PJFP.com

Pursuit of Joy, Fulfillment, and Purpose

Tag: Nemotron

  • Why the Markets Are Pricing AI Wrong: Gavin Baker on the July 2026 Selloff, GPU Spot Prices, Memory LTAs, and Nvidia’s Credit Wrapper

    Gavin Baker of Atreides Management returned to Invest Like the Best with Patrick O’Shaughnessy days after one of the strangest months the AI trade has ever produced. AI and semiconductor names fell 40 to 60 percent in a straight line while, by Baker’s account, not a single quantitative metric on the ground deteriorated. He spent the week in Silicon Valley hunting for a bearish data point and came back with almost nothing except credit. This conversation is the result: a detailed argument that the market has mispriced the gap between contracted compute and spot compute, that open source is growing the infrastructure pie rather than shrinking it, and that the one risk actually worth fearing is political rather than financial.

    TLDW

    Gavin Baker describes July 2026 as “2022 packed into a single month,” a violent AI and semiconductor drawdown that happened while hyperscaler operating cash flow accelerated from roughly 28 percent growth to 32 percent, or closer to 35 percent adjusting for unusual legal charges. His core claim is that the installed base of GPU compute is locked into long-term contracts priced far below the current spot market, so as those contracts roll off, compute reprices higher, operating cash flow accelerates, and the buildout can be funded internally rather than with the debt that widening credit default swap spreads and a poorly received Meta bond have made look expensive. He walks through each catalyst of the selloff: Meta renting out compute (misread as a capex cut), the open source capability leap from GLM 5.2 and Kimi K3 (misread as deflationary when a token is a token and costs the same flops, watts, and memory to produce), China acquiring a domestic deep ultraviolet lithography machine (real but 25 years behind), and rising real yields (the only genuine negative). He covers the game theory of breaking a memory long-term agreement in a world where market share is set by supply allocations, Nvidia’s new credit wrapper plus revenue share model and why it is misunderstood, the router and fine-tuning stack from Fireworks and Baseten that turns “ChatGPT wrappers” into defensible AI natives, continual learning as the one technical development that could disrupt training demand, SRAM accelerators for disaggregated inference, SpaceX as an underappreciated compute company with orbital ambitions, and his view that regulation, not fundamentals, is the biggest risk because the industry has done a terrible job telling its own story. He also makes an unusual observation about market structure: everyone now feeds news into Claude, and Claude has become a kind of Walter Cronkite for the stock market, collapsing the diversity of interpretation that normally keeps markets stable.

    Thoughts

    The load-bearing claim in this episode is the spread between contracted and spot compute, and to Baker’s credit it is falsifiable in a way most bull cases are not. He is not arguing that AI will be transformative or that demand feels strong. He is arguing something narrow and checkable: hyperscalers and neoclouds signed multi-year GPU contracts in 2024 and 2025 at prices that assumed a gentle decline, prices instead went vertical, and the installed base is therefore systematically under-earning. A startup rented several thousand B200s in the mid two dollars per GPU hour range and expects to pay just under four dollars for an identical cluster seven months later. If that repricing is real and broad, hyperscaler operating cash flow mechanically accelerates and roughly 700 billion dollars of projected credit demand evaporates. If GPU rental prices roll over and stay down for two consecutive quarters, the thesis is dead. That is the number to watch rather than any earnings headline. The caveat he steps past quickly is that the open source mix shift he describes as bullish does not eliminate margin, it relocates it, out of the frontier labs and down into the infrastructure layer. Excellent if you sell GPUs, power, and memory. Considerably more awkward for the labs whose projected cash flows are the reason anyone believes the compute gets paid for at all.

    The Claude as Walter Cronkite observation deserves more attention than it got, where it passed as a joke. Baker is describing a genuine change in market microstructure. Every institutional and retail participant now feeds the same news into roughly the same models, and while those models are probabilistic, they are not producing meaningfully diverse readings of the same headline. He connects this to Michael Mauboussin’s argument that a breakdown in diversity, not leverage alone, is what produces bubbles and crashes. If that is what happened in July, then the Japanese capacitor stock chart he cites, an entire three-year cycle compressed into six weeks before the fundamentals had even arrived, is not a curiosity. It is the signature of a market where thousands of participants share one interpretive engine. That makes drawdowns faster and deeper without making them more informative, which argues for holding through machine-generated narrative cascades rather than trading them.

    The middle of the conversation contains the most consequential business idea in it, and it is one that got almost no coverage during the selloff: memory long-term agreements and Nvidia’s credit wrapper are the same move executed at two different layers of the stack. Both trade near-term upside for durability. The memory companies stopped maximizing spot price and started signing prepaid agreements with floors and ceilings, and the reason those agreements will hold is that the penalty for breaking one has changed category. Apple could renege on memory pricing for years because its volume was overwhelming and it had no equivalent competitor. In a world with four buyers that matter and where AI market share is set by supply allocation rather than product quality, a supplier can answer a broken price agreement by breaking the volume commitment and handing your allocation to a rival, in an industry where oversupply is always followed by undersupply. Nvidia is running the same play one layer up. The credit wrapper with a revenue share above a price floor converts a cyclical one-time chip sale into a royalty on recurring compute revenue, financed on someone else’s balance sheet, which is a materially better business than selling hardware. It also widens the moat, because a startup accelerator pays more at the foundry, pays more for high bandwidth memory, and cannot finance its chips at Nvidia’s rate. Baker is right that this is misunderstood, and it is a strange thing for a stock at a ten-year-low forward multiple to be quietly doing.

    The technical material in the back half reveals an asymmetry worth naming. Baker treats two efficiency developments very differently. Continual learning and sample efficient learning, which several labs believe are close, would collapse the token budget required to produce a capable model, and he handles this by asserting that training asymptotes to a small but nonzero share of compute and that the outcome would be wonderful for the world anyway. SRAM-based accelerators for disaggregated inference, running prefill on one chip, attention on a high-memory chip, and the feed forward network on SRAM, he embraces enthusiastically as a return-on-investment improvement across the installed base. Both are efficiency gains. One is treated as neutral, the other as clearly positive, and Jevons paradox is doing all the work in both directions. That is probably correct given everything we have observed so far, but it is an assumption rather than a finding, and it is the assumption on which the entire “cheaper compute is bullish for compute” framework rests. Worth noting too that the SRAM disaggregation point is genuinely underdiscussed: those chips sit on older nodes and do not compete for leading-edge capacity, so they are additive supply rather than substitute supply.

