with them.” OpenAI wants to go direct rather than through hyperscalers, a major strategic shift, but until ink hits paper, this remains upside risk, not certainty.

🛡️ CUDA Moat Gets Quantified: Kress revealed that software alone contributes 2X of Blackwell’s 10-15X performance improvement, a rare quantification of NVIDIA’s software advantage. CUDA is now on its 13th version with backward and forward compatibility. She was emphatic: “Today, everybody is on our platform. All models are on our platform, both in the cloud as well as on-premise.”

📉 Gross Margins Holding Steady: Despite concerns about HBM cost pressures and rising bill of materials, Kress committed to maintaining mid-70s gross margins into next year. She cited improvements in cycle times, yields, and manufacturing costs, noting that even “one more day of efficiency” at NVIDIA’s scale moves the needle significantly.

🔮 Vera Rubin Has Taped Out: The next-generation chip is no longer vaporware. Kress confirmed: “Vera Rubin, we’re pleased to say that it has been taped out. We have the chips and are working feverishly right now to get ready for the second half of next year.” She expects another “X factor increase in performance.”

🚩 Model Builder Financing Concerns Linger: When pressed about model builders with limited revenue committing massive capacity, Kress acknowledged reality: “They’re gonna have to work through, have I earned enough in terms of profitability? Can I raise more capital?” She emphasized NVIDIA vets customers for purchase orders and ability to pay, but this doesn’t eliminate the risk that startup funding could dry up.

🌍 Middle East as Growth Vector: Kress hinted at sovereign AI infrastructure deals as upside to guidance, mentioning Middle East opportunities multiple times and teasing: “You may even hear about another one today as well.”

Why AI/Semiconductor Investors Should Care: This conversation revealed NVIDIA operating from a position of extraordinary strength, $500 billion in visible demand, supply chain commitments suggesting management has locked in hypergrowth through mid-2026, and competitive positioning that Kress described in near-monopolistic terms.

The unsigned OpenAI LOI represents both the biggest upside catalyst and the most significant execution risk; if that deal closes as a direct relationship, NVIDIA’s trajectory steepens further, but its absence from current guidance means the stock’s valuation already relies heavily on confirmed hyperscaler demand. For investors, the $25 billion inventory surge is the green flag that matters most, it’s not guidance or vision, it’s committed capital translating to imminent revenue.

Join WhatTheChippHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Amazon (NASDAQ: AMZN)
⚡ AWS Flexes Its Chip Muscle: Trainium Goes Multi-Billion, AI Factories Come Home

What The Chip: At AWS re:Invent 2025 on December 2nd, CEO Matt Garman unveiled a sweeping set of announcements that position Amazon as a vertically integrated AI powerhouse, from custom silicon to quantum computing to enterprise-grade AI agents, while revealing that its homegrown Trainium chips have quietly become a multi-billion dollar business powering the majority of Amazon Bedrock inference.

Details:

💰 AWS hit $132 billion in revenue, accelerating 20% year-over-year. Garman put the scale in perspective: “The amount we grew in the last year alone was about $22 billion. That absolute growth over the last 12 months is larger than the annual revenue of more than half of the Fortune 500.” The cloud giant added 3.8 gigawatts of data center capacity in 2024, more than any competitor globally.

🔥 Trainium is no longer a side project; it’s a juggernaut. AWS deployed over 1 million Trainium chips and ramped Trainium 2 volumes 4x faster than any previous chip in AWS history. Garman didn’t mince words: “We’re selling those as fast as we can make them. Trainium already represents a multi-billion dollar business today.” Even more striking: the majority of Amazon Bedrock inference now runs on Trainium. “If you’re using any of Claude’s latest generation models in Bedrock, all of that traffic is running on Trainium, which is delivering the best end-to-end response times compared to any other major provider.”

⚡ Trainium 3 Ultra servers hit general availability with AWS’s first 3-nanometer AI chip. The specs are staggering: 4.4x more compute, 3.9x memory bandwidth, and 5x more AI tokens per megawatt versus Trainium 2. The largest configurations pack 144 Trainium 3 chips delivering 362 FP8 petaflops of compute with 700+ terabytes per second of aggregate bandwidth. AWS can scale these clusters to hundreds of thousands of chips.

🚀 AWS previewed Trainium 4, already in active development. Projections show 6x FP4 compute performance, 4x memory bandwidth, and 2x high-bandwidth memory capacity versus Trainium 3, designed to handle the largest models in existence.

🏭 AWS AI Factories bring AWS infrastructure into customer data centers. Think of it as a private AWS region: customers leverage their own real estate and power while getting access to Trainium 3 UltraServers, NVIDIA GPUs, SageMaker, and Bedrock. This directly addresses enterprises with existing data center investments who want AWS’s AI stack without migrating workloads.

🟢 NVIDIA remains a key partner. AWS announced P6e GB300 instances powered by NVIDIA’s GB300 NVL72 systems, delivering over 20x the compute compared to previous-generation P5en instances. AWS continues positioning itself as a multi-vendor shop rather than going all-in on Trainium alone. Tranium 4 will be using Nvidia’s NVLINK Fusion technology.

🤖 Nova 2 models arrived with competitive positioning against frontier models. Nova 2 Pro targets agentic workloads, matching or beating GPT-5.1, Gemini 3 Pro, and Claude 4.5 Sonic on instruction following and tool use benchmarks according to Artificial Analysis. Nova 2 Omni breaks new ground as the industry’s first reasoning model supporting text, image, video, and audio input with text and image output.

⚛️ AWS unveiled Ocelot, its first quantum computing chip prototype, claiming it reduces quantum error correction costs by over 90%, a potential breakthrough for making quantum computing commercially viable.

🔵 Intel and AMD get new instances too. C8a instances (AMD EPYC) deliver 30% higher performance, while C8ine instances (Intel Xeon 6) offer 2.5x higher packet performance per vCPU. For the third consecutive year, over half of new CPU capacity came from AWS’s own Graviton processors, signaling continued in-house silicon momentum.

Why AI/Semiconductor Investors Should Care: The Trainium deployment numbers, 1 million chips, 4x faster ramp than any prior chip, majority of Bedrock inference, suggest AWS has cracked both the manufacturing scale and software optimization challenges that plague custom silicon efforts. For NVIDIA investors, the P6e GB300 announcement confirms AWS remains a massive customer, but the explicit Trainium-vs-GPU competitive positioning and the “tokens per megawatt” efficiency metrics signal AWS is building optionality to reduce GPU dependence over time but crucial to note the TRN4 will use Nvidia’s technology.

Garman’s framing of AI agents as the inflection point where “technical wonder” becomes “real value” echoes a broader industry concern: enterprises have spent heavily on AI infrastructure but haven’t seen proportional returns. If agentic AI delivers on that promise, the 3.8 GW of capacity AWS added becomes revenue-generating infrastructure. If it doesn’t, hyperscaler spending becomes a harder sell to shareholders.

Subscribe now

Marvell (NASDAQ: MRVL)
🔮 Marvell Goes All-In on Light: $5.5B Bet on Photonics Signals AI’s Next Infrastructure Wave

What The Chip: Marvell announced Q3 FY2026 earnings on December 2, 2025, beating expectations with $2.075 billion in revenue (up 37% YoY) while simultaneously revealing a transformational $3.25 billion acquisition of Celestial AI—a photonics startup whose optical interconnect technology has already landed a major hyperscaler design win that Marvell believes will reshape how AI data centers are built.

Details:

💡 The Celestial AI deal is structured as $3.25 billion upfront plus up to $2.25 billion in contingent payments tied to revenue milestones through FY29. For context, this mirrors Marvell’s 2021 Inphi acquisition, which CEO Matt Murphy called “an absolute home run.” Celestial’s photonic fabric technology delivers greater than 2x power efficiency versus copper interconnects with nanosecond-class latency, critical as AI clusters scale and electrical connections hit physical limits. Revenue contributions begin in the second half of FY28, targeting a $500 million annualized run rate by Q4 FY28 and $1 billion by Q4 FY29.

📈 Management significantly raised their outlook. FY27 data center revenue growth expectations jumped to over 25% YoY (previously lower), driven by cloud CapEx growth now expected to exceed 30%. FY28 looks even stronger: data center revenue projected to accelerate to approximately 40% growth, with total company growth around 30%. Murphy explicitly stated these are “base case assumptions, not dream the dream”

🔌 The interconnect business (50% of data center revenue) is firing on all cylinders. 1.6T transceivers began shipping in H2 FY26 with “exceptionally strong demand.” AEC and retimer revenue is expected to more than double from FY26 to FY27. The company demonstrated 400 gig per lane technology for 3.2T transceivers at OFC, with production deployments expected in calendar 2028. Marvell claims it’s “first to market, first to ramp” consistently in optical categories.

🎯 The XPU attach market is emerging as a sleeper growth driver. Marvell has secured over 15 design wins with “line of sight to revenue exceeding $2 billion by fiscal 2029.” Five unique CXL sockets are secured across two Tier 1 U.S. hyperscalers, with the first already entering volume production. Murphy noted attach rates are “exceeding our initial expectations” as hyperscalers plan to deploy custom NICs across AI server fleets that “can exceed 1 million units or more annually” at large customers.

⚠️ The custom business (25% of data center) carries concentration risk. Management acknowledged it’s tied to “a handful of programs today” with a product transition occurring at the lead customer. The second major XPU customer won’t contribute meaningfully until FY28, and Murphy noted this customer “doesn’t have a history of ramping big ASIC programs.” Growth guidance of “at least 20%” reflects this caution, with Q3 showing a sequential decline “due to lumpiness in demand.”

📊 Q3 financials were solid: non-GAAP EPS of $0.76 (beat by $0.02), non-GAAP gross margin of 59.7%, and operating cash flow of $582 million. The company executed $1 billion in accelerated stock repurchases plus $300 million in ongoing buybacks. Q4 guidance calls for $2.2 billion in revenue (up 21% YoY as reported, 24% excluding the divested automotive business).

Why AI/Semiconductor Investors Should Care: Marvell is positioning itself as the “one-stop shop” for everything inside the AI rack, interconnects, switches, custom XPUs, XPU attach solutions, and storage, rather than competing on individual component specs. This rack-level strategy creates meaningful switching costs and deeper customer relationships, which explains why hyperscalers are providing multi-year visibility that enables Marvell to guide two full years out with unusual confidence.

The Celestial AI acquisition bets on copper-to-optics transition as AI clusters scale beyond electrical interconnects’ physical limits, targeting an estimated $10+ billion addressable market by 2030. The risk? Celestial represents an unproven commercial deployment at scale, and the aggressive earnout signals both opportunity and execution pressure. As Murphy put it: “We are everywhere in the AI rack. And we are just getting started.”

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Join WhatTheChippHappened Community — 33% OFF Annual Use Code CYBER
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

What The Chip Happened?
🔥 AMD: Helios, OpenAI & a $1T TAM — the growth math gets real
🧠 “Sold‑Out Cloud” energy: Meta + Microsoft deals power a 2026 sprint
🌐 From Backlog to Backstop: Q3 demand soars, one data‑center snag nudges 2025 into ’26

Read time: 7 minutes
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

AMD (NASDAQ: AMD)
🔥 AMD: Helios, OpenAI & a $1T TAM — the growth math gets real

What The Chip: On November 11, 2025, AMD laid out a bold Analyst Day plan: >35% company‑wide revenue CAGR for the next 3–5 years, non‑GAAP EPS > $20, and a sprint to AI data‑center scale anchored by MI450 “Helios” in Q3’26 and MI500 in 2027. Management framed the prize as a $1T compute market by 2030 with a line of sight to $100B+ annual data‑center revenue within five years.

Details:

📊 “Money slides” that matter. AMD targets >35% revenue CAGR, non‑GAAP operating margin >35%, and non‑GAAP EPS >$20 over 3–5 years. Segmentally, data center grows >60% CAGR and data‑center AI >80%. AMD also shared a 55–58% gross‑margin band and >25% FCF margin for the model. Quote: “AMD is entering a new era of growth…” — CEO Lisa Su. We have never been better positioned.

🖥️ $1T TAM, $100B run‑rate ambition. Su put data‑center compute on a path to $1T by 2030 and said AMD can scale data‑center revenue to ~$100B annually within five years—underscoring how AI training/inference is driving a secular capex wave.

🤝 Mega‑deals de‑risk the ramp.

• OpenAI: multi‑year, multi‑generation pact to deploy 6 gigawatts of AMD GPU capacity, starting with 1 GW in 2H’26 on MI450.

• Oracle: 50,000 MI450 GPUs begin rolling into OCI in Q3’26, expanding in 2027 and beyond. Quote: “Together, AMD and Oracle are accelerating AI with open, optimized, and secure systems built for massive AI data centers.” — Forrest Norrod (EVP, AMD).

• U.S. Dept. of Energy (ORNL): two AMD‑accelerated AI supercomputers (Lux in early 2026; Discovery later) extend sovereign/scientific AI momentum.

🧱 Helios + MI450 (Q3’26) → MI500 (’27): rack‑scale, memory‑heavy AI. Helios is AMD’s open, rack‑scale platform: 72 GPUs per rack, 432GB HBM4 per GPU, and up to 20TB/s memory bandwidth per accelerator—translating to bigger models in‑memory and less sharding. AMD says Helios delivers rack‑level performance leadership, and the roadmap steps again with MI500 in 2027.