    The final twenty minutes hold both the largest unpriced upside and the largest unpriced risk, and neither is in consensus estimates. On the upside, only the hyperscalers, CoreWeave, Crusoe, and SpaceX have ever brought more than 500 megawatts online in a single year, and SpaceX has done it fastest and cheapest. When it dumped a large block of compute into the market, the market absorbed it without a blip, which tells you more about demand than any survey. Baker’s sanity check on orbital compute is the sharpest reasoning move in the episode: Benchmark, from entirely outside the Elon ecosystem and without the benefit of internal launch costs, funded StarCloud at a real valuation, so the set of people who would all have to be wrong keeps growing. On the downside, regulation is the risk he names first and it is the one his own framework cannot arbitrage. New York’s data center moratorium is not a fundamentals problem, and no amount of operating cash flow acceleration fixes a permitting ban. His diagnosis is that the industry finds the benefits so obvious that it never learned to explain them, which is how a water usage figure overstated by four orders of magnitude became conventional wisdom. Proposing a foundation that buys World Series ad time is a tell about how far behind he thinks the industry is. Every other risk in this conversation is priced somewhere. That one is not.

    Key Takeaways

    • Baker characterizes July 2026 as “2022 in a month,” with AI names down 40 to 60 percent from their highs in a straight line while underlying fundamentals improved.
    • He spent the week in Silicon Valley explicitly hunting for a negative quantitative metric and found essentially one: third-party data suggesting Anthropic’s growth curve came slightly off trajectory, a data point Anthropic shareholders reportedly dispute.
    • Nvidia was trading at its lowest forward price to earnings multiple in ten years at the time of recording. The only cheaper moments were the DeepSeek shock and Liberation Day, both of which proved to be V-bottoms.
    • A low forward multiple means the market believes these companies are significantly over-earning. Baker’s counter is that they are under-earning because their installed compute is contracted below spot.
    • Combined operating cash flow at Microsoft, Meta, and Amazon accelerated from roughly 28 percent to 32 percent growth, or to about 35 percent after adjusting for an unusual quarter of legal and regulatory charges.
    • Nobody in 2024 or 2025 modeled old GPU prices going vertical in 2026. The bull case assumed a slow decline in rental rates and the bear case assumed a steep one.
    • A concrete example: a well-known startup rented several thousand Blackwell B200s in the mid two dollars per GPU hour range and expects to pay just under four dollars for an identical cluster seven months later, a 50 to 60 percent increase.
    • One inference cloud stated publicly that it plans to pay roughly 100 percent more for Blackwells when its current contract expires.
    • Neoclouds were often forced into below-market long-term contracts because they needed an offtake agreement to finance the GPUs in the first place.
    • Consensus models hyperscalers monetizing Blackwell and Rubin at roughly Ampere rates, two generations behind, producing about 1.3 to 1.4 trillion dollars of hyperscale operating cash flow. Assuming monetization merely at a discount to current Blackwell rates pushes that closer to two trillion and removes roughly 700 billion dollars of credit demand.
    • The credit concerns are real and undeniable: real yields are up, spreads have widened, credit default swap levels for the large buyers have blown out, and a recent Meta bond did not price where a Meta bond should price.
    • Baker’s response is that debt-fueled buildouts demand immediate repayment and unwind violently, which is what happened in the internet buildout, but this buildout is still overwhelmingly funded from operating cash flow.
    • If credit is not available, he argues the existing flops simply become more valuable, which is self-correcting rather than catastrophic.
    • The Meta selloff catalyst was a misread. Meta renting out compute was interpreted as excess capacity and a capex cut. Meta did not cut capex, and the actual motivation appears to have been demonstrating strong internal rates of return on a small slice of capacity ahead of a capital raise.
    • The open source panic was also a misread. Open source taking token share moves margin dollars out of the frontier model layer, but a token still requires the same flops, memory, and watts to produce, so infrastructure demand rises rather than falls.
    • Frontier tokens carry gross margins somewhere in the 80 to 95 percent range. Open source tokens might carry 30 percent. The customer’s savings come almost entirely out of that margin, not out of compute consumption.
    • Baker calls open source “dark matter to the public markets,” growing rapidly through GLM 5.2, Kimi K3, and Nvidia’s Nemotron, but nearly impossible for public investors to measure since it runs through private inference clouds.
    • Jensen Huang being the world’s loudest supporter of open source is itself evidence that open source is good for Nvidia’s business.
    • Enterprises that blow through their AI budget in three months set up a router, which cuts their spend but often increases total GPU hours consumed by shifting volume to cheaper open source tokens.
    • Adoption is happening in staggered waves: AI natives are all in and hiring very few humans, coastal public companies are optimizing, East Coast and non-coastal companies have barely adopted, and Europe is trying to regulate AI before using it.
    • Roughly 500,000 people worldwide use agentic AI, and perhaps half that number use it seriously, yet the world is already in an acute compute shortage. The relevant question is what happens at 100 million or 500 million users.
    • Token spend at the most AI-forward companies now runs 20 to 25 percent of total compensation spend, with individual examples at 30 percent and reports as high as 50 percent, against a roughly 25 trillion dollar global knowledge work market.
    • Founder-controlled companies are not conducting large-scale layoffs, which suggests the cash flow to pay for AI is expected to come from growth rather than from labor substitution.
    • Memory is the dominant variable in token economics. More memory per unit of compute yields more tokens out, which lowers cost per token, which is why demand has shown no negative elasticity to memory pricing.
    • Memory suppliers have shifted from maximizing near-term price to signing long-term agreements with prepayments, floors, and ceilings, trading short-term upside for durability.
    • Breaking a memory long-term agreement is now potentially fatal. With four buyers that matter at scale and market share determined by supply allocation, a supplier can respond by breaking the volume commitment and handing your allocation to a competitor.
    • This is structurally different from the Apple era, when a single dominant buyer could break pricing agreements without consequence.
    • Nvidia’s new model is best described as a credit wrapper with a revenue share triggered when GPU prices exceed a floor. It is not vendor financing, since a third party lends the money, and it could produce a very large cloud-scale royalty business quickly.
    • Baker thinks this model is badly misunderstood, meaningfully increases Nvidia’s revenue per gigawatt, and strengthens its competitive position against startup accelerators that pay more at the foundry, pay more for high bandwidth memory, and cannot finance their chips as cheaply.
    • Nvidia has taken equity stakes across the ecosystem, and Baker’s read is that every time they have not taken a stake it has proven to be a mistake.
    • The scenario that would genuinely frighten him: hyperscaler operating cash flow stops accelerating, forcing the buildout onto debt, or a sustained sharp contraction in GPU rental prices. Nobody he has spoken to says they have too many GPUs.
    • Continual learning and sample efficient learning are the technical developments most likely to disrupt training demand, and several new labs including Safe Superintelligence are focused on them. Baker still thinks training asymptotes to a small share of compute rather than to zero, and that the change would be enormously good for the world regardless.
    • Fireworks launched a product called Nexus that plugs into Claude Code, OpenAI Codex, or Grok in roughly three lines of code, ingests a customer’s data, applies reinforcement learning to a model, and routes queries appropriately.
    • This stack is what converts an alleged “ChatGPT wrapper” into a defensible company. Shifting 30 to 60 percent of token consumption to a customized open model on top of frontier orchestration produces better outcomes at roughly half the cost.
    • Cheap, capable open source models may actually inflate the value of the very best frontier model, since a 160 IQ orchestrator becomes more valuable when it has an army of cheap 120 IQ models to direct.
    • The inference clouds are growing almost as fast as the frontier labs did in their early days while burning very little cash, which is extraordinary by any conventional software metric.
    • China obtaining a domestic deep ultraviolet lithography machine is a genuine phase transition and should not be dismissed, but the technology is roughly 25 years behind extreme ultraviolet, and lithography progress is learning by doing that cannot be teleported through.
    • Baker considers regulation the biggest single risk to AI, citing New York’s data center moratorium as the first of many and describing the current environment as post-factual and post-logical.
    • The public narrative that data centers raise power bills, drain water, and destroy jobs is largely wrong. Behind the meter deals typically lower local electricity prices, and modern community agreements include hospitals, schools, police and fire stations.
    • The widely cited data center water figure originated in a published error overstating usage by roughly 10,000 times, since acknowledged by the author, which Baker likens to the decimal point error that created the myth that spinach is exceptionally high in iron.
    • He argues data centers are among the best things to happen to blue collar wages in his lifetime, with ongoing rather than one-time employment from maintenance, replacement, and upgrade cycles.
    • SRAM-based accelerators built on older nodes and free of high bandwidth memory constraints could substantially improve return on investment by allowing disaggregated inference: prefill on one chip, attention on a high-memory chip, and the feed forward network on SRAM.
    • SpaceX has improved fundamentally since going public, and Baker believes the market does not yet understand it as a compute company. Only the hyperscalers, CoreWeave, Crusoe, and SpaceX have ever brought on more than 500 megawatts of power in a single year, and SpaceX has done it fastest and cheapest.
    • A widely circulated report claims SpaceX intends to bring on eight gigawatts of compute in 18 months. Baker doubts the number but notes that at roughly 50 billion dollars of monetization per gigawatt, even a fraction of it dwarfs the current consensus estimate.
    • When SpaceX dumped a large block of compute into the market, it was absorbed without a blip, which Baker reads as one of the more bullish demand signals of the year.
    • Orbital compute feels more real every day. Benchmark funding StarCloud, from outside the Elon ecosystem and without access to internal launch costs, functions as a useful sanity check on the idea.
    • Dark horse names Baker flags for the next phase: Lip-Bu Tan, Lin Qiao at Fireworks, and Scott Wu at Cognition.

    Detailed Summary

    A Selloff That Contradicted Every Fundamental

    Baker opens by describing July 2026 as 2022 compressed into a single month. AI names fell 40 to 60 percent from their highs in a nearly straight line. What made the month unusual was not the magnitude but the absence of a legible cause. In 2022 the market feared recession, rising rates, and inflation. During the DeepSeek shock and Liberation Day you knew exactly what the market was reacting to. This time the fundamentals moved in the opposite direction from the tape. GPU availability tightened, GPU rental pricing rose, DRAM spot prices rose, and token growth accelerated. Baker asked Patrick, who had also spent the summer in Silicon Valley, whether he had heard a single negative quantitative metric or a single instance of deceleration. The answer was nothing.

    Part of the problem is visibility. Public markets cannot see Anthropic or OpenAI directly, and they cannot see the American open source inference clouds like Fireworks, Baseten, Modal, and Together that monetize inference. Everyone stares at the same chart of semiconductor cash flow rising while hyperscaler free cash flow falls, and that chart omits the private companies entirely. It also omits the repricing dynamic Baker considers the most important fact in the market.

    The Spot Versus Contract Gap

    In 2024 and 2025 every serious forecast assumed GPU rental prices would decline, with the only debate being how fast. Neoclouds locked in long-term contracts partly out of prudence and partly because they needed offtake agreements to finance the hardware at all. The result is a large installed base of contracted compute trading at a steep discount to today’s spot market. Baker’s argument is that as those contracts roll off, compute reprices higher even if spot itself declines from current levels, and that repricing flows directly into hyperscaler operating cash flow.

    The anecdotes are stark. A prominent startup rented several thousand B200s in the mid two dollar per GPU hour range and expects to pay just under four dollars for an identical cluster seven months later. One inference cloud said publicly it plans to pay roughly double for Blackwells at contract renewal. Baker’s read is that hyperscalers are therefore under-earning across the board, which is the exact opposite of what a ten-year-low forward multiple implies the market believes.

    Financing the Buildout and the Credit Question

    Credit is the one bearish input Baker concedes is real. Real yields have risen, spreads have widened, credit default swap levels have blown out across the large buyers, and a recent Meta bond did not price the way a Meta bond should. Sophisticated private capital investors told him this is just banks hedging commitments, but he acknowledges the optics are bad and the facts are undeniable. His concern is the classic capital cycle: debt-financed buildouts demand immediate repayment, so when supply and demand slip out of alignment the unwind is fast and brutal, exactly as it was in the internet buildout.

    The math he ran is the counterweight. Consensus effectively models hyperscalers monetizing Blackwell and Rubin at Ampere rates, two generations behind, producing 1.3 to 1.4 trillion dollars of operating cash flow. Assume instead that they monetize merely at a modest discount to current Blackwell rates and the figure approaches two trillion, taking about 700 billion dollars of credit demand off the table. Better cash flow also improves the credit ratios, which makes debt cheaper if they choose to use it. And if credit disappears entirely, the flops already installed simply become more valuable. Microsoft brought on a large slug of capacity in June that did not even appear in second quarter results.

    How the Month Actually Unfolded

    Baker walks the sequence of catalysts. First, Meta announced it would rent out compute, which the market read as excess capacity and an imminent capex cut. Meta did not cut capex. What Meta appears to have seen was SpaceX selling trading-optimized clusters into the market at an enormous premium to contracted rates, and the plan was likely to demonstrate strong returns on a small slice of capacity before raising equity capital and increasing capex. Shortly afterward Meta released its best model in a long time, overshadowed by a competing release but a clear signal it was not easing off.

    Next came the open source freakout. Kimi K3 arrived, the widely watched token index dipped and flattened, and the two were connected: the index captures mix, and a shift from expensive frontier tokens toward open source tokens looks like weakness even when total compute consumption is rising. Then China’s deep ultraviolet lithography news triggered a broad selloff in semicap equipment. Finally, rising real yields and widening spreads gave the market a genuine reason to worry. Baker’s summary is that with the sole exception of credit, every one of these narratives was factually wrong, and a friend at Fidelity described the winning strategy of the past three years as doing the dumbest, most superficial thing as fast as possible and cycling between them.

    Open Source as Dark Matter

    The most important conceptual argument in the episode is that a token is a token. Regardless of which model produces it, a token consumes the same flops, the same memory, and the same watts. Open source taking share therefore does not reduce compute demand. It transfers margin from the frontier model layer, where gross margins might be 90 percent, to open weights inference at perhaps 30 percent, and the resulting price decline drives elasticity in token volume. Since frontier labs and open source models both run on the same underlying cloud infrastructure at the same compute cost, the effect is to push margin dollars down into the infrastructure layer.