🧠 CPU cadence & process lead‑ins. The next‑gen EPYC “Venice” CPU—TSMC N2 (2nm)—is on the Helios bill of materials, lining up with MI400‑class deployments in 2026. (Why it matters: more cores/threads per socket and better perf/W shrink total cluster count per job.)

💻 AI PC scale‑out. Ryzen’s AI PC portfolio grew 2.5× since 2024 and now sits in 250+ platforms, with adoption at over half of the Fortune 100. AMD guides a ~10× improvement in on‑device AI performance by 2027 with “Gorgon” and “Medusa.” (Translation: local copilots get far faster, cheaper to run.).

🧰 Open software flywheel. ROCm downloads surged ~10× YoY, and AMD keeps adding features/perf across releases. The open stack (ROCm + UALink + Ultra Ethernet‑aligned fabric) is a practical way to lower switching costs from CUDA‑first estates.

⚠️ What could go wrong (and what to watch):

• Supply/complexity risk: Helios depends on HBM4, liquid cooling, and 800G‑class fabrics at scale; execution, yields, and serviceability will matter.

• Policy & China exposure: 2025 export rules hurt shipments/margins; AMD’s Q4’25 outlook embeds ~54.5% non‑GAAP GM as a near‑term reality versus the long‑term 55–58% target.

• Competition: Nvidia’s next racks (Vera Rubin/OBERON‑class) set a high bar; AMD’s open approach must translate to time‑to‑train, memory fit, and TCO wins in customer POCs.

Why AI/Semiconductor Investors Should Care

This isn’t just “another node + another GPU.” The 6‑GW OpenAI commitment, 50k‑GPU Oracle cluster, and sovereign AI wins give AMD multi‑year shipment visibility and prove customers will buy an open, memory‑rich, rack‑scale alternative when it hits real‑world perf/TCO. If AMD executes Helios + MI450 on time and ROCm keeps compilers/kernels in sync, the model supports margin expansion toward 55–58% GM and a credible path to double‑digit AI accelerator share—with $100B data‑center revenue the north star. Just remember the timing: the heavy lift starts 2H’26, so near‑term numbers still ride MI350 and EPYC share gains while policy/supply remain watch items.

NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Nebius Group (NASDAQ: NBIS)
🧠 “Sold‑Out Cloud” energy: Meta + Microsoft deals power a 2026 sprint

What The Chip: On Nov 11, 2025, Nebius said it inked a $3B / 5‑year AI‑infrastructure deal with Meta, on the heels of its $17.4–$19.4B Microsoft contract. Management stressed demand was “overwhelming” and the Meta deal size was capped by available capacity, not appetite.

🤝 Mega‑deals as accelerants, not the end game. Meta’s $3B and Microsoft’s $17.4–$19.4B agreements validate Nebius with tier‑1 buyers and act as financing flywheels for the core AI cloud. As the shareholder letter puts it, these partnerships are “commercial and financial accelerators” for the cloud business. Quote (CEO Arkady Volozh): demand for Meta capacity was “overwhelming… limited to the amount of capacity that we had available.”

⚡ Capacity is the constraint—and the unlock. Nebius lifted its contracted power target from 1 GW to >2.5 GW by YE26, with 800 MW–1 GW connected by then (220 MW connected by YE25). That’s the gating item for revenue in 2025–26.

📈 2026 revenue runway got longer. Management now targets $7–$9B ARR by YE26 (ARR = annualized run‑rate revenue), up from $551M ARR exiting September. “Sold out” remained the mode in Q3.

💸 CapEx throttle‑up + staged risk control. 2025 CapEx guidance jumped to ~$5B (from ~$2B) to lock down hardware, power, land and sites. CFO also reiterated a three‑stage build—land & power → shell & facilities → GPUs—to avoid overspending and preserve flexibility if the market shifts. Cash stood at $4.8B (9/30/25) after raising $4.3B via converts + equity in September.

📊 Quarter in numbers (Q3’25): Revenue $146.1M (+355% Y/Y, +39% Q/Q); adjusted EBITDA loss $(5.2)M (an 89% Y/Y improvement). Core AI infrastructure (~90% of revenue) grew ~400% Y/Y / 40% Q/Q with ~19% adjusted EBITDA margin. Cost structure tightened: cost of rev 29% of sales, R&D 31% (vs 98%), SG&A 61% (vs 149%).

🧾 Guidance keeps tightening. For 2025, Nebius reiterated $500–$550M revenue and $900M–$1.1B year‑end ARR; management still expects adjusted EBITDA to turn slightly positive at group level by year‑end (full‑year still negative).

🧱 Execution risk = timing risk. Management said Q3 landed at the midpoint because revenue is tied to the exact timing of capacity going live. CFO Dado Alonso flagged incremental ARR of ~$12M in Q3 (vs $180M in Q2) due to supply constraints—underscoring how tightly growth couples to deployment.

🧩 Enterprise software push (Aether 3.0 + Token Factory). Nebius rolled out Aether with SOC 2 (incl. HIPAA) and ISO 27001 certifications plus enterprise‑grade IAM/observability, and launched Token Factory to industrialize inference/post‑training. Management’s message: this is the foundation for a stronger 2026 enterprise mix.

🖥️ Technical edge: early Blackwell at scale. Nebius is among the first to deploy NVIDIA B200/B300 broadly (UK B300s; US/Israel B200s). It earned NVIDIA Exemplar Cloud status on H200 (training) and posted top MLPerf v5.1 inference marks: GB200 NVL72 up ~6.7% (offline) and ~14.2% (server) on Llama‑3.1‑405B vs prior bests; B200 delivered ~3× / ~4.3× tokens‑per‑second vs H200 in Nebius’ submissions. Customers include Cursor and Black Forest Labs.

Why AI/Semiconductor Investors Should Care

Nebius’ print reinforces a simple reality: AI compute is still supply‑constrained, and those who secure power, land, and Blackwell‑class GPUs early enjoy pricing power and multi‑year visibility. If Nebius executes its >2.5 GW plan and monetizes Microsoft/Meta capacity on schedule, the path to multi‑billion ARR becomes tangible—good read‑through for NVIDIA demand and for the broader AI data‑center supply chain. The flip side: this is capital‑intensive and timing‑sensitive; as more neoclouds/hyperscalers bring capacity online into 2026–27, pricing could compress—so watch build timing, financing mix, and unit margins closely.

Subscribe now

CoreWeave (NASDAQ: CRWV)
🌐 From Backlog to Backstop: Q3 demand soars, one data‑center snag nudges 2025 into ’26

What The Chip: On Nov 10, 2025, CoreWeave posted record Q3 results and a $55.6B revenue backlog, then trimmed 2025 guidance after a third‑party data‑center developer slipped, pushing revenue into early 2026 rather than losing it. Management also flagged 2026 CapEx set to be well above double 2025 to fuel surging AI compute demand.

Details:

📈 Backlog goes parabolic. Contracted revenue backlog hit $55.6B (+271% YoY), propelled by multi‑year deals: Meta up to $14.2B, OpenAI +$6.5B (total $22.4B), plus additional hyperscaler wins. Translation: multi‑year demand is locked in.

🧱 Timing hit, not demand hit. Because one developer ran behind, CoreWeave cut FY25 revenue to $5.05–$5.15B and CapEx to $12–$14B; the affected customer extended timelines to preserve full contract value, deferring rev mostly to Q1’26.

🚀 CapEx blast‑off next year. Management said 2026 CapEx will be “well in excess of double” 2025—think $25–$30B+—to stand up capacity for that backlog. Read‑through: power, racks, cooling, and GPUs remain the bottleneck—and the opportunity.

🧩 Customer risk dramatically lower. Single‑customer concentration fell from ~85% at the start of 2025 to ~35% in Q3; 60%+ of backlog is now with investment‑grade customers. That’s real durability for revenue quality.

💰 Profit engine revs (with high interest drag). Q3 revenue $1.36B (+134% YoY); Adj. EBITDA $838M (61%); Adj. Op. Income $217M (16%); GAAP net loss $110M. FY25 interest expense guided to $1.21–$1.25B, a key earnings headwind to watch.

🔌 Infrastructure scale‑up continues. Active power ~590 MW (+120 MW q/q), contracted power ~2.9 GW, 41 data centers. Q3 CapEx $1.9B with construction‑in‑progress $6.9B (+$2.8B q/q)—that’s future capacity not yet in service.

🤝 NVIDIA “backstop” becomes a growth wedge. A $6.3B strategic collaboration lets CoreWeave interrupt and resell NVIDIA‑reserved capacity to smaller AI labs while NVIDIA underwrites unused capacity—mitigating utilization risk and opening a long‑tail of customers.

🎯 Year‑end operating targets. Company now expects >850 MW of active power by year‑end; FY25 adjusted operating income guided to $690–$720M.

Why AI/Semiconductor Investors Should Care

CoreWeave’s print screams structural demand: a $55.6B backlog, Blue‑chip customers, and a 2026 CapEx surge point to sustained GPU, networking, and power gear orders—bullish for suppliers up and down the AI stack. The bear case is execution risk: industry‑wide constraints in powered shells, transformers, and contractors inject volatility, while $1.2B in annual interest keeps the EPS bridge tight. The NVIDIA backstop/interruptible structure lowers utilization risk and broadens TAM into startups—an underappreciated edge—yet the market will keep scoring CoreWeave quarter‑to‑quarter on how fast new MWs energize and how quickly deferred revenue shows up in Q1’26.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

What The Chip Happened?
🤖 AMD Levels Up—OpenAI Mega-Deal Fuels an AI Renaissance
⚡ Navitas Powers Up with Strategic Pivot to High-Voltage AI Future
🔥 Super Micro’s Mega-Scale Moment: Q2 Revenue Set to Double After Historic $13B Order Wave

Read time: 7 minutes
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Advanced Micro Devices (NASDAQ: AMD)
🤖 AMD Levels Up—OpenAI Mega-Deal Fuels an AI Renaissance

What The Chip: AMD dropped its Q3 earnings on November 4, 2025, reporting record revenue of $9.2 billion (+36% YoY, +20% QoQ). A colossal multi-year partnership with OpenAI to deploy 6 gigawatts of AMD’s Instinct GPUs, kicking off with 1 GW of MI450 Series accelerators in 2H26. CEO Lisa Su confidently projects AI-driven revenues to soar into “tens of billions” by 2027, transforming AMD into an undisputed heavyweight in AI computing.

Details:

🚀 AI’s New Heavyweight: AMD’s multi-year partnership with OpenAI is no ordinary deal, it’s massive, deploying 6 gigawatts of Instinct GPUs starting mid-2026. Management boldly called out potential revenues exceeding $100 billion in coming years from similar deals, positioning AMD as a dominant player in the AI compute market.

🖥️ Unexpected CPU Renaissance: AI isn’t just GPUs anymore. Hyperscalers, including major cloud providers, are now significantly ramping up CPU builds to handle the surging AI workload demand. This newfound CPU boom, described as a “multi-quarter phenomenon,” aligns perfectly with AMD’s rollout of its high-performance 5th Gen EPYC Turin processors, already representing nearly half of EPYC revenues.

📈 Record-Breaking Financial Performance: AMD delivered robust Q3 results, posting non-GAAP diluted EPS of $1.20 (+30% YoY). Total revenue reached $9.2 billion (+36% YoY), with the Data Center segment leading the charge at a record $4.34 billion (+34% QoQ). Impressively, AMD generated $1.53 billion in free cash flow, more than triple last year’s figure.

🔧 AI Scale and Supply Challenges: Scaling 6 gigawatts isn’t trivial. AMD acknowledges “tight” supply chain constraints across silicon, memory, power, and advanced packaging, which introduces execution risks over the next two years. While management remains optimistic, investors should keep an eye on these critical bottlenecks.

⚔️ David vs. Goliath in AI Accelerators: AMD faces intense competition from NVIDIA, especially as it battles for massive hyperscale deployments. The current guidance notably excludes revenue from MI308 GPU shipments to China due to export controls—a significant headwind investors should monitor closely.

🎮 Client and Gaming Rebound: AMD’s client and gaming segments crushed it, hitting a record $4.05 billion (+73% YoY). Gaming alone surged 181% YoY, underlining strong demand and market leadership in consumer markets.

🖥️ Full-Stack AI Platform Advantage: AMD’s secret weapon? Its unique full-stack approach. AMD isn’t just offering GPUs—it’s delivering fully integrated rack-scale solutions (Helios) combining GPUs (MI450), CPUs (2nm Venice), and Pensando networking, optimized for performance and efficiency. This strategy, bolstered by recent acquisitions like ZT Systems, could cement long-term customer lock-in.

💡 Software Breakthrough with ROCm: AMD’s previously lagging software stack, ROCm, is finally gaining developer traction. Third-party heavyweights like Hugging Face and vLLM are directly contributing to ROCm 7. This signals a crucial tipping point for AMD, potentially narrowing the software ecosystem gap with NVIDIA’s CUDA.

Why AI/Semiconductor Investors Should Care:
This quarter’s results signal that AMD isn’t just riding the AI wave, it’s carving out a substantial chunk of the market. The OpenAI deal is transformative, not only due to sheer size but because AMD is uniquely positioned to deliver integrated solutions that could deepen customer lock-in and raise competitive barriers. While execution risks from the ambitious scale-up remain real, AMD’s aggressive R&D investments, software improvements, and CPU resurgence driven by AI workloads collectively suggest a strong, durable growth trajectory.

NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Navitas Semiconductor (NASDAQ: NVTS)
⚡ Navitas Powers Up with Strategic Pivot to High-Voltage AI Future

What The Chip: Navitas Semiconductor just announced a decisive pivot away from consumer and mobile markets towards high-power applications, as new CEO Chris Allexandre called this “a crucial moment” for the company. The move, unveiled in their Q3 earnings on November 4, 2025, involves restructuring operations to fully embrace AI data centers, grid infrastructure, performance computing, and industrial electrification.

Details:

📉 Painful Pivot in the Short-term: Navitas expects Q4 revenue to hit a bottom of about $7.0M—down 31% sequentially and 67% year-over-year. Management openly acknowledged intentionally walking away from revenue, particularly in China mobile, due to lower margins and tariff pressures on silicon carbide (SiC).

🤖 Validated by NVIDIA: Navitas received a critical endorsement from NVIDIA at the recent OCP Global Summit, being named a preferred power semiconductor partner for NVIDIA’s cutting-edge 800V DC AI data center architecture. CEO Chris Allexandre highlighted that Navitas is uniquely positioned to handle the power path “from the grid to the GPU,” marking a strong validation of its technology.

🔋 Material Revenue Pushes to 2027: While Navitas already ships into AI data centers, substantial revenue won’t appear on the financial statements until 2027, when the 800V DC architecture adoption becomes mainstream. Investors should expect 2026 to be a foundational year of design wins rather than significant revenue growth.

🚀 AI-Electrification Tailwind: Allexandre underscored AI as a powerful catalyst driving accelerated demand not just in data centers but also high-performance computing and industrial markets. Navitas believes this trend will accelerate demand well ahead of the major 800V ramp, creating multiple growth opportunities throughout 2026.

🎯 Complete High-Power Portfolio Advantage: Navitas differentiates itself with a comprehensive suite of power products—from GaN ICs at lower voltages (100V for mid-voltage AI server power) to high-voltage SiC modules (2.3kV-3.3kV) for energy storage and grid infrastructure. This breadth is critical, as it allows deeper, multi-subsystem engagements, increasing margins and revenue predictability.

💸 Financial Health Check:

• Q3 revenue dropped to $10.1M, down 53% YoY, as the mobile business shrinks.

• Non-GAAP gross margin was relatively stable at 38.7% despite plunging revenues.

• Net loss of $10.2M (non-GAAP), mitigated by reduced operating expenses and strengthened cash position ($150.6M cash, zero debt) thanks to a recent $100M capital raise.

Why AI/Semiconductor Investors Should Care:
Navitas’ pivot is bold and painful in the short term, but it positions the company squarely in the high-value future of power semiconductors driven by AI. The strategic NVIDIA partnership and the shift towards 800V DC data center architectures are transformative trends investors cannot ignore. While 2026 will test investor patience with gradual, foundational growth, the real acceleration and payoff emerge in 2027, making Navitas a name to closely track for investors targeting the next big wave in AI infrastructure growth.

Subscribe now

Super Micro Computer (NASDAQ: SMCI)
🔥 Super Micro’s Mega-Scale Moment: Q2 Revenue Set to Double After Historic $13B Order Wave

What The Chip: On November 4, 2025, Super Micro Computer announced Q1 FY2026 results featuring a surprising revenue miss at $5 billion—but that’s not the headline. A massive $1.5 billion revenue shift from Q1 to Q2, sparked by customer upgrades to its cutting-edge GB300 AI platform, turns next quarter into a blockbuster. With full-year guidance boosted to at least $36 billion (up from $33 billion), this quarter marks a pivotal moment in Super Micro’s journey into mega-scale territory.

Details:

🚀 Historic $13 billion order backlog: Super Micro secured a record $13 billion in new orders during Q1—its largest-ever deal pipeline. Most of this backlog is tied to NVIDIA’s Blackwell Ultra (GB300) AI platform, setting the stage for rapid growth through FY2026.

🌐 Revenue doubles next quarter: After the Q1 shortfall, Super Micro guided Q2 revenue to a massive $10-11 billion, doubling sequentially. This isn’t just a bounce-back—it’s driven by large-scale AI platform deployments to hyperscalers, showing strong customer demand.

⚡ Mega-scale manufacturing ramp: Production capacity is scaling aggressively, now reaching 6,000 racks per month, half featuring advanced Direct Liquid Cooling (DLC). With theoretical annual capacity surpassing $100 billion, Super Micro is positioned for substantial market-share gains, albeit with near-term margin pressures.

📉 Margins under near-term pressure: Q2 gross margins are expected to drop 300 basis points to approximately 6.5% due to customer mix and upfront investment in mega-scale deals. CFO David Weigand explained, “We’re ramping one of the largest clusters in the world,” indicating deliberate short-term margin sacrifice to secure strategic mega-scale relationships.

💸 Cash flow strains intensify: Operating cash flow swung sharply negative (-$918 million), pushing Super Micro from net cash to a $575 million net debt position. Management acknowledged the strain, securing an $1.8 billion receivables facility and promising tighter control of growth aligned with available working capital.

🔍 DCBBS: High-margin future: CEO Charles Liang highlighted their high-margin Data Center Building Block Solutions (DCBBS) as essential for future profitability. Unlike commodity servers, DCBBS solutions—including advanced cooling and power management—are key to margin recovery, especially as they ramp up capacity.

🌏 Geographic shift—Asia rising: Revenue from Asia soared 143% YoY to 46% of total revenue, driven by a U.S.-based hyperscaler expanding aggressively in Asia. Meanwhile, U.S. revenues fell sharply (down 57% YoY), highlighting shifting customer deployments globally.

🎯 Strategic confidence despite hurdles: Even amid short-term pressures, CEO Liang described the revised $36 billion full-year target as “very conservative,” emphasizing growing confidence in sustained demand far beyond FY2026.

Why AI/Semiconductor Investors Should Care:
Super Micro’s strategic bet—sacrificing near-term margins to win mega-scale AI deals—could significantly reshape the company’s competitive position. While cash flow and margin deterioration pose immediate risks, the massive GB300 backlog and advanced DCBBS solutions provide long-term margin expansion opportunities. Investors must closely monitor the company’s execution: if Super Micro can convert short-term pain into sustained mega-scale customer relationships, its current strategy could unlock substantial long-term gains.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

What The Chip Happened?
🚧 Capacity Is the New Currency — Q1 FY26: Azure Soars, OpenAI Locks in $250B
🎛️ Alphabet’s $100B Moment: All-In on AI, Tight on Supply
🧠 Meta’s Super-Compute Sprint: Big Brain, Bigger Bill
[Microsoft FY26 Q1: Cloud & AI Lift a $77.7B Quarter]

Read time: 7 minutes
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Microsoft (NASDAQ: MSFT)
🚧 Capacity Is the New Currency — Q1 FY26: Azure Soars, OpenAI Locks in $250B

What The Chip: Microsoft posted Q1 FY2026 results on October 29, 2025 with Azure up 40% and a fresh warning: demand is running ahead of supply through fiscal year-end. A day earlier (Oct 28), Microsoft and OpenAI signed a definitive deal that includes $250B of incremental Azure commitments and extends Microsoft’s access to OpenAI model IP, reshaping revenue visibility for years.

Details:

💥 Demand > Supply. CFO Amy Hood said, “In Azure, we expect Q2 revenue growth of approximately 37% in constant currency as demand remains significantly ahead of the capacity we have available… we now expect to be capacity constrained through at least the end of our fiscal year.” Translation: Microsoft is deliberately throttling some external workloads to feed first‑party apps, R&D, and required server replacements.

🏗️ Capex dialed up—again. Q1 capex hit $34.9B (+74% y/y). Roughly half went to short‑lived GPUs/CPUs, and Microsoft recorded $11.1B in finance leases for longer‑lived assets. Management now expects FY26 capex growth to exceed FY25, reflecting an all‑out build to meet AI demand.

🧱 Building a bigger factory. Satya Nadella (CEO) said Microsoft will increase total AI capacity by >80% this year and plans to roughly double its data‑center footprint in two years—evidence the spending isn’t just GPUs, it’s entire regions coming online.

🤝 OpenAI deal = moat + visibility. The new agreement gives Microsoft an approx. 27% stake in OpenAI’s public‑benefit corp and includes $250B incremental Azure consumption. Microsoft retains rights to OpenAI models/products through 2032, with an independent panel now validating any claim of AGI. (Microsoft also dropped a prior “right of first refusal” even as it tightens the tie on Azure usage.)

📚 Bookings & backlog exploded. Commercial bookings +112% y/y, primarily on Azure commitments from OpenAI. Commercial RPO reached $392B (+51%) with a ~2‑year weighted average duration—short enough to matter for near‑term revenue.

📊 Margin math: investment first. Overall gross margin stayed high at ~69%, but Microsoft Cloud margin was 68% this quarter, and Hood guided ~66% next quarter as AI scale‑out mixes in and depreciation ramps.

🧠 AI usage at platform scale. Nadella: “We increased token throughput for GPT‑4.1 and GPT‑5 by >30% per GPU this quarter.” First‑party Copilots now >150M MAUs; Microsoft says AI features reach ~900M monthly users; GitHub Copilot tops 26M users.

🗣️ How Microsoft is thinking about monetization. Nadella framed AI as expanding categories, not just raising ARPU: “You could say our ARPUs are too low for M365—or you could say AI lets us be much more expansive… coding goes from tools to one of the most expansive AI systems… same with security and consumer—it’s ads plus subscriptions.”

Why AI/Semiconductor Investors Should Care

Microsoft’s message is clear: it will sacrifice some near‑term margin to secure a durable AI edge in infrastructure and the application layer (Copilot/agents). The $250B OpenAI commitment plus $392B RPO compresses uncertainty on future Azure revenue, while persistent capacity constraints imply upside when supply actually catches up. Watch for signals that constraints ease (H2 FY26/FY27) and for agentic‑AI attach rates across M365, GitHub, and Security—if usage converts to paid seats at scale, Microsoft’s margin headwinds can flip to operating leverage.

NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Alphabet (NASDAQ: GOOGL)
🎛️ Alphabet’s $100B Moment: All-In on AI, Tight on Supply

What The Chip: Alphabet posted its first-ever $100B+ quarter on October 29, 2025, and promptly raised 2025 CapEx to $91–93B as it races to build AI capacity. Management says 2026 spend goes higher from here.

Details:

💾 CapEx = commitment. 2025 CapEx guide jumps to $91–93B (from $85B) to fund data centers, GPUs/TPUs, and networking. Q3 purchases of property & equipment hit $23.95B, up ~83% YoY (vs. $13.06B in Q3’24). CFO Anat Ashkenazi (Alphabet CFO since July 2024) added 2026 will see a “significant increase.” Depreciation rose to $5.6B (+41% YoY) and will “accelerate slightly” in Q4. Translation: near‑term margin pressure in service of capacity.

☁️ Cloud is the growth engine. Google Cloud revenue: $15.2B (+34% YoY); operating income: $3.6B (+85%); operating margin: 23.7%. The Cloud backlog hit $155B and, importantly, management said it was up 46% QoQ and 82% YoY—evidence of enterprise AI demand converting into signed commitments.

🚦 Capacity remains the bottleneck. Alphabet expects to “remain in a tight demand‑supply environment” in Q4 and into 2026 for AI infrastructure. That’s bullish for long‑term revenue but constrains near‑term monetization if customers can’t get capacity when they want it.

🔎 AI is expanding Search, not just shifting it. CEO Sundar Pichai said overall and commercial queries grew faster in Q3 thanks to AI Overviews and AI Mode. AI Mode now has >75M daily active users, with queries doubling in Q3. CBO Philipp Schindler added “AI Max unlocked billions of net new queries” in the quarter—fresh ad inventory, not just reshuffled demand. (AI Mode = conversational, agentic search; AI Overviews = generative summary atop results.)

📺 Ads solid—but Q4 comps get tricky. YouTube ads: $10.3B (+15% YoY), yet management cautioned Q4 advertising comps will be tough due to heavy 2024 U.S. election spend, especially on YouTube. Expect deceleration optics even if fundamentals hold.

🧠 Model/compute scale is staggering. Pichai: first‑party models like Gemini now process ~7B tokens per minute via direct API usage; across Google surfaces the company is processing >1.3 quadrillion tokens per month (>20× YoY). Gemini app surpassed 650M MAUs; paid subs >300M. That’s durable usage to monetize across Cloud, Ads, and Workspace.

🧱 Regulatory and cost headwinds linger. Q3 included a $3.5B EC fine, which dragged reported margin to 30.5% (would be 33.9% ex‑fine). Management also flagged higher technical infrastructure operating costs (e.g., energy) as AI build‑out scales.

🧩 Full‑stack differentiation. Pichai emphasized Alphabet as “the only Cloud provider offering our own leading generative AI models,” alongside custom TPUs and NVIDIA GPUs—vertical integration that can sustain Cloud margins as scale ramps. Also: 13 Cloud product lines now run at >$1B ARR each.