    Baker calls open source dark matter to public markets. It is real, it is accelerating on the back of capability leaps from GLM 5.2 and Kimi K3, Nvidia continues to push Nemotron closer to the frontier, and yet none of it appears in audited financials that public investors can underwrite. He also notes the tell that should have settled the debate: Jensen Huang is the world’s most vocal supporter of open source, which would be an odd position for the largest beneficiary of frontier concentration to hold if open source actually threatened the business. Baker adds a normative point, that a world with only one or two dominant frontier models charging 90 percent margins is not good for humanity, and that many models is the better outcome.

    Routers, Fine-Tuning, and the End of the Wrapper Insult

    The practical mechanism behind the open source surge is the router plus fine-tuning stack. Inference clouds have become genuinely good at supervised fine-tuning and reinforcement learning, so a company can take its proprietary data, customize an open weights model, put it behind a router, and have the router send most queries to that model while escalating to a frontier model for verification or harder work. The result is often slightly better outcomes at half the cost. Fireworks shipped a product called Nexus that connects to Claude Code, OpenAI Codex, or Grok in roughly three lines of code and handles ingestion, reinforcement learning, and routing.

    This changes the durability question for AI natives. Two years ago the criticism was that these companies were thin wrappers with no defensibility. Now a company with domain-specific proprietary data can train on it, own the model serving 30 to 60 percent of its tokens, and get off the frontier lab treadmill it previously had no choice but to accept. Baker points to Cursor, Harvey, and others leaning hard into this. He also raises the counterargument fairly: some believe that once a frontier model achieves recursive self-improvement it will serve every intelligence level more cheaply through distillation, leaving no room for open source. He does not dismiss it, but he thinks the proprietary data held by AI natives and the orchestration value of the single smartest model make the multi-model future more likely. Cheap 120 IQ models arguably make a 160 IQ orchestrator more valuable, not less.

    Where the Money Comes From

    The pushback Baker gets on X is fair: even if hyperscalers are under-earning, where does the customer revenue ultimately come from? Definitionally it must come from faster economic growth through productivity or from labor substitution. He sees labor substitution happening at AI natives, though not through firing. They simply never hire the humans, and gross profit dollars per full-time employee at these companies is vertical compared with prior startup generations. Token spend now runs 20 to 25 percent of total compensation spend at the most aggressive companies, with individual examples at 30 percent and reports as high as 50 percent, against a roughly 25 trillion dollar global knowledge work market.

    The encouraging signal is that founder-controlled companies, the ones most likely to move fast on efficiency, are not conducting large-scale layoffs once you adjust for pandemic-era overhiring. That suggests they see continued opportunity for people plus large token budgets rather than a straight substitution. Data from Cognition, Ramp, and Stripe indicates that companies spending the most on AI are growing meaningfully faster, though Baker acknowledges the skeptics’ point that these datasets do not control for industry.

    The Memory Supply War and LTA Game Theory

    Everything is currently in shortage, and Baker argues the constraint is energizing gigawatts rather than manufacturing. Turbine makers and diesel generator makers are ramping, old aircraft turbines are being stripped and reconditioned for data center power, and regulatory policy is moving favorably. The transition he says he got wrong is the shift, especially in memory, from maximizing short-term pricing to signing long-term agreements with customer prepayments, price floors, and price ceilings.

    The reason those agreements will hold is game theory. Memory is the axis around which everything else revolves, because more memory per unit of compute means more tokens out, which lowers cost per token, which is why demand has shown essentially no negative elasticity. Market share among the four buyers that matter (Amazon with Trainium, Google with TPUs, AMD, and an Nvidia bigger than all of them combined) will be determined for years by supply chain allocation. Break a long-term agreement to chase a lower price in an oversupply year and the supplier can break the volume commitment in return and hand your allocation to a competitor. Since oversupply in this industry is reliably followed by undersupply, that is a decision that can end a franchise. Apple could get away with this historically because its volume was overwhelming and it had no equivalent competitor. That world is gone.

    Nvidia’s New Playbook

    Baker finds Nvidia’s low multiple hard to reconcile with how thoroughly the current environment favors it. If chips need to be financed, nothing on earth is more financeable than an Nvidia GPU. If land and power are the constraint, Nvidia has been playing the matchmaking chess game well. On top of that they have rolled out what Baker describes as a credit wrapper with a revenue share that kicks in when GPU prices sit above a floor. It is not vendor financing, since someone else lends the buyer the money. What it does is give Nvidia a royalty on recurring compute revenue, which could amount to a very large cloud business built entirely out of royalties, while helping bridge the cash flow mismatch between an industry that has gone free cash flow negative and a supplier collecting all the cash.

    Asked what he would do as a memory CEO, Baker says he would do exactly what Nvidia is doing: approach GPU and accelerator buyers, participate in the credit wrapper, perhaps put up cash upfront to make lenders comfortable, and take a cut of ongoing revenue. He expects firms like Blackstone and Apollo are pitching variants of this to the memory companies already. He also thinks the arrangement quietly widens Nvidia’s competitive moat, since startup accelerator companies pay more at the foundry, pay more for high bandwidth memory, and cannot finance their chips at Nvidia’s rate. And he notes that essentially every time Nvidia has declined to take an equity stake in something, it has turned out to be a mistake.

    What Could Break the Thesis

    Pressed for the scenario that would flip him, Baker names two. The first is operating cash flow failing to accelerate, which would force the buildout onto debt and validate the credit bears. That outcome depends largely on whether the combined trajectory of Anthropic, OpenAI, Grok, Cursor, and open source keeps compounding. The second is a sustained sharp contraction in GPU rental prices. The market would react instantly, and it would mean the compute shortage had broken. As of the recording, not a single person he has spoken with says they have too many GPUs.

    The technical wildcard is continual learning and sample efficient learning. Many researchers believe both are close. A human learns effectively on something like 20 billion tokens while frontier models train on 300 trillion, so a model that could be trained on 10 trillion tokens and then learn efficiently in the world would represent a discontinuity in training demand. Baker thinks training will asymptote to a small but nonzero share of compute regardless, and that the development would be extraordinarily good for the world. He also notes Nvidia is deeply involved with essentially all of the labs pursuing it.

    China, Lithography, and Decoupling

    On China’s deep ultraviolet lithography machine, Baker holds both views at once. It is a genuine phase transition, comparable to going from having no propeller plane to having one, because they did not have it before and now allegedly they do. It is also roughly 25 years behind extreme ultraviolet, and lithography is learning by doing, so you cannot teleport through the required cycles. He suspects the market overreacted and that if it ever affects ASML’s order book it will be years out, by which time the market will have forgotten and rediscovered the concern several times.

    He is careful about certainty here. It is very hard for an American to have real clarity on what is happening inside China, the people there are extremely capable and work brutally hard, and they consider this existential for the country. There are unverified reports that an extreme ultraviolet machine was smuggled in, which he treats as noise. His larger point is that decoupling is now self-reinforcing on both sides, it is unfortunate, and neither side is going to stop.

    Regulation, Data Centers, and a Failure of Storytelling

    Asked for the worst thing that could happen to AI, Baker answers regulation without hesitation. New York’s data center moratorium feels like the first of many, and even deep red pro-growth states are telling the industry it is doing a poor job explaining itself. The political narrative among ordinary Americans is that data centers will raise electricity prices, drain water supplies, and eliminate jobs. Baker’s counter is that behind the meter deals generally lower local electricity prices, that community agreements now routinely include hospitals, schools, police stations, and fire stations rather than the old model of buying the fire department new trucks, and that the jobs are ongoing rather than one-time because of continuous maintenance, replacement, and upgrade cycles.

    The water claim is the clearest case of a myth outrunning the correction. An author overstated data center water usage by roughly 10,000 times, has acknowledged the error repeatedly, and the figure still circulates. Patrick offers the parallel of the spinach iron myth, created by a misplaced decimal point in an academic text and still believed 80 years later. Baker’s proposed remedy is blunt: a foundation or political action committee running ads during the Final Four, NFL games, and the World Series explaining what a data center actually does for a community, alongside the story of AI accelerating medical research and improving outcomes for people with serious illness. The people building this find the benefits so obvious that they assume everyone already knows, and they cannot process how divergent their view is from most Americans.

    SRAM Accelerators and Disaggregated Inference

    An underdiscussed development, Baker argues, is what happens when SRAM-based accelerators arrive at scale. These chips are not constrained by high bandwidth memory and are often built on older nodes, so they do not compete for the leading edge capacity that GPUs consume. Inference disaggregates into prefill and decode, and decode splits further into attention and the feed forward network. The holy grail is running prefill on a chip without high bandwidth memory, attention on a high-memory chip, and the feed forward network on SRAM, which nothing beats for that workload. Since workloads keep changing, no single chip can get the ratio of compute to high bandwidth memory to on-die SRAM permanently right, which is precisely the argument for disaggregation. Baker expects this to be strongly positive for the return on investment across the installed base and on new compute.

    SpaceX, Orbital Compute, and Dark Horses

    Baker does not think the market understands SpaceX as a company yet, and he considers it the most important new public company. The fundamentals have improved since the IPO, and the compute story is the part being missed. Only the hyperscalers, CoreWeave, Crusoe, and SpaceX have ever brought more than 500 megawatts of power online in a single year, and SpaceX has done it fastest and cheapest while building clusters customers actually like. When SpaceX dumped a large block of compute into the market, it was absorbed without a blip, which Baker treats as one of the most bullish demand datapoints available. A circulating Substack report claims eight gigawatts within 18 months. He doubts that figure and quotes it only because it is public, but at roughly 50 billion dollars of monetization per gigawatt against a 73 billion dollar consensus estimate, even partial delivery would overwhelm expectations. There is a well-known New York hedge fund short case built on spot compute prices falling 90 percent.

    On orbital compute, Baker says time at Starbase left him thinking it feels more real every day, and the Starship landing reinforced it. His sanity check is that Benchmark, from entirely outside the Elon ecosystem and without the benefit of internal launch costs, chose to fund StarCloud at a real valuation, with SpaceX partnering to provide the Starlink laser technology that orbital compute requires. As he puts it, maybe he is crazy, maybe Elon is crazy, maybe Benchmark is crazy, and maybe the SpaceX engineers are crazy too, but all of that being true simultaneously does not seem probable. Asked for dark horses who could become as consequential as the current giants, he names Lip-Bu Tan, Lin Qiao at Fireworks, and Scott Wu at Cognition. The episode was recorded at Benchmark’s offices, at the table where their dinners are held.

    Notable Quotes

    “I want to be scared. I don’t want to feel like a lunatic watching these stocks get cheaper thinking the expected forward returns are going up.”

    Gavin Baker, on why he spent the week in Silicon Valley hunting for bearish data

    “I would describe July as 2022 in a month.”

    Gavin Baker, characterizing a 40 to 60 percent drawdown in AI names that happened in a straight line

    “Have you heard a single negative quantitative metric about AI? A single instance of deceleration?”

    Gavin Baker to Patrick O’Shaughnessy, framing the central contradiction of the month

    “A token is a token, and you need the exact same amount of compute to make a token. It takes the same amount of flops, the same amount of memory, the same amount of watts.”

    Gavin Baker, on why the open source panic misread infrastructure demand

    “Open source is kind of dark matter to the public markets. It’s hard for public markets to measure it.”

    Gavin Baker, on why the fastest-growing part of inference demand is invisible in audited financials

    “Claude is kind of Walter Cronkite for the stock market and everybody just believes whatever it says. And by the way, it’s really smart, but it’s not always right.”

    Gavin Baker, on the collapse of interpretive diversity among investors

    “Nvidia is actually, as we record this, at its lowest forward PE of the last 10 years.”

    Gavin Baker, noting the only cheaper moments were the DeepSeek shock and Liberation Day, both V-bottoms

    “If you break your LTA and then in the next two or three years for any reason leverage shifts back to the memory guys, you’re out of business.”

    Gavin Baker, on why long-term agreements will hold through the next memory cycle

    “If you need to be able to finance the chips, and you do, nothing’s more financeable than an Nvidia GPU. Nothing.”

    Gavin Baker, on why the current environment favors Nvidia more than its multiple suggests

    “Data centers are in a lot of ways the best thing to happen for blue collar wages in my lifetime.”

    Gavin Baker, on the gap between the political narrative and the local economics

    “A lie could go around the world faster than truth gets out of bed.”

    Gavin Baker, on a data center water usage figure overstated by roughly 10,000 times that still circulates

    “One of Elon’s phrases is we specialize in making the impossible late.”

    Gavin Baker, on why he doubts the eight gigawatt figure without betting against SpaceX

    Watch the full conversation here: Why the Markets Are Pricing AI Wrong with Gavin Baker on Invest Like the Best.

    Related Reading

    • Invest Like the Best on Colossus the show’s home, where the full episode archive and transcripts live.
    • Atreides Management Gavin Baker’s firm and the vantage point behind these compute and semiconductor calls.
    • More Than You Know by Michael Mauboussin, the source of the diversity breakdown framework Baker invokes to explain why markets crash when everyone reasons the same way.
    • High Bandwidth Memory (Wikipedia) background on the memory technology that sits at the center of the long-term agreement game theory.
    • Fireworks AI the inference cloud whose routing and fine-tuning stack Baker credits with making open source models competitive for production workloads.