Why AI/Semiconductor Investors Should Care

Alphabet is turning AI from promise to P&L. Cloud backlog acceleration (+46% QoQ) and ads fueled by AI‑driven query growth point to multi‑year revenue compounding, while 23.7% Cloud margin and full‑stack control (models + TPUs + infra) support structurally better unit economics than a resell‑only approach. The trade‑off: CapEx and depreciation will compress margins near‑term, and capacity tightness into 2026 could defer some Cloud revenue. If Alphabet executes its build‑out on time, investors get a bigger, higher‑quality cash engine on the other side—one where AI features monetize at roughly Search‑like rates today, with upside as new ad formats mature.

Subscribe now

Meta Platforms (NASDAQ: META)
🧠 Meta’s Super-Compute Sprint: Big Brain, Bigger Bill

What The Chip: On October 29, 2025, Meta lifted its 2025 CapEx outlook to $70–$72B (from $66–$72B), flagged even larger CapEx in 2026, and warned that EU ad rule changes could ding revenue as soon as Q4. The thread through it all: Meta says its compute needs expanded meaningfully vs. last quarter’s plan.

Details:

💾 Compute floodgates opening. CFO Susan Li said “capex dollar growth will be notably larger in 2026 than 2025,” with total expenses growing at a significantly faster rate next year—driven by infrastructure (cloud + depreciation) and AI-heavy headcount. Translation: the data-center build is just getting started. 2025 CapEx: $70–$72B.

💸 Profit optics vs. tax reality. Q3 revenue hit $51.24B (+26% y/y) and operating margin landed at 40%. Reported EPS fell to $1.05 after a $15.93B one‑time tax charge tied to the One Big Beautiful Bill Act; ex‑charge EPS would be $7.25. Free cash flow: $10.62B; Q3 CapEx: $19.37B; buybacks: $3.16B; dividends: $1.33B.

📈 Ad engine keeps humming. Ad impressions +14%; average price/ad +10%; FoA ad revenue $50.08B (+26%). The family hit 3.54B daily active people in September. Inside the apps: time spent +5% on Facebook and +10% on Threads; video time on Instagram +30% vs. last year.

🧲 Three AI revenue magnets. Reels now runs at >$50B annualized; Meta’s end‑to‑end AI ad tools crossed $60B run rate; Click‑to‑WhatsApp ads +60% y/y. CEO Mark Zuckerberg added: “More than a billion monthly actives already use Meta AI.” He also teased a “single unified AI system” spanning Facebook, Instagram, and ads.

👓 AI wearables find a pulse. New Ray‑Ban Meta Display glasses “sold out in almost every store within 48 hours,” with demo slots booked through next month. Management plans to increase manufacturing.

🧑‍⚖️ Regulatory clouds. Meta cautioned that European Commission changes to its Less Personalized Ads offering “could have a significant negative impact” on EU revenue as early as this quarter; youth‑related U.S. trials in 2026 “may ultimately result in a material loss.”

🧠 Compute hunger, not a fad. Zuckerberg: Meta is “building what we expect to be an industry‑leading amount of compute” and the core business “continues to be able to profitably use much more compute than we’ve been able to throw at it.” That’s the strategic rationale for front‑loading spend.

🔭 Q4 outlook. Meta guides Q4 revenue to $56–$59B, assumes a ~1% FX tailwind, and expects a 12–15% tax rate. Also noted: Q4 Reality Labs revenue likely down y/y due to headset timing (holiday channel fill pulled into Q3).

Why AI/Semiconductor Investors Should Care

Meta just put a multi‑year purchase order on compute—servers, data centers, and network infrastructure—and says 2026 spend accelerates further. That supports demand not only for accelerators/GPUs but also for HBM memory, optical networking, power gear, and data‑center construction. The near‑term trade‑off is margin pressure from depreciation and cloud costs, plus EU ad uncertainty that could shave top‑line in the next quarter—but if Meta’s unified AI and rec‑systems roadmap keeps lifting ad ROI and engagement, the infrastructure will have a long runway to monetize.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

What The Chip Happened?
🏭 Amkor Goes All-In on Arizona—Bets Big on AI Packaging Boom
🤖 Cadence Hits Record Backlog as AI Wave Boosts Outlook
🟢 Intel’s AI Revival: Capacity Crunch and Strategic Wins
[Intel Q3’25: Margins Rebound, AI Partnerships Heat Up]

Read time: 7 minutes
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Amkor Technology (NASDAQ: AMKR)
🏭 Amkor Goes All-In on Arizona—Bets Big on AI Packaging Boom

What The Chip: Amkor Technology just turned up the heat on its U.S. expansion plans, announcing on its Q3 earnings call (October 27, 2025) an increased Arizona investment of $7 billion, highlighting a dramatic uptick in demand for domestic semiconductor packaging and testing, especially driven by AI and advanced computing.

Details:

🏗️ Mega Expansion in Arizona: Amkor is significantly ramping up its Arizona campus investment to $7 billion, featuring 750,000 square feet of cleanroom space and promising up to 3,000 jobs. Phase 1 is targeted for completion in mid-2027, with production starting early 2028. CEO Giel Rutten emphasized surging U.S. manufacturing interest and customer commitments as key drivers.

📈 Margin Boost Ahead: CFO Megan Faust outlined a strategic path to expand corporate gross margins by ~100 basis points by the end of 2027. This improvement hinges on optimizing manufacturing in Japan, efficiency gains in Vietnam, mainstream recovery, and ramping advanced packaging technologies.

⚠️ Near-Term Margin Pressure: Despite robust revenue growth, near-term gross margins face headwinds due to product mix and higher upfront costs for scaling advanced packaging technologies. Gross margins for Q4 are expected to be constrained by higher manufacturing expenses and unfavorable product mix compared to last year.

💰 Strong Financial Performance: Q3 2025 revenue hit $1.99 billion (up 31% sequentially, 7% year-over-year), surpassing guidance. EPS was $0.51, and gross margin improved to 14.3% (up 230 basis points sequentially), with advanced products revenue climbing sharply to $1.68 billion.

💼 Solid Financial Position: The company remains well-capitalized with $2.1 billion in cash and short-term investments and total liquidity of $3.2 billion. Amkor raised its 2025 CapEx guidance from $850 million to $950 million, largely for the Arizona campus.

🚧 Capacity Crunch is Real: CEO Rutten noted existing tightness in advanced packaging capacity, particularly for flip chip and wafer-level packaging technologies, underscoring ongoing constraints despite aggressive expansions.

🤖 AI Demand Surge: Rutten highlighted broad-based growth driven by accelerated AI adoption across multiple end markets, including edge devices, communications, and data centers, reinforcing the long-term secular growth trajectory for advanced packaging technologies.

🚀 High-Density Fan-Out (HDFO) Ramp: Amkor is aggressively scaling its HDFO capacity, with investments in highly flexible equipment. It’s currently ramping multiple products with major customers, signaling significant near-future revenue growth.

Why AI/Semiconductor Investors Should Care: Amkor’s substantial investment boost in Arizona signals a seismic shift toward domestic advanced packaging—a critical component for AI and semiconductor infrastructure. For NVIDIA and other AI chip giants, this campus significantly de-risks the advanced packaging supply chain. Meanwhile, capacity constraints in advanced packaging highlight a crucial bottleneck that Amkor is actively addressing, positioning it as a vital player in the AI hardware ecosystem. Investors should note both the short-term margin pressures from heavy upfront investment and the substantial long-term payoff as advanced packaging transitions from a niche necessity to a central pillar of the semiconductor industry.

NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Intel (NASDAQ: INTC)
🚀 Intel’s AI Revival: Capacity Crunch and Strategic Wins

What The Chip: Intel’s Q3 2025 earnings, announced recently, highlighted persistent supply constraints despite strong demand, a robust AI-driven growth strategy, and significant cash inflows bolstering its manufacturing capabilities.

Details:

📈 Demand Exceeding Supply: CFO David Zinsner confirmed ongoing manufacturing constraints, particularly on Intel 10 and Intel 7 nodes, limiting full demand fulfillment in Q3. These limitations are expected to continue into 2026, suggesting potential revenue upside as Intel ramps up capacity.

💡 AI’s Dual Boost: CEO Lip-Bu Tan underscored how AI is not just accelerating demand for new architectures but is also rejuvenating traditional compute markets. This dual opportunity strengthens Intel’s market position across both cutting-edge and legacy workloads.

💰 Cash Boost from Key Partnerships: Intel significantly improved its balance sheet with nearly $20 billion in new cash, including a notable $5 billion investment from NVIDIA, $5.7 billion from the U.S. government, $2 billion from SoftBank, and $5.2 billion from Altera and Mobileye asset sales, dramatically enhancing operational flexibility.

💻 Solid Segment Performance: Client Computing revenue reached $8.5 billion, boosted by new product launches like Lunar Lake and Arrow Lake. Data Center and AI revenue stood strong at $4.1 billion, reflecting increased AI demand. However, Intel Foundry continues to struggle, recording a $2.3 billion operating loss, though improved sequentially by $847 million.

⚠️ Margin Pressure Ahead: Intel faces short-term headwinds due to product launches and unfavorable cost structures. CFO Zinsner noted Lunar Lake’s initial margin dilution and ongoing competitive pressures, especially acute in the data center segment.

🔧 Foundry Execution Risk: Intel’s foundry business remains unprofitable, posting a -54.8% operating margin. Successfully competing against established foundries like TSMC and Samsung hinges on flawless execution of Intel’s upcoming 18A and 14A nodes.

🤖 AI Infrastructure Demand: Intel sees significant growth in CPU demand driven by the expanding AI infrastructure market. Executives emphasized CPUs’ critical role, forecasting a greater than 10x increase in AI capacity by 2030, indicating ample future opportunities for Intel.

🤝 NVIDIA Partnership: Intel’s strategic collaboration with NVIDIA, featuring a $5 billion investment and manufacturing NVIDIA’s custom AI chips, validates Intel’s advanced foundry capabilities. This partnership could transform Intel into a key supplier in the advanced AI chip market, diversifying NVIDIA’s manufacturing beyond TSMC.

Why AI/Semiconductor Investors Should Care:

Intel’s strategic pivot to harness AI-driven growth presents substantial opportunities but also underscores ongoing execution risks. The NVIDIA partnership, especially leveraging Intel 18A, could redefine Intel’s competitive stance in advanced chip manufacturing, offering potential revenue diversification and long-term growth in AI infrastructure. Investors should carefully monitor Intel’s execution in foundry operations and margin management, as these will critically influence the company’s turnaround trajectory.

Subscribe now

Cadence Design Systems (NASDAQ: CDNS)
🚀 Cadence Hits Record Backlog as AI Wave Boosts Outlook

What The Chip: Cadence delivered robust Q3 2025 earnings on October 27, 2025, achieving a record backlog of $7.0 billion and raising full-year guidance, driven by accelerating AI infrastructure investments and strategic partnerships.

Details:

📈 Record Backlog: Cadence reached a new backlog high of $7.0 billion, boosted partly by a $150 million catch-up from China. CFO John Wall noted the momentum positions Cadence strongly into FY26.

🤖 AI Infrastructure Boom: CEO Anirudh Devgan emphasized accelerating AI infrastructure builds as critical drivers. Cadence is essential to AI chip designs, collaborating closely with tech giants like NVIDIA, AMD, Broadcom, and the “Mag 7.”

📊 Impressive Q3 Financials:

• Revenue grew to $1.339 billion, up 10% YoY.

• Non-GAAP operating margin expanded to 47.6% from 44.8%.

• Non-GAAP EPS reached $1.93, a significant jump from $1.64 in Q3 2024.

• Free cash flow was strong at $277 million.

🚧 Short-term Supply Challenges: Demand for Cadence’s hardware verification platforms is robust, causing near-term hardware capacity constraints. The company is actively scaling manufacturing and inventory to manage these supply chain hurdles.

⚠️ Geopolitical Risks: Ongoing compliance with export controls, especially around China, introduces short- and long-term uncertainty. Regulatory oversight remains a critical watchpoint.

💡 Custom Silicon Opportunity: AI-driven complexity and custom silicon are surging, now representing 45% of Cadence’s business. The company’s deep integration into AI-centric chip design provides significant growth opportunities.

🌐 Strategic Acquisitions: The Hexagon acquisition positions Cadence to dominate growth in physical AI, 3D-IC, and high-performance computing. Expanded partnerships, including with OpenAI, underline this strategic focus.

🟢 Nvidia Collaboration: Cadence benefits significantly from its deep partnership with Nvidia, witnessing substantial performance gains in verification workloads. Regardless of market shifts toward custom or merchant silicon, Cadence’s broad collaboration model ensures sustained growth.

🎯 AI-driven Productivity: Cadence’s dual AI strategy—Design for AI and AI for Design—delivers measurable results, with verification tools like SimAI achieving 5x to 10x efficiency improvements, validated by leading customers such as Samsung, Qualcomm, and NVIDIA.

Why AI/Semiconductor Investors Should Care: Cadence’s growing backlog, fueled by the AI chip design surge and deep collaborations across the industry, positions it uniquely to capture outsized gains from both merchant and custom silicon trends. However, investors should remain cautious about geopolitical exposure and short-term supply constraints impacting near-term performance.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
NEW SEMICONDUCTOR COMMUNITY — 50% OFF FOUNDING RATE

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!