  • Jensen Huang Says the AI Apocalypse Is ‘Complete Nonsense’: NVIDIA’s CEO on AI Jobs, China, Open Source Models, the AI Bubble, and the Trillion-Agent Future (Axios Behind the Curtain)

    Sitting on the floor of a brand new chip factory in Fort Worth, Texas, NVIDIA CEO Jensen Huang gave Axios reporter Mike Allen one of his most combative and quotable interviews yet. In this episode of Behind the Curtain, the head of the world’s most valuable company dismisses AI doom scenarios as “complete nonsense,” argues that AI is creating jobs rather than destroying them, defends Chinese open source models like Kimi and DeepSeek, explains why the AI build out is not a bubble yet, and calls for Anthropic’s most powerful model to be made available to everyone.

    TLDW

    Huang covers the full sweep of the AI moment: Chinese export control threats and why he wants open research flows in both directions, why the world needs both closed models (Anthropic, OpenAI) and open models (Kimi, Qwen, DeepSeek, NVIDIA’s own Nemotron), why Wall Street misread the Kimi selloff exactly as it misread DeepSeek, the sovereign AI argument that no company or country should “outsource its alpha,” his evidence that AI is increasing jobs for radiologists, paralegals, and manufacturing workers, a sustained attack on AI doomers and the “made up” narratives of singularity, simulation, and machine consciousness, the CapEx-heavy economics of manufacturing intelligence via tokens, his claim that the bubble is not coming in the next five years because physical constraints (chips, memory, power, construction workers) are pacing the build out, his warm relationship with President Trump and his warning against knee-jerk regulation, his position that Claude Mythos should be available to all users, the coming era of a trillion AI agents, the “ChatGPT moment” for robots having already arrived, and closing life lessons on pain, suffering, practice, immigration, and why he refuses to wear a watch because “now is the most important time.”

    Thoughts

    The first thing to hold in mind while watching this: every single position Huang takes, without exception, maps to selling more GPUs. Open models are good (more diffusion, more compute). Closed models are also good (more services, more compute). Chinese models are good (more use, more compute). Doom talk is bad (fear slows adoption, which slows compute). The bubble is far away (keep buying compute). That perfect alignment between worldview and order book does not make him wrong, but it means his arguments deserve scrutiny on the merits rather than deference to his position. He is the most effective anti-doomer in the industry partly because he is the person with the most to lose if the world gets scared.

    That said, his strongest material is empirical, and it lands. The radiologist example is a direct rebuttal to one of the most famous predictions in AI history, Geoffrey Hinton’s 2016 claim that we should stop training radiologists. Huang’s version of events, that automating the scan-reading task let radiologists see more patients and demand for them grew, is a textbook case of what economists call the Jevons effect applied to labor. Whether his specific numbers (20 percent more radiologists, 10 percent more paralegals, 50 percent more manufacturing jobs) survive fact-checking, the structural argument that automating a task can grow the profession around it is historically well supported, and it is the single most useful reframe in the interview: your job is not your task, and when the task gets automated, the purpose remains.

    The open source security argument is the most intellectually serious part of the conversation and the one most directly aimed at his own customers. Huang praises Anthropic and OpenAI as businesses in one breath and then dismantles the “closed models are safer” position in the next: Linux runs the world’s digital infrastructure precisely because millions of people can inspect and harden it, and a world defended by one closed model is a world with a single point of failure. His call for “massively distributed, diverse defense” via open models in the hands of cybersecurity experts everywhere is a real policy position with real stakes, and it puts him closer to Meta’s historical stance than to the labs he supplies.

    The bubble section is where the skeptic should lean in. Allen hands him the most famous cursed phrase in financial history, “this time is different,” and Huang takes the bait enthusiastically: it is different, he says, because the demand is industrial rather than cyclical. Every bubble in history was justified by exactly this argument, including the railroads and the dot-com fiber build out that Huang implicitly invokes as precedent. But his supply-side observation deserves weight: bubbles pop when supply overshoots demand, and right now everything (chips, memory, packaging, power, land, construction labor) is short. A market that cannot build fast enough is at least not overbuilt yet. His own concession that “the bubble will come someday” and his refusal to vouch for years five through ten is more honest than the rest of the answer.

    Finally, notice the tension he never resolves. He says warnings about AI’s power are “well heeded,” that safety is the leaders’ responsibility, and that Anthropic must fix jailbreaks fast. He also says consciousness, singularity, and existential risk are “all made up,” and shrugs off the referenced Mythos jailbreak with “everything was fine, you and I are here having a conversation.” Those two postures, take the technology seriously enough to harden it but never seriously enough to fear it, are held together mostly by confidence. It is a bet that capability and controllability scale together. The doomers he mocks are making the opposite bet, and nothing in this interview actually settles which one is right.