[Paid Subscribers] Intel Q3’25: Margins Rebound, AI Partnerships Heat Up

Event date: October 23, 2025

Executive Summary

*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences

Intel reported third‑quarter revenue of $13.7 billion, up 3% year over year, with GAAP earnings per share (EPS) of $0.90 and non‑GAAP EPS of $0.23. GAAP gross margin improved to 38.2% (non‑GAAP 40.0%), a sharp recovery from last year’s trough. Management emphasized two themes: disciplined balance‑sheet repair and a widening set of artificial intelligence (AI) opportunities across central processing units (CPUs), accelerators, and Intel Foundry. “AI is accelerating demand for compute and creating attractive opportunities across our portfolio,” CEO Lip‑Bu Tan said, while CFO David Zinsner added, “Current demand is outpacing supply, a trend we expect will persist into 2026.”

Intel guided Q4’25 revenue to $12.8–$13.8 billion and non‑GAAP EPS to $0.08 (GAAP EPS of –$0.14), noting that guidance excludes Altera following its deconsolidation in September. The company bolstered liquidity with $5.7B in U.S. government funding this quarter (part of a broader $8.9B commitment), plus strategic equity from NVIDIA ($5.0B, expected to close in Q4) and SoftBank Group ($2.0B). Cash and short‑term investments ended the quarter at $30.9B.

Growth Opportunities

AI everywhere, x86 in the mix. Intel’s pitch is less “GPU or bust” and more “hybrid compute,” where x86 CPUs remain the traffic cops for data‑center AI—routing, orchestrating, and increasingly handling inference alongside accelerators. On the PC side, the ongoing Windows 11 refresh and early AI PC adoption are tailwinds; Intel expects the 2025 client consumption total addressable market (TAM) to approach 290 million units, marking a second straight year of growth.

Read more

What The Chip Happened?
🔌 Texas Instruments: Navigating a Slow Recovery by Cutting Wafer Production to Safeguard Cash Flow
🤖 Intuitive: da Vinci 5 Drives Robust Q3 Results, Raises Outlook
🟢 “$505M and 800MW” — Galaxy’s Crypto–AI Flywheel Spins Faster
[Texas Instruments Q3 2025 Results: Playing it Smart in a Sluggish Market]

Read time: 7 minutes
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Texas Instruments (NASDAQ: TXN)
🔌 Texas Instruments: Navigating a Slow Recovery by Cutting Wafer Production to Safeguard Cash Flow

What The Chip:
On October 21, 2025, TI reported Q3 revenue of $4.74 billion (+14% year-over-year) and earnings per share (EPS) of $1.48, including an unexpected $0.10 reduction due to restructuring charges and operational adjustments. Management guided Q4 revenue lower to $4.22–$4.58 billion, and EPS down to $1.13–$1.39, signaling plans to reduce wafer starts to control inventory amid a slower-than-expected market recovery.

Details:

📈 Growth Across Segments: CEO Haviv Ilan highlighted steady broad-based demand, noting revenue grew 7% sequentially and 14% year-over-year, driven primarily by the Analog (+16%), Embedded Processing (+9%), and Other (+11%) segments.

🧭 Market Breakdown:

• Industrial: Up roughly 25% year-over-year but slowed to low single-digit growth sequentially, reflecting cautious customer investments.

• Automotive: Rose high single digits year-over-year and around 10% sequentially, benefiting from content-driven secular growth.

• Personal Electronics: Modest low single-digit year-over-year growth, improving sequentially in the high single digits.

• Enterprise Systems: Strong performance, surging about 35% year-over-year and 20% sequentially due to robust data-center demand.

• Communications Equipment: Jumped around 45% year-over-year and 10% sequentially, driven partly by data-center-related investments.

🧮 Margin Pressure and Inventory Management: Gross margins slightly compressed to 57% (down 50 basis points sequentially). CFO Rafael Lizardi explained the decline was due to lower utilization rates as TI intentionally trims wafer loadings. Depreciation, now running at $1.8–$2.0 billion annually and projected at $2.3–$2.7 billion next year (toward the lower end), is further pressuring margins. Management implied Q4 gross margins would likely trend towards the mid-50s percentage range.

🧱 Inventory Strategy: Inventory was essentially flat at $4.8 billion (215 days, down 16 days sequentially). Lizardi emphasized that TI aims to maintain or slightly reduce inventory levels in Q4, using moderated wafer production to prevent excess inventory build-up and protect free cash flow.

🔌 Accelerating Data Center Opportunity: TI disclosed plans to separately break out the Data Center segment in early 2026, noting it now represents about a $1.2 billion annual revenue run-rate and is growing rapidly at over 50% year-to-date. The growth spans multiple products including high-voltage DC power delivery, optical modules, and switching and interconnect solutions, driven by booming AI infrastructure investment.

🏭 Manufacturing Transitions and Restructuring: TI recorded a restructuring charge of $0.08 per share tied to shutting down its final 150mm (6-inch) fabs and consolidating certain R&D sites. Management reiterated their commitment to the significantly more cost-effective 300mm (12-inch) fabs, which allow more chips per wafer and superior economics long-term.

🌍 Macro Environment and Outlook: Ilan described the current semiconductor market recovery as moderate and slower-paced than typical cycles, pointing to customer hesitancy around CapEx spending due to lingering trade and tariff uncertainties. Customer inventories remain lean, pricing remains stable (declining at low single-digit percentages year-to-date), and TI maintains competitive lead times.

Why AI/Semiconductor Investors Should Care:
Investors should closely watch TI’s deliberate trade-off—accepting near-term margin pressure to strategically reduce wafer loadings and safeguard long-term free cash flow per share growth. More importantly, the growing significance of TI’s data-center business highlights a major AI-driven growth opportunity. If TI sustains its accelerated growth in data-center power and connectivity solutions while benefiting from improving scale efficiencies of its 300mm fabs, the company’s long-term earnings and free cash flow trajectory look compelling despite near-term margin volatility. However, uncertainty around industrial investment and geopolitical tariff headwinds could temper the near-term upside.

Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Intuitive Surgical (NASDAQ: ISRG)
🤖 Intuitive: da Vinci 5 Drives Robust Q3 Results, Raises Outlook 🚀

What The Chip: On October 21, 2025, Intuitive Surgical reported strong Q3 revenue of $2.51B (+23% YoY) driven by ~20% worldwide procedure growth and strong adoption of the new da Vinci 5 and Ion systems. Management increased full-year 2025 guidance, expecting da Vinci procedure growth of 17–17.5% and non-GAAP gross margin of 67–67.5%, despite a ~70 bps tariff impact.

Details:

📈 Procedure Momentum: Total procedures grew ~20%, led by da Vinci procedures (+19%) and Ion procedures (+52%, nearly 38,000 procedures). Utilization improved across platforms: multiport +4%, single-port (SP) +35%, and Ion +14%.

🧰 System Placements & Mix: Installed 427 da Vinci systems (+13% YoY), including 240 da Vinci 5 systems and 30 SP systems. Ion system placements were 50 units, slightly down from 58 units last year as customers prioritized utilization. Leasing accounted for 54% of placements.

💰 Financial Performance:

• Revenue segments: Instruments & Accessories up 20% to $1.52B, Systems revenue up 33% to $590M, and Service revenue up 20% to $396M.

• Non-GAAP net income rose to $867M ($2.40/share) from $669M ($1.84/share). GAAP net income reached $704M ($1.95/share).

• $1.92B spent on repurchasing 4.0M shares; free cash flow was $736M. Cash and investments totaled $8.43B.

🌐 Global Trends: U.S. procedures rose 18%, driven by da Vinci (+16%) and Ion (+48%). International procedures grew 25%, benefiting from strong uptake in India, Korea, Taiwan, and Brazil, offset by budget constraints in Japan and the UK, and competitive pressures in China.

⚙️ Tech Enhancements:

• da Vinci 5: New FDA-cleared features, including remote software updates, Force Gauge visual feedback, and integrated insufflation technology (used in ~90% of da Vinci 5 cases).

• Ion: AI-driven navigation significantly improved diagnostic accuracy. A recent Swiss study showed Ion achieving an 84.6% diagnostic yield, dramatically outperforming traditional bronchoscopy (23.1%).

• SP platform: Procedure growth surged 91%, driven by increased adoption in Korea and early U.S. utilization of new SP staplers.

📊 Updated 2025 Guidance:

• Procedure growth: raised to 17–17.5%.

• Non-GAAP gross margin: raised to 67–67.5%, inclusive of ~70 bps tariff impact.

• Operating expenses: expected to grow 11–13%.

⚠️ Investor Considerations:

• Margins: Pressured by tariffs and higher mix of lower-margin da Vinci 5 and Ion systems.

• Bariatric Weakness: U.S. bariatric surgeries declined high-single digits due to ongoing GLP-1 drug competition.

• China Competition: Continued price pressure amid slow tender processes.

Why AI/Semiconductor Investors Should Care: Intuitive Surgical continues to leverage robotics and AI-driven innovation, notably through its da Vinci 5 and Ion systems. These technological advancements demonstrate increased efficiencies, driving adoption and utilization. Despite margin pressures from tariffs and competitive international markets, Intuitive’s sustained growth in procedures and strategic portfolio diversification—including refurbished systems—indicates robust potential. The company’s ability to deliver consistent growth and upgrade its financial outlook underscores its strength and strategic position within the rapidly expanding surgical robotics market.

Subscribe now

Galaxy Digital (NASDAQ: GLXY; TSX: GLXY)
🟢 “$505M and 800MW” — Galaxy’s Crypto–AI Flywheel Spins Faster

What The Chip: On October 21, 2025, Galaxy reported Q3 net income of $505M, diluted EPS $1.01 (adjusted $1.12), and adjusted EBITDA $629M. Management also firmed up the Helios AI/HPC buildout with CoreWeave—now contracted across the full 800MW—and secured financing to bring Phase I online in 1H26.

Details:

💰 Blowout quarter: Net income hit $505M (up 1,546% Q/Q), with adjusted EBITDA at $629M (up 198% Q/Q). Cash + stablecoins rose to $1.9B and total equity to $3.2B as of Sept 30.

🏗️ Helios = AI/HPC growth option (but 2026 revenue start):
• Phase I: 133MW critical IT load, delivery 1H26 (lease with CoreWeave).
• Phase II: 260MW, 2027 delivery (lease executed).
• Phase III: 133MW, 2028 start.
• Total committed critical IT load: 526MW (of 800MW gross power across Phases I–III).
• Financing: $1.4B project facility secured—fully funds the $1.7B Phase I build.
Galaxy expanded the campus to >1,500 acres with 2.7GW of additional potential power under ERCOT study. Management still guides to immaterial P&L from Data Centers until 1H26.
Novogratz on the ramp: “We start cash flowing… and 18 months after that, it’s hundreds of millions of dollars per quarter.”

🛠️ Product + capital: Galaxy launched GalaxyOne (Oct 6) to offer U.S. individuals high‑yield cash plus crypto and equities trading in one platform. On Oct 10, the company announced a $460M equity investment from a large asset manager, with $325M net proceeds earmarked for Helios and general purposes.

⚠️ Watch the risks:
• Concentration: CoreWeave is currently the sole contracted Helios tenant (800MW). Galaxy may diversify tenants on future power approvals.
• Cycle risk: Novogratz cautioned, “It feels like we’re in some stage of a bubble around data centers and AI… there will be an inevitable pullback… I don’t know if [it’s] in 3 months or 3 years.”

Why AI/Semiconductor Investors Should Care

Galaxy is morphing from a pure‑play crypto operator into a crypto–AI infrastructure hybrid: robust trading/asset‑management cash generation today, with a contracted, utility‑like lease ramp at Helios beginning 1H26. If CoreWeave’s AI/HPC demand persists, Helios could become a multi‑year free‑cash‑flow engine, diversifying away from crypto cyclicality. The bear case: data center/AI overbuild, tenant/credit and ERCOT/power risks, plus crypto drawdowns, could pressure cash generation and delay returns. Near‑term, track Helios construction milestones, power energization and cash‑flow ramp cadence, along with AUM/staking growth and loan‑book health—the levers that will determine how quickly Galaxy converts momentum into durable FCF.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!

[Paid Subscribers] Texas Instruments Q3 2025 Results: Playing it Smart in a Sluggish Market

Event date: October 21, 2025

Executive Summary

*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences

On October 21, Texas Instruments (Nasdaq: TXN)—or TI, as investors fondly call it—released third-quarter results that showcased solid growth despite a cautious semiconductor recovery. TI posted $4.74 billion in revenue, marking a healthy 14% increase year-over-year and 7% higher than the previous quarter. Net income came in at $1.36 billion, translating into earnings per share (EPS) of $1.48. However, there was a minor hiccup: EPS included an unexpected $0.10 reduction, primarily from restructuring charges linked to efficiency improvements and the planned closure of older fabs.

CEO Haviv Ilan kept things straightforward: “Revenue increased 7% sequentially and 14% from the same quarter a year ago, with growth across all end markets.” TI’s well-known cash-generation machine hummed along, delivering $6.9 billion in operating cash flow over the past year, and $2.4 billion in free cash flow, including incentives from Uncle Sam’s CHIPS Act. TI continued to reward shareholders generously, returning $6.6 billion over the last year through dividends and stock buybacks, marking its 22nd consecutive year of dividend hikes—talk about commitment.