    Key Takeaways

    • On reports that Chinese regulators may tighten export controls on AI models and semiconductors to keep them from the West: Huang hopes it does not happen, notes half the world’s AI researchers are Chinese, and says both sides should de-escalate and let the technology advance.
    • He opposes any US ban on Chinese models like Kimi: American companies should absolutely be allowed to use them, because downloaded open models can be fine-tuned, guardrailed, and run inside secure sandboxes and harnesses, and the “back door” fear is a misconception.
    • The world needs both closed and open models: use closed services (Anthropic, OpenAI) as much as possible because they are excellent and convenient, but science, cybersecurity, and sovereignty require open models.
    • Regulate applications of AI (medicine, transportation, autonomous vehicles), not the underlying technology, which is dual use and should advance as fast as possible.
    • NVIDIA’s China sales are “approximately zero today” and he has told investors to expect none; he would consider it an honor to return if both governments allow it.
    • The market misunderstood DeepSeek and is now misunderstanding Kimi the same way: great open models, wherever they come from, drive more AI use, which drives more NVIDIA computers, more data centers, and more services.
    • Open models are not adversarial to closed models: the most likely customer to upgrade to Anthropic or OpenAI is someone who already uses AI and wants it more convenient and better.
    • NVIDIA’s Nemotron open model exists for companies that must build their own AI for sovereignty, regulatory, privacy, or IP reasons. “We don’t have to be the frontier. We have to be at the frontier.”
    • The large language model is the brain; a harness (he names OpenClaw and Claude Code as examples) turns it into a working agent. With the right harness, Nemotron can be world-class for specific skills.
    • Cheap or free open source tokens are “fantastic” for the proprietary labs: free AI grows the population of people who realize they need AI, and running even a free model yourself usually costs more than renting a service.
    • Echoing the viral Palantir CEO interview: “Nobody should outsource their alpha.” Companies and countries should rent AI wherever they can but must build their own AI for domain-specific, proprietary, sovereign, secret, or regulated work.
    • For non-differentiating work (marketing automation, legal department productivity), outsource to the frontier labs as much as possible.
    • Nothing AI has done has truly surprised him; what society needs to realize is that automating tasks is increasing the number of jobs the world needs.
    • His jobs evidence: radiologists up roughly 20 percent because AI-automated scan reading lets them see far more patients; paralegals up roughly 10 percent for the same reason; US manufacturing jobs up roughly 50 percent in recent years because AI data centers require industrial might.
    • On the demonstrated ability of Anthropic’s Mythos to break into hardened systems: “it surprised me that people were surprised.” An AI that can write and debug software can necessarily find vulnerabilities; the same capability powers cyber defense.
    • His security architecture argument: one single model is one single point of attack and failure. Open models in the hands of cybersecurity experts worldwide create “massively distributed, diverse defense,” the same reason Linux is trustworthy.
    • Whether China has “caught up” does not matter: the race-with-a-finish-line framing is wrong, China manufactures more AI researchers than the rest of the world combined, holding China back is ill-conceived, and neither side can hold back the other.
    • “AI is not going to destroy all of our jobs. Someone who uses AI is going to take our jobs.” The biggest risk to the US is scaring industries and society out of adopting AI.
    • On doomer AI CEOs: warning is fine, warning with a solution is better, and making things up is “absolutely inappropriate.” End-of-humanity and half-of-jobs-destroyed claims are “complete nonsense” contradicted by all the evidence.
    • Asked why Asia loves him while America is anxious: “the doomers spend too much time theorizing about these science fiction outcomes, maybe it makes them sound smart.”
    • OpenAI and Anthropic are not in trouble from Chinese competition: “zero possibility” China runs US companies off the road, both labs are thriving, and their IPOs will be the most successful in human history.
    • On chip stocks down 18 percent after Kimi dropped: free AI is great for hardware, chips, and data centers; the market got it wrong with DeepSeek (NVIDIA fell about 30 percent) and is getting it wrong again.
    • AI cannot have peaked because diffusion into society and industry has barely begun; useful AI has finally arrived, and useful AI is profitable AI, citing coding agents companies happily pay hundreds of millions a year for.
    • The new IT industry is CapEx heavier than software because intelligence must be manufactured: machines produce the tokens behind every answer, image, protein, and robot maneuver, and the resulting productivity will more than pay for the build out.
    • A token is an embedding of knowledge and intelligence, and unlike pi it gets smarter over time; smarter tokens are more valuable, which is why token economics keep improving.
    • On the bubble: “The bubble will come someday. It’s just not today.” Very unlikely in the next five years; five to ten years depends on how fast the industry can build.
    • The build out is constrained in every direction (chips, memory, land, power, construction workers), and that constraint is healthy: it pushes out the day supply exceeds demand.
    • This cycle is “industrial-driven,” not seasonal or consumer-demand-driven: the world needs a new intelligence infrastructure layer on top of energy, internet, roads, and railroads, and the semiconductor industry needs to be 5 to 10 times larger within ten years.
    • He is not worried about customers issuing hundreds of billions in debt to buy his chips: these companies generate enormous cash, the compute platform shift is real, and the ROI question has been answered because AI is now demonstrably profitable.
    • He would use Kimi himself, with fine-tuning, guardrails, sandboxing, and access control, the same way the world already trusts open source software like Linux.
    • On Trump: they text, the president “remembers everything” including H20, H200, Blackwell, and Rubin, and the Fort Worth factory they are sitting in is a direct result of their first conversation about reindustrializing America.
    • His warning to the administration: do not over-correct based on science fiction narratives about AI consciousness; talk to many CEOs and scientists, not one or two, and take time to be informed before regulating.
    • On the government taking an equity stake in NVIDIA: unnecessary, because the US already has a stake via $10 billion in taxes paid last year, job creation, and the stock market holdings of most Americans.
    • Claude Mythos should “absolutely be available to everyone,” not just selected institutions; it is Anthropic’s job to harden it and patch jailbreaks fast, and he notes that when it was jailbroken “everything was fine.”
    • On distillation of closed models: learning from other intelligence is fundamental (soon the internet will be 99 percent AI-generated content anyway), but violating terms of service or privacy is not okay and should be handled through existing legal channels.
    • NVIDIA has 6,500 employee families in Israel he is concerned for; he remains bullish on the UAE reinventing itself from an oil economy into an AI hub.
    • NVIDIA runs about 50,000 employees and may reach only 75,000 in ten years, “as small as possible,” because strategy means maximizing impact per unit of resource.
    • Jobs that are a single task (customer service call centers) will be automated; jobs with purpose survive because purpose does not change when the task is automated. “Don’t mistake your task for the job.”
    • In 10 to 20 years, photos of people typing at keyboards will look like old photos of typing pools with IBM Selectrics: typing was never the job, solving problems and creating value was.
    • The ChatGPT moment for robots has already arrived (a robot can reason through “put the apple in the drawer,” including opening the drawer first); useful robots in ordinary life within 3 to 4 years would not surprise him.
    • The agentic era’s capability has arrived and diffusion is next: the future holds 100 billion to a trillion agents running constantly, and agents will not become computers, they will use computers, which is why compute demand explodes.
    • $300 billion has been invested into US venture capital startups in the last six months, and he tells his nieces and nephews that great fortunes will be created on a laptop.
    • Life lessons: greatness requires “plenty of pain and suffering” and practice when nobody is watching; under maximum stress, time slows down the way athletes describe, and that comes from repetition.
    • He advises every bright mind in the world to come to America, the country built by immigrants that will need amazing immigrants in the future.
    • He wears no watch and refuses to let Outlook manage his life: “now is the most important time.” His perfect Saturday: dogs, work, family dinner, a cocktail, and he notes every weekend is exactly like that.

    Detailed Summary

    Export Controls Cut Both Ways

    The interview opens on a Financial Times report that Chinese regulators are considering export controls of their own, restricting Chinese AI models and semiconductors from reaching the West. Huang’s response is de-escalation in both directions: half the world’s AI researchers are Chinese, groundbreaking research flows from both countries, and once one side reaches for export controls, everyone starts thinking in those terms. He is confident the US will continue to lead as long as government supports rather than constrains its companies. Asked whether the US should ban Chinese models like Kimi, he rejects the premise: downloaded open models run inside harnesses and sandboxes with security, privacy, and access controls, and the idea of hidden back doors phoning home to China is a misconception. His China sales, he notes pointedly, are approximately zero today, so his position is not about protecting revenue he does not have.

    Open and Closed Models Both Win

    Huang’s framework is consistent: rent closed models (Anthropic, OpenAI, which he personally uses along with Perplexity) whenever you can because they are excellent and convenient, and build on open models only when you must, for sovereignty, regulation, privacy, or proprietary domain reasons. This is the pitch for NVIDIA’s own Nemotron open model family, which he positions not as a frontier competitor but as raw material for companies that need custom AI: “We don’t have to be the frontier. We have to be at the frontier.” He describes the modern stack in plain terms: the large language model is the brain, and a harness (he cites OpenClaw and Claude Code) turns it into a working agent. Open, cheap, and free models are on-ramps that grow the total population of AI users, which is why he insists the labs should not fear them: the person most likely to pay for Claude is someone already using AI who wants it better and easier.

    Kimi, DeepSeek, and Wall Street’s Repeated Mistake

    Chip stocks fell 18 percent in the month after Kimi dropped, echoing the roughly 30 percent NVIDIA drawdown when DeepSeek landed. Huang says the market got it wrong both times and for the same reason: free and open AI is great for hardware, because great models drive use, use drives data centers, and data centers drive chips. He runs through the models he considers extraordinary (Kimi 3, Qwen, Nemotron, GPT 5.6, Codex, Claude Code) and lands on his core claim about this moment: useful AI has finally arrived, and useful AI is profitable AI. Companies like NVIDIA happily pay hundreds of millions of dollars a year for coding agents doing high-value work, which funds more AI, which he describes as a flywheel that has now started.

    Don’t Outsource Your Alpha

    Allen raises the viral Palantir CEO warning about handing your intellectual property to frontier labs, noting Huang’s unique position as both a top customer and top supplier of those labs, including using their models for chip design. Huang agrees with the principle without hesitation: nobody, no company, no country should outsource its alpha or its intelligence. His dividing line is specificity: work that is domain-specific, proprietary, sovereign, secret, or regulated must be done in-house on your own models, while generic productivity work like marketing automation or legal department support should be outsourced to the labs as aggressively as possible. The same logic scales to nations, which he says cannot outsource their fundamental intelligence to a third party.

    The Jobs Evidence

    Asked what AI has done that scared or awed him, Huang says essentially nothing surprised him, including the demonstrated ability of Anthropic’s Mythos to penetrate hardened systems (“it surprised me that people were surprised,” since an AI that debugs software can obviously find vulnerabilities). What he wants the world to notice instead is the labor data. Radiology reading has been substantially automated, and the number of radiologists is up roughly 20 percent because they can now see the enormous backlog of patients. Paralegals are up roughly 10 percent by the same mechanism. Manufacturing jobs are up roughly 50 percent in recent years because AI data centers require industrial construction. His formulation of the real risk: AI will not take your job, someone who uses AI will, and the worst thing America could do is scare its own industries out of adopting the technology.

    Against the Doomers

    This is the section that gives the interview its title. Huang says warning people is fine, warning with a solution is better, and making things up is absolutely inappropriate. The end of humanity: complete nonsense. Half of American jobs destroyed: complete nonsense. The singularity, living in a simulation, machine consciousness: “all made ups,” fun science fiction he enjoys hearing from “many of those leaders and my friends,” but Hollywood, not ground truth. Asked why he is mobbed by fans in Asia while the American mood is hostile, he suggests the doomers theorize about science fiction outcomes because “maybe it makes them sound smart.” His prescription for the industry is to tell the factual story, that AI is creating millions of jobs, rather than a made-up narrative that frightens the public and, more dangerously in his view, frightens policymakers. His closest thing to a concession: the closest thing to true AI is R2-D2 and C-3PO, “and who doesn’t want R2-D2 and C-3PO?”

    CapEx, Tokens, and the Bubble Question

    Huang’s economic argument for the build out runs through the token. Unlike the CapEx-light software era, intelligence must be manufactured: machines generate the tokens behind every answer, every image, and eventually every protein, chemical, and robot movement. A token is an embedding of knowledge, and unlike a static number it gets smarter over time, which makes it more useful, more valuable, and worth paying more for. On the bubble, he does not deny one is possible: “The bubble will come someday. It’s just not today.” He rules it out for roughly five years and hedges on five to ten. His reasoning is that this cycle is industrial-driven rather than consumer-cyclical: the world is adding an intelligence layer on top of energy, internet, roads, and railroads, the semiconductor industry needs to be 5 to 10 times larger within a decade, and everything (chips, memory, optical interconnects, packaging, TSMC capacity, land, power, construction workers) is short. Those constraints pace the CapEx and push out the day supply overtakes demand. As for customers issuing hundreds of billions in debt to buy his chips, he says the companies are extraordinary cash generators and the ROI question has been settled by profitable coding agents.

    Trump, Washington, and the Over-Correction Risk

    Huang describes a genuinely warm relationship with President Trump: they text, the president remembers chip model numbers (H20, H200, Blackwell, and next-generation Rubin), and the Fort Worth factory hosting the interview traces directly to their first conversation about restoring American manufacturing. He praises Susie Wiles, Secretary Bessent, and Secretary Lutnick. But his message to the administration is a warning: signs point toward more restrictive AI policy, and he fears policymakers falling for science fiction narratives (consciousness, an imminent finish line in a US-China race) pushed partly by companies hoping regulation will advantage them. His advice: talk to many CEOs and scientists, not one or two, take time, and do not over-correct. He rejects the 100-meter-dash framing of the China race entirely, arguing the win is diffusion, not invention: America did not invent electricity or manufacturing, it applied them with more enthusiasm than anyone, and that is what made the country. Asked about the government taking equity stakes in AI companies, he calls it unnecessary: the US already holds a stake in NVIDIA through $10 billion in annual taxes, job creation, and the stock market.

    Mythos for Everyone, and the Distillation Question

    In the most newsworthy exchange, Allen asks whether the world is ready for Anthropic’s most powerful model, Claude Mythos, to be available to everyone rather than selected institutions. Huang’s answer is unambiguous: it should absolutely be available to everyone, it is Anthropic’s responsibility to harden it, and jailbreaks are the nature of software, to be patched as fast as they are found. He points to the referenced jailbreak incident and observes that “everything was fine,” while noting that holding Anthropic back serves no American interest, especially since open models are available regardless. On distillation, he splits the question: AIs learning from other AIs is fundamental and inevitable (within a few years, he predicts, the internet will be 99 percent AI-generated content, so every model is distilling other AIs anyway), but violating terms of service or privacy is not acceptable, and aggrieved providers should pursue the conventional legal remedies that already exist.

    Robots, Agents, and the Next Era

    Huang argues the ChatGPT moment for robots has already happened, on his definition: the 2022 ChatGPT moment was not when AI became useful (that took four more years) but when it did something surprising, and a robot that can reason through “put the apple in the drawer,” including opening the drawer first, clears that bar today. Useful everyday robots within three to four years would not surprise him. On the agentic era, capability has arrived and diffusion is what comes next: where perhaps 100 million humans use computers at any given moment today, the future holds 100 billion to a trillion agents of every kind running constantly. His line: agents are not going to become computers, agents are going to use computers, and that is the deepest driver of compute demand.

    Life Lessons from 33 Years at the Helm

    The closing stretch turns personal. On keeping NVIDIA at roughly 50,000 employees (maybe 75,000 in ten years, “as small as possible”) while peers run six figures, he says strategy is using limited resources with maximum precision, a craft he has practiced longer than any CEO in tech history: “this is my kung fu.” On which jobs disappear, he distinguishes task from job from purpose: call center tasks will be automated, but a radiologist’s purpose (ending human suffering) survives the automation of scan reading, and typing was never the job in the first place. Born in Taiwan and sent to a rough American boarding school at nine