Looking ahead to the holiday quarter, TI projected revenues between $4.22 billion and $4.58 billion, with EPS ranging from $1.13 to $1.39. The outlook incorporates changes from new U.S. tax legislation, setting an expected tax rate near 13%.

Now, let’s dig deeper into what’s driving TI’s steady performance, the bumps along the way, and how the semiconductor giant plans to navigate uncertain market waters.

Growth Opportunities: Betting Big on Industrial, Auto, and Data Center

Read more

What The Chip Happened?
🚀 TSMC: AI foundry engine hums — Q3 beat, ~60% GM, N2 on deck
☁️ AWS Outage Rocks the Cloud—And Everyone Else’s Boat, Too
⚡️ Google Cloud Powers Up with NVIDIA’s Blackwell GPUs—Say Hello to G4!
[TSMC’s Q3 2025: Record Profits Powered by AI and Advanced Chips]

Read time: 7 minutes
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

TSMC (NYSE: TSM; TWSE: 2330)
🚀 TSMC: AI foundry engine hums — Q3 beat, ~60% GM, N2 on deck

What The Chip: On October 16, 2025, TSMC flexed its silicon muscle, reporting Q3 EPS of NT$17.44 (US$2.92 per ADR) on revenues hitting $33.1 billion, soaring 40.8% year-over-year and rising 10.1% sequentially. The AI gold rush keeps driving explosive growth at the foundry giant, with leading-edge fabs booked solid. Management confidently projects Q4 revenue of $32.2–$33.4 billion and rock-solid margins near 60%.

Details:

📈 Headline numbers & profitability. Revenue NT$989.9B ($33.1B); net income NT$452.3B; gross margin 59.5%, operating margin 50.6%, net margin 45.7%; ROE 37.8%. Cash from ops NT$427B, CapEx NT$287B, implying ~NT$140B FCF; cash & marketable securities NT$2.8T (~$90B). CFO Wendell Huang: “Supported by strong demand for our leading‑edge process technologies.”

🧠 Node mix shows AI tilt. N3 = 23%, N5 = 37%, N7 = 14%; nodes ≤7nm = 74% of wafer revenue. Translation: most sales come from the processes used for AI accelerators and top smartphone chips.

🖥️ Platform mix (Q3). HPC (AI/data center) held 57% of revenue (flat QoQ); Smartphone rose 19% QoQ to 30%; IoT up 20% to 5%; Auto up 18% to 5%; DCE down 20% to 1%.

🎯 Q4 outlook. $32.2–$33.4B revenue (-1% QoQ / +22% YoY midpoint), GM 59–61%, OPM 49–51%. Gross margin should tick ~50 bps higher at midpoint helped by FX; inventory days 74 and A/R days 25 support healthy flow‑through.

🔥 AI demand still “insane.” CEO C.C. Wei said AI demand is stronger than three months ago, adding full‑year 2025 revenue should grow “close to mid‑30s%” in USD. On the GPU vs. ASIC mix: “No differentiation in front of TSMC—we support all types.”

📦 Advanced packaging = bottleneck (and opportunity). CoWoS (2.5D/3D packaging that stitches compute dies to HBM memory) remains tight. TSMC plans two advanced‑packaging fabs in Arizona and is partnering with a major OSAT whose Arizona plant breaks ground earlier. Wei also noted advanced‑packaging revenue is approaching ~10% of total—material and growing.

🧪 Next nodes explained. N2 (gate‑all‑around nanosheets for better speed/power) enters volume later this quarter; N2P adds efficiency in 2H’26; A16 with Super Power Rail (backside power delivery that frees front‑side routing for higher density) targets 2H’26. These nodes matter because they cut energy per computation, a key cost in AI training/inference.

🌍 Costs, FX & policy risk. Overseas‑fab dilution trimmed to ~2% in 2H’25 (vs. prior 2–3%); over the next several years expect 2–3% early, widening to 3–4% later. Every 1% USD/TWD move shifts GM by ~40 bps. Management also flagged tariff policy risk for price‑sensitive consumer end‑markets, though smartphone inventories look healthy (no “prebuild” concern).

Why AI/Semiconductor Investors Should Care
TSMC remains the gatekeeper of AI compute: 74% of revenue now comes from ≤7nm, and N2/A16 extend its lead as packaging scales in the U.S. The company narrowed 2025 CapEx to $40–$42B—a tell that multi‑year AI demand stays robust—while preserving ~60% gross margins. Key watch‑items are packaging capacity and tariff/FX swings; if TSMC eases bottlenecks and ramps N2 on schedule, it should continue to capture an outsized share of the AI infrastructure economics.

Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Amazon.com, Inc. (NASDAQ: AMZN)
☁️ AWS Outage Rocks the Cloud—And Everyone Else’s Boat, Too

What The Chip: Early Monday morning, October 20, 2025, AWS hit some stormy weather in its core region, US-EAST-1 (N. Virginia), causing a ripple effect felt across millions of users and thousands of businesses worldwide. The disruption began at 12:11 a.m. PT due to issues with DNS resolution for DynamoDB API endpoints, and while AWS claimed victory over the initial problem by 2:24 a.m. PT, lingering issues meant the internet’s Monday blues lasted much longer.

Details:

🕰 Timeline turbulence: Things went sideways at exactly 12:11 a.m. PT, when AWS customers began seeing errors spike across numerous services. AWS quickly traced it back to a DNS resolution glitch affecting DynamoDB endpoints. By 2:24 a.m. PT, the initial DNS snafu was resolved—but recovering fully proved trickier, causing service hiccups into the afternoon.

🛠 Why was the recovery so messy? Turns out, the DNS issue cascaded downstream into deeper AWS infrastructure layers, specifically affecting Lambda invocation, Network Load Balancer (NLB) health checks, and internal networking for EC2 instances. Basically, AWS’s internal plumbing sprang leaks all over the place.

🌊 Ripple effects were massive: The list of impacted services reads like the guest list to a tech celebrity party—Snapchat, Roblox, Fortnite, Coinbase, Robinhood, Lyft, Venmo, and even Amazon’s very own Prime Video, Alexa, Ring, and the retail website. The widespread nature of this outage highlights just how interconnected—and fragile—today’s digital world really is.

📈 Just how big? Huge. Reports ranged widely, with 6.5 million to 11 million user outage reports across 1,000 to 2,500 companies, making it the biggest cloud meltdown since the infamous CrowdStrike incident of 2024. If outages had their own Richter scale, this one would be pushing seismic.

🚨 Not a cyberattack (this time): AWS quickly ruled out a malicious breach or cyberattack. This outage was purely an inside job—a technical fault that triggered a domino effect through AWS’s massive service ecosystem.

🌍 The gravitational pull of US-EAST-1: AWS’s oldest and largest data-center cluster, located in Northern Virginia, remains the backbone for countless global operations. It’s the “New York City” of cloud computing—everything passes through it, and when it sneezes, the whole internet catches a cold.

🏛 Regulators are watching closely: UK financial authorities quickly raised eyebrows and renewed their calls to treat AWS as a “critical third-party provider,” potentially increasing regulatory scrutiny and oversight. In other words, AWS might soon feel the heat of being deemed “too big to fail.”

Why AI/Semiconductor Investors Should Care
This event is a wake-up call for investors in AI-driven companies heavily reliant on cloud infrastructure, especially those married exclusively to AWS via platforms like Amazon Bedrock (hello, Anthropic and its Claude models). The outage underscores the critical importance of investing in multi-cloud, multi-region capabilities—companies able to rapidly reroute workloads between different providers or regions will command a resilience premium.

Subscribe now

Alphabet (NASDAQ: GOOGL)
⚡️ Google Cloud Powers Up with NVIDIA’s Blackwell GPUs—Say Hello to G4!

What The Chip: On October 20–21, 2025, Google Cloud officially flipped the switch on its newest virtual machines—G4 VMs, turbocharged by NVIDIA’s RTX PRO 6000 Blackwell GPUs. Designed to crush high-throughput AI inference, fine-tuning, visual computing, and real-time simulations (think Omniverse and robotics).

Details:

🎯 Blackwell Arrives in the Cloud: Every G4 VM packs NVIDIA’s RTX PRO 6000 Blackwell GPUs, each boasting 96 GB of ultra-fast GDDR7 memory, 5th-gen Tensor Cores (featuring FP4 precision for optimized inference), and 4th-gen RT cores for buttery-smooth ray-traced visuals. Choose your power level from 1, 2, 4, or even 8 GPUs per VM—with fractional GPUs coming soon for cost-savvy customers.

🚦 Fast Lane for Multi-GPU Performance: Google didn’t just throw GPUs together; they built a software-defined PCIe peer-to-peer fabric that supercharges multi-GPU teamwork. Collective operations like All-Reduce and All-Gather see performance boosts up to 2.2× faster. Tensor-parallel inference gets a sweet 168% throughput jump with latency dropping by 41%.

🏎️ The Leap from G2 is No Joke: Compared to the previous-gen G2 (L4 GPUs), G4 VMs deliver a staggering 9× throughput increase. Positioned as the sweet spot, G4 fills the performance gap perfectly between the budget-friendly G2 series and heavyweight A-series training monsters.

🧰 Built on Solid Ground (and Fast Disks!): G4 VMs run atop AMD’s latest-and-greatest EPYC Turin (5th Gen) CPUs. One twist—Google replaced traditional Persistent Disks with Hyperdisk for turbocharged I/O performance, and you can pile on Titanium local SSDs up to 12 TB.

🌌 Omniverse & Isaac at Your Fingertips: The cherry on top—NVIDIA’s flagship simulation platforms, Omniverse and Isaac Sim, are now ready for one-click deployment via the Google Cloud Marketplace.

🎬 Graphics and Media, Upgraded: G4 isn’t just for inference—it’s ready to roll as a virtual workstation powerhouse, fully supporting NVIDIA’s RTX virtual workstation (RTX vWS) stack for stunning graphics, real-time rendering, and high-quality media encoding.

💬 Customer Cheers Already Pouring In: Creative giants like WPP call Omniverse on G4 “a true engine for our creative transformation,” while simulation experts at Altair praise G4 for handling their “most demanding fluid dynamics workloads.”

Why AI/Semiconductor Investors Should Care

The arrival of G4 VMs signals Google’s serious bet on AI inference and visual computing, precisely where market demand is surging. Google’s unique, built-in P2P GPU fabric boosts real-world performance metrics—throughput and latency—which are critical for enterprise-scale deployments. For NVIDIA investors, this launch expands Blackwell GPUs far beyond traditional workstations, deepening NVIDIA’s cloud footprint and boosting adoption of Omniverse and Isaac ecosystems. Investors should keep eyes on availability, pricing, and potential tariff headwinds—but overall, G4 is perfectly timed to ride the wave of AI’s next big growth spurt.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!

[Paid Subscribers] TSMC’s Q3 2025: Record Profits Powered by AI and Advanced Chips

Event date: October 16, 2025

Executive Summary

*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences

If you’ve been wondering just how much the world loves AI, TSMC’s latest results might answer that clearly. Taiwan Semiconductor Manufacturing Company (TSMC), the world’s most influential chipmaker, has just reported another blockbuster quarter, thanks to surging demand for cutting-edge chips.

In the third quarter, TSMC posted revenue of NT$989.92 billion (around US$33.10 billion), jumping nearly 41% from last year. Net income soared to NT$452.30 billion, lifting earnings per share (EPS) to NT$17.44 (US$2.92 per ADR). In short: business is booming, and the AI wave is carrying TSMC higher.

AI remains the superstar driving TSMC’s growth. CEO Dr. C.C. Wei highlighted that demand for AI-related technology isn’t just strong—it’s accelerating. Companies are racing to deploy new AI capabilities, and TSMC is perfectly positioned to supply the crucial advanced chips.

TSMC isn’t just talking to chipmakers these days; they’re directly engaging with their customers’ customers. Everyone from enterprise tech giants to governments focused on sovereign AI is knocking at TSMC’s door, securing their spot in line for these indispensable semiconductors.

Growth Opportunities

Read more

What The Chip Happened?
🚀 Starcloud: Taking Nvidia Data Centers Off‑Planet
💵 AMD Gets Dual Street Highs: HSBC to $310, Wedbush to $270 on AI GPU Momentum
🧠 ASML’s Q3: Big Q4 Setup, High‑NA Matures, and a Bold AI Bet
[ASML Q3 2025: EUV Shines, AI Fuels the Future, China Softens October 15, 2025]

Read time: 7 minutes
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Starcloud (Private) Nvidia (NVDA)
🚀 Starcloud: Taking Nvidia Data Centers Off‑Planet

What The Chip: On October 15, 2025, Starcloud—a Redmond, WA startup in NVIDIA’s Inception program—detailed plans for Starcloud‑1, launching November 2025 with an NVIDIA H100 on board. The company projects 10× lower energy costs and 10× CO₂ savings over a data center’s life by shifting compute to orbit.

Details:

🛰️ First H100 in space. Starcloud‑1 is a ~60‑kg (small‑fridge‑size) satellite that the company says will deliver ~100× more GPU compute than any prior space mission—marking the H100’s “cosmic debut.”

☀️ Power + cooling advantage. Orbit provides near‑continuous solar exposure (terminator‑line orbit) and the vacuum of space as a heat sink, avoiding evaporative cooling and conserving fresh water that terrestrial data centers would otherwise need.

💸 Cost & carbon math. “In space, you get almost unlimited, low‑cost renewable energy,” said cofounder/CEO Philip Johnston. Starcloud projects 10× cheaper energy (even after launch costs) and 10× CO₂ savings over the facility’s life versus running on Earth.

🧱 Scale vision. The roadmap calls for a ~5‑gigawatt orbital data center with super‑large solar and radiator panels roughly 4 km × 4 km—a size they argue unlocks hyperscale economics off‑planet.

🔥 Real‑time edge in orbit. Initial workloads target Earth observation, including wildfire detection and distress‑signal response—running inference on‑orbit to cut alert times from hours to minutes.

🛰️📡 Data deluge, processed in place. SAR (synthetic‑aperture radar) feeds can reach ~10 GB/s; doing inference in space trims downlink needs and speeds up insights for 3D mapping and monitoring.

🧠 LLMs in orbit. The team plans to run Gemma (Google’s open model) on the H100 to prove large‑model inference off‑planet; they credit NVIDIA Inception for technical support, access to experts, and GPU resources.

🧩 Roadmap & risks. Future missions aim to integrate NVIDIA Blackwell, which Johnston expects could deliver up to 10× the in‑orbit AI performance vs. Hopper (H100). Watch headwinds: radiation hardening, bandwidth/downlink limits, on‑orbit servicing, regulatory/spectrum approvals, debris risk, and launch cadence/capex.

Why AI/Semiconductor Investors Should Care
Energy and cooling now anchor data‑center economics; if Starcloud proves H100‑class compute can operate reliably with dramatically lower OPEX in orbit, it opens a new capex cycle in space‑based AI infrastructure while easing grid and water constraints on Earth. Near‑instant analytics for Earth observation is a real commercial wedge today—and if Blackwell’s step‑function arrives on‑orbit, it extends NVIDIA’s datacenter GPU TAM beyond Earth. The opportunity is bold, but execution will hinge on launch costs, radiation‑tolerant hardware lifetimes, and regulatory throughput—key variables to track before betting big.

Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Advanced Micro Devices (NASDAQ: AMD)
💵 AMD Gets Dual Street Highs: HSBC to $310, Wedbush to $270 on AI GPU Momentum

What The Chip: On October 15, 2025, HSBC hiked AMD’s price target to a Street‑high $310 (Buy) and Wedbush raised to $270 (Outperform), citing accelerating visibility from new, large AI GPU deals—most notably OpenAI’s 6‑gigawatt buildout and Oracle’s 50,000‑GPU MI450 cluster.

Details:

📈 Two big upgrades, one day. HSBC’s Frank Lee lifted AMD’s target from $185 → $310 (+~68%) and kept a Buy; Wedbush’s Matt Bryson went $190 → $270 with Outperform. Both moves lean on rising confidence in AMD’s AI GPU pipeline.

🤝 OpenAI x AMD: 6 GW of compute. AMD and OpenAI signed a multi‑year, multi‑generation pact for 6 gigawatts of AMD Instinct GPUs, with the first 1 GW of MI450 deployments starting 2H’26. CEO Lisa Su: “We are thrilled to partner with OpenAI to deliver AI compute at massive scale.” CFO Jean Hu added the deal is “expected to deliver tens of billions of dollars in revenue for AMD.”

☁️ Oracle supercluster: 50,000 MI450s. At Oracle AI World, Oracle said OCI will launch a publicly available AI supercluster with 50,000 AMD Instinct MI450 GPUs beginning Q3’26, expanding through 2027. AMD’s Forrest Norrod said Oracle customers gain “powerful new capabilities for training, fine‑tuning, and deploying the next generation of AI.”

🔎 HSBC’s math is more aggressive. Lee models up to $37.5B in 2027 AI GPU revenue (vs. Street $26.8B), arguing AMD is narrowing the performance gap with Nvidia and gaining supply visibility as TSMC capacity improves.

🧮 Wedbush’s valuation frame. Bryson applies ~30× P/E to FY2027 EPS of $9 (plus net cash) to justify $270, grounded in multi‑year AI hardware ramps. Market commentary also flags AMD’s current rich multiple (around 133× P/E), a reminder execution must land on time.

💵 Deal economics: big per‑watt revenue. Wedbush pegs OpenAI’s expansion as a $20B per gigawatt revenue pool for AI hardware—illustrating the magnitude of the AI buildout if AMD captures meaningful share.

🆚 Competitive backdrop stays fierce. While HSBC argues the gap is closing, independent testing highlighted by Barron’s still shows Nvidia’s latest GB200 NVL72 system leading AMD on inference performance and efficiency—one reason bulls and bears both have ammo.

🧩 Supply & timing risk. The MI450 ramp begins in 2H’26 (OpenAI) and Q3’26 (Oracle); any slip in CoWoS/HBM capacity or platform readiness could push revenue right on the curve that underpins these targets. (Press releases and analyst notes emphasize forward‑looking uncertainties.)

Why AI/Semiconductor Investors Should Care

These calls reflect a step‑change in revenue visibility. The OpenAI (6 GW) and Oracle (50k GPUs) wins move AMD from “credible second source” to strategic supplier at hyperscale—bolstering multi‑year order books and supporting more ambitious 2026–2027 AI revenue and EPS trajectories. The bull case hinges on supply ramp (TSMC/packaging/HBM) and competitive parity trending in AMD’s favor; the bear case points to execution risk and Nvidia’s continued performance lead. Positioning and risk management aside, the size and timing of these deployments explain why targets jumped to $310 and $270 today.

Subscribe now

ASML Holding N.V. (NASDAQ: ASML)
🧠 ASML’s Q3: Big Q4 Setup, High‑NA Matures, and a Bold AI Bet

What The Chip: On Oct 15, 2025, ASML reported Q3 net sales of €7.5B, 51.6% gross margin, and €2.1B net income, with €5.4B in bookings (€3.6B EUV). Management guided Q4 sales to €9.2–9.8B (GM 51–53%) and reiterated FY25 ~+15% y/y sales growth and ~52% GM—plus a key line: “We do not expect 2026 total net sales to be below 2025.”

Details:

📊 Quarter in line; bigger Q4 ahead. Q3 sales €7.5B (down ~2.3% QoQ), GM 51.6% (down 210 bps QoQ), EPS €5.49. ASML expects a very strong Q4, with the midpoint implying ~27% QoQ revenue step‑up; FY25 net sales ~€32.5B and ~52% GM. CFO Roger Dassen (ASML’s CFO) said the print was “within guidance” and Q4 is “as planned.”

📦 Bookings healthy and EUV‑heavy. €5.4B of Q3 net bookings, about two‑thirds EUV (€3.6B); mix 53% logic / 47% memory. Net system sales in Q3 skewed 65% logic / 35% memory, with Installed Base Management (service + options) at €2.0B (about 26% of total net sales).

💡 Litho intensity rising—especially in DRAM. CEO Christophe Fouquet (ASML’s CEO) said, “We see litho intensity continue to develop positively as EUV adoption gains momentum.” Management pushed back on fears that DRAM’s move from 6F² to 4F² reduces EUV layers: they do not expect EUV layers to drop; over time, they expect layers to grow, with some boost to advanced DUV masks as structures get more complex.

🔬 High‑NA EUV: maturity improving; two EXE tools in Q4. Customers have now run >300,000 wafers on High‑NA platforms, and early maturity looks ahead of Low‑NA at a similar stage. ASML plans to recognize two High‑NA EXE systems in Q4—helping revenue but dilutive to GM. Dassen said High‑NA margins are “very low positive” today and will improve with volume, though still dilutive through 2030.

🧱 Advanced packaging beachhead—XT:260. ASML shipped its first product for 3D integration, the TWINSCAN XT:260 (i‑line). It offers up to 4x productivity vs. existing solutions and targets advanced packaging—critical as chipmakers stitch together more compute/HBM in 2.5D/3D stacks. Fouquet: “There will be more” products for 3D integration, and customer interest into 2026 looks strong.

🤝 Strategic AI move with Mistral AI. ASML closed a partnership to embed AI across its holistic portfolio (scanners, metrology, and inspection) to boost tool performance, productivity, and customer yield—and to speed ASML’s own development. ASML also invested €1.3B as lead investor in Mistral’s Series C, taking ~11% and a strategic committee seat. Dassen called it a “very strategic partnership.”

🌏 2026 setup: EUV up, DUV down; China normalizes. Management expects China sales to decline significantly in 2026 vs. 2024–25 (after two very strong years), weighing on DUV, while EUV should increase on leading‑edge logic and advanced DRAM demand. ASML flagged that tariff clarity versus last quarter has reduced uncertainty, aiding customer planning.

💸 Cash, dividend, and buyback. Cash & ST investments fell to €5.1B (from €7.2B in Q2), reflecting working capital tied to High‑NA cycle times, the Mistral investment, dividends, and buybacks. ASML will pay a €1.60 interim dividend on Nov 6, 2025. Q3 buybacks totaled ~€148M, bringing 2022–2025 program totals to 9.0M shares / €5.9B; ASML does not expect to finish the €12B program by year‑end and intends to announce a new program in Jan 2026.

CEO Fouquet: “We do not expect 2026 total net sales to be below 2025.”
CFO Dassen on Mistral: “Our partnership allows us to embed AI across our entire portfolio… and improve time to market.”
Fouquet on tech progress: “High‑NA maturity is well ahead of where Low‑NA was at the same stage.”

Why AI/Semiconductor Investors Should Care

EUV demand is broadening (advanced logic and DRAM), lithography layers are rising, and ASML’s Q4 surge plus 2026 ≥ 2025 outlook suggest the AI capex cycle is translating into tools. The High‑NA ramp (though margin‑dilutive near term) and 3D‑integration entry expand ASML’s strategic choke points in performance‑hungry AI chips and HBM stacks. Risks: a China DUV reset could trim mix and gross margin, High‑NA remains volume‑ and cycle‑time‑sensitive, and working capital will stay elevated as ASML builds ahead of demand. Net‑net, the order mix (EUV ~67% of Q3 bookings), technology positioning, and the AI software tie‑in with Mistral put ASML in the critical path of the AI infrastructure build—exactly where investors want it.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!

[Paid Subscribers] ASML Q3 2025: EUV Shines, AI Fuels the Future, China Softens October 15, 2025

Event date: October 15, 2025

Executive Summary

*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences

ASML just wrapped up its third-quarter earnings for 2025, and the story remains compelling: steady financial execution, promising tech breakthroughs, and some shifting dynamics, especially regarding China. Net sales for the quarter landed neatly at €7.5 billion, right within the company’s previous guidance. Gross margins held firm at 51.6%, while net income clocked in at a robust €2.1 billion, translating to earnings of €5.49 per share. Not bad for a quarter’s work.

CEO Christophe Fouquet summed it up clearly: “Our Q3 results were exactly where we said they’d be, a solid quarter for ASML.” He highlighted exciting momentum around AI-driven investments, particularly boosting demand in cutting-edge logic and advanced memory (DRAM). The continued uptake of ASML’s extreme ultraviolet (EUV) technology underpins this trend, making chips faster, smaller, and smarter. In short, the semiconductor race is hotter than ever, and ASML remains squarely at the center of the action.

Growth Opportunities

The wave of AI continues reshaping the chip market, and ASML is riding high. More customers are now tapping into AI’s immense potential, pushing lithography intensity to new heights. ASML’s EUV systems, critical for the industry’s top-tier chip makers, saw a healthy €3.6 billion in new orders in Q3 alone.

Read more

What The Chip Happened?
🔥 Aehr Test Systems: Hyperscaler Burn‑In Orders Signal AI Ramp
⚡ Vertiv + NVIDIA push 800V DC from concept to “engineering‑ready” for AI factories
🚀 OpenAI’s Three‑Way Chip Play: From Buyer to Builder
[Aehr Test Systems Starts Fiscal 2026: AI Buzz Builds, Margins Take a Hit]

Read time: 7 minutes
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Aehr Test Systems (NASDAQ: AEHR)
🔥 Aehr Test Systems: Hyperscaler Burn‑In Orders Signal AI Ramp

What The Chip: On October 6, 2025, Aehr reported Q1 FY26 results showing smaller revenue but growing momentum in AI chip test & burn‑in across both package and wafer levels. A “world‑leading hyperscaler” placed follow‑on production orders for Aehr’s new Sonoma packaged‑part burn‑in systems, while Aehr highlighted the first production WLBI (wafer‑level burn‑in) systems for AI processors at a premier OSAT.

Details:

⚡ Hyperscaler ramp on Sonoma: A lead hyperscaler “placed multiple follow‑on volume production orders…and requested shorter lead times to support higher‑than‑expected volumes,” with plans to expand capacity and “introduce new AI processors…to be tested and burned in on our Sonoma platform” at a top OSAT, said CEO Gayn Erickson. Aehr is co‑developing burn‑in for future gens at both package and wafer levels.

🧪 WLBI beachhead at scale: Aehr says it delivered the world’s first production WLBI systems for AI processors, installed at a premier global OSAT—a visible showcase that is already drawing approaches from other AI suppliers. Aehr is also partnering with this OSAT on advanced wafer‑level test & burn‑in for HPC/AI, offering a turnkey path from design to high‑volume production.

🛠️ Power & automation upgrades: Since acquiring Incal, Aehr raised Sonoma’s power‑per‑device to 2,000W, increased parallelism, and added full automation via an integrated device handler. Ten companies visited Fremont last quarter to see the enhanced system; feedback was “very positive,” and management expects these features to drive new applications and orders.

📡 Silicon photonics & HDD tailwinds: Aehr upgraded a major silicon photonics customer’s FOX‑XP to 3.5kW per wafer in a nine‑wafer configuration with a fully integrated WaferPak Aligner for single‑touchdown wafer‑level burn‑in—more orders expected this fiscal year. In storage, multiple FOX‑CP systems shipped to a top HDD supplier to stabilize next‑gen read/write heads as AI data growth boosts HDD demand.

⚙️ Power semis (GaN & SiC): Engagements expanded in GaN (data center power, auto, energy); Aehr is building a “large number” of WaferPaks for new designs. SiC growth should skew 2H FY26 with upgrades, WaferPaks, and capacity additions. Aehr highlighted shipment of its first 18‑wafer high‑voltage FOX‑XP (±2,000V) to test 100% of EV inverter devices in one pass.

🧠 Memory opportunity—HBF on deck: Aehr’s NAND WLBI benchmark progressed with a new fine‑pitch WaferPak. Meanwhile, High‑Bandwidth Flash (HBF)—pitched as 8–16× the capacity of HBM at similar cost—is shifting tester requirements upward in power and parallelism; Aehr is preparing proposals to meet these needs within its 18‑wafer FOX‑XP infrastructure. Separately, a paid WLBI evaluation with a top‑tier AI processor supplier includes a custom high‑power WaferPak and production test program.

💰 Numbers that matter (Q1 FY26, ended Aug 29): Revenue $11.0M (↓ 16% YoY); GAAP net loss $(2.1)M or $(0.07); non‑GAAP net income $0.2M or $0.01. Non‑GAAP GM 37.5% (down from 54.7% on lower volume/mix and more third‑party content). Bookings $11.4M (~1.04× book‑to‑bill); backlog $15.5M, effective backlog $17.5M. Cash $24.7M, no debt. Facility renovation now complete ($6.3M total), boosting manufacturing capacity ≥5×. CFO Chris Siu: “We believe investment in this facility renovation has increased our overall manufacturing capacity by at least 5x…and we are more ready than ever to support the growth of our customers.”

🧭 Guidance & watch‑outs: Management beat Street for the quarter but withheld formal guidance due to 2025 tariff‑related uncertainty. Mix is a swing factor: consumables were $2.6M (24%) vs $12.1M (92%) last year’s Q1, pressuring margin. Concentration risk remains (lead hyperscaler), WLBI adoption/evaluations take time, and early AI programs can shift between package‑level and wafer‑level approaches. Still, Erickson summed it up: “We’re excited about the year ahead…nearly all our served markets will see order growth…though we remain cautious due to tariff‑related uncertainty.”

Why AI/Semiconductor Investors Should Care

Reliability is now a first‑order constraint in AI data centers—high‑power processors and advanced packages (multi‑die compute + HBM) can destroy yield if early failures slip through. Aehr sells the screening step: burn‑in at package level (Sonoma) and increasingly at wafer level (FOX‑XP). Shifting left (WLBI) can cut scrap of costly CoWoS‑style modules and reduce rack‑level burn‑in power/time—clear economics if volumes scale. The OSAT showcase, hyperscaler orders, and 5× capacity set Aehr up for multi‑year AI demand, with optionality in silicon photonics, HDD, GaN, and a potential HBF wave. Balance that with margin mix, order timing, and tariff risks—but if WLBI becomes standard for AI/HPC, Aehr’s high‑power niche looks strategically leveraged.

Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Vertiv (NYSE: VRT)
⚡ Vertiv + NVIDIA push 800V DC from concept to “engineering‑ready” for AI factories

What The Chip: On October 13, 2025, Vertiv said its 800 VDC power platform has advanced from concept to engineering readiness, with first releases planned for 2H26—timed to support NVIDIA’s Rubin Ultra rollout in 2027. The work stems from a collaboration announced earlier this year.

Details:

🚀 Timeline & scope. Vertiv moved its 800 VDC “unit of compute” platform toward production, targeting H2’26 availability to match NVIDIA’s rack‑scale roadmap for Rubin Ultra in 2027.

🔌 Why 800 VDC now. Today’s 54 VDC in‑rack distribution—built for kilowatt racks—hits limits as AI racks push toward megawatt class. Higher voltage slashes current for the same power, which cuts resistive losses and copper mass while reducing conversion stages. NVIDIA frames 800 VDC as the path to support ≥1 MW per rack and improve end‑to‑end power efficiency (they cite up to ~5% efficiency gain and up to ~30% TCO reduction from architectural changes).

🧩 What Vertiv is building. System‑level platform designs include centralized rectifiers, high‑efficiency DC busways, rack‑level DC‑DC converters, and energy‑storage integration to stabilize large, synchronous AI loads.

🗣️ Management voice. Scott Armul (EVP, Vertiv) says, “Larger AI workloads are reshaping every aspect of data center design,” adding Vertiv’s AC/DC systems know‑how positions it to meet unprecedented power demand. NVIDIA’s Dion Harris underscores the need for a “fundamental shift in power architectures” to unlock next‑gen AI infrastructure.

🏭 Designs under test. Vertiv says its reference and product designs are being validated on real projects and scaled against gigawatt‑scale campus requirements—an early indicator that customers are planning for multi‑GW “AI factories.”

🛡️ Service moat. Safety and serviceability are pivotal at 800 VDC. Vertiv highlights its 4,000+ field engineers and DC/AC service track record as a differentiation point for mission‑critical operations.

📈 Market reaction. Shares of VRT jumped intraday to a record high on the news; Reuters cited a move to $184.36 intraday and ~7% up on the day.

🧠 NVIDIA tie‑in & scale. NVIDIA’s OCP‑tied roadmap points to Kyber racks connecting 576 Rubin Ultra GPUs by 2027, with broad ecosystem support for 800 VDC infrastructure—context for why power vendors like Vertiv are aligning now.

Why AI/Semiconductor Investors Should Care

Power has become the gating factor for AI compute. If 800 VDC really trims conversion losses and copper usage while enabling ≥1 MW racks, it lifts usable power density and floor‑space efficiency—key drivers of $/TFLOP capex and opex. That creates a larger TAM for Vertiv across rectification, busway, DC‑DC, energy storage, controls, and long‑tail services, and ties the company tightly to NVIDIA’s rack‑scale roadmap. Risks: execution and safety/standards at 800 VDC, a multi‑year product ramp (H2’26), and competitive responses from other power OEMs. But if NVIDIA’s claimed efficiency/TCO gains prove out at scale, 800 VDC becomes foundational to AI factory economics—and Vertiv’s early readiness positions it to capture outsized share of that transition.

Subscribe now

OpenAI (Private), Broadcom (NASDAQ: AVGO)
🚀 OpenAI’s Three‑Way Chip Play: From Buyer to Builder

What The Chip: On October 13, 2025, OpenAI capped a three‑deal run to secure ~26 gigawatts (GW) of AI compute by 2029: 10 GW of custom inference silicon with Broadcom (announced today), 10 GW of NVIDIA systems backed by a $100 billion NVIDIA investment (Sept 22), and 6 GW of AMD GPUs tied to an equity warrant (Oct 6). The strategy shifts OpenAI from a pure software shop to a vertically integrated AI infrastructure operator.

Details:

🔧 Broadcom = custom chips + Ethernet at scale. OpenAI will co‑design and deploy 10 GW of OpenAI‑designed accelerators with Broadcom’s end‑to‑end Ethernet stack, with first racks arriving 2H26 and full rollout by end‑2029. As CEO Sam Altman put it, “Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential.” Broadcom CEO Hock Tan called it “a pivotal moment” as the companies co‑develop and deploy 10 GW of next‑gen accelerators and networks. OpenAI also disclosed “over 800 million weekly active users.”

🌐 Why Broadcom’s network matters. The systems lean on Broadcom’s Tomahawk switches and Jericho fabric to scale standards‑based Ethernet clusters far beyond a single GPU box. Jericho‑class silicon has been positioned to stitch hundreds of thousands to ~1 million accelerators across sites; Tomahawk Ultra/6 push radix and bandwidth for rack/cluster scale‑out. By contrast, NVIDIA’s NVLink domain tops out at 72 GPUs for all‑to‑all coupling—great for tight training, less ideal for internet‑scale inference sprawl.

🟩 NVIDIA = capital + compute. Under a letter of intent, OpenAI will deploy at least 10 GW of NVIDIA systems, starting 2H26 on the Vera Rubin platform; NVIDIA intends to invest up to $100 billion in OpenAI as each GW comes online. Jensen Huang called it “the biggest AI infrastructure project in history.” Importantly, OpenAI buys directly from NVIDIA for the first time (not via a cloud intermediary), giving OpenAI more control over cost and timing.

🟥 AMD = equity‑for‑access. OpenAI will deploy 6 GW of AMD Instinct GPUs over multiple generations, beginning with MI450 in 2H26. AMD granted OpenAI a warrant for up to 160 million shares (~10%), vesting on deployment, stock‑price, and technical milestones—directly aligning OpenAI’s financial upside with AMD’s success. AMD CFO Jean Hu framed the deal as “tens of billions of dollars in revenue” and accretive to EPS. Context: AMD’s MI300X has been powering Azure OpenAI Service since 2024, with Microsoft citing leading price/performance on GPT inference.

📊 Economics: enormous scale, real trade‑offs. Reuters pegs total all‑in data‑center costs at $50–$60 billion per GW; Broadcom’s custom route aims to lower unit inference cost versus off‑the‑shelf GPUs. At the same time, critics flag a “circular” AI economy—NVIDIA invests so OpenAI can buy NVIDIA; AMD gives OpenAI equity so OpenAI buys AMD—stoking bubble questions. Meanwhile, OpenAI disclosed 700 million weekly users in Sept (NVIDIA LOI) and 800 million today, but press reports show ~$5 billion 2024 losses on ~$4 billion revenue and a 2025 revenue outlook of ~$11.6–$12.7 billion. Execution must close the cash‑flow gap.

⚠️ Policy headwinds. New tariff threats (e.g., proposed 100% U.S. tariffs on China goods) and evolving AI chip export controls add supply‑chain and pricing risk for components (optics, substrates, HBM). These could raise TCO or delay deployments across 2026–2029.

Why AI/Semiconductor Investors Should Care

OpenAI just created a three‑lane moat around compute. If it works, OpenAI improves unit economics (custom chips + vendor leverage), reduces single‑supplier risk, and keeps pace with Big Tech peers building in‑house silicon. For investors:

• NVDA gains demand certainty and equity exposure; the direct‑buy model should deepen wallet share even as it funds a key customer.

• AMD gets a flagship anchor customer and strategic alignment via the warrant—powerful, but it raises execution risk (deliver 6 GW on time, scale software and networking).

• AVGO monetizes both custom accelerators and AI Ethernet/optics, riding the secular pivot from proprietary fabrics to open, huge‑scale networks.

The bear case: circular financing and huge cash burn could unravel if AI monetization lags or policy shocks hit supply chains; first‑gen custom silicon often needs iterations. The bull case: 800 M weekly users, enterprise adoption, and lower inference cost curves compound into sustainable margins by the time H2‑2026 deployments come online. Today’s moves don’t just buy GPUs—they buy optionality across the full compute stack.

Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Get 15% OFF FISCAL.AI — ALL CHARTS ARE FROM FISCAL.AI —

Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.

The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!

[Paid Subscribers] Aehr Test Systems Starts Fiscal 2026: AI Buzz Builds, Margins Take a Hit

Date of Event: October 6, 2025

Executive Summary

*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences

When you think of the AI gold rush, chipmakers like NVIDIA and AMD probably spring to mind first. But behind the scenes, companies like Aehr Test Systems (NASDAQ: AEHR) play a crucial role, supplying equipment to stress-test semiconductors, ensuring they survive the intensity of modern AI workloads. Aehr kicked off fiscal 2026 (ending August 29, 2025) with some mixed financial results but plenty of optimism for future growth.

Financial Snapshot

Revenue landed at $11.0 million, a 16% drop from last year’s $13.1 million. On a GAAP basis, Aehr posted a net loss of $2.1 million, or $0.07 per diluted share—a swing from the $0.7 million profit ($0.02 per share) in the same period last year. Non-GAAP earnings (which exclude things like stock-based compensation and restructuring costs) were just above breakeven at $0.2 million ($0.01 per share), significantly lower than last year’s $2.2 million ($0.07 per share).

On a brighter note, bookings hit $11.4 million, nudging past quarterly revenue and growing the backlog to $15.5 million. Adding early Q2 orders pushes this backlog even further to $17.5 million. Aehr closed the quarter with a solid cash cushion of $24.7 million, carrying no debt.

CEO Gayn Erickson put it plainly, “We finished ahead of street consensus for both revenue and our bottom line.” Translation: it wasn’t a perfect quarter, but better than Wall Street anticipated.

Riding the AI Wave

Aehr’s real story is the buzz building around its AI-driven products. The company’s new Sonoma systems, designed for packaged-part burn-in (think of this as quality control on steroids), continue to attract big players in the AI space. One “world-leading hyperscaler” (likely a major cloud giant) placed multiple follow-on orders, speeding up deliveries to handle growing demand for their own cutting-edge AI processors.

Read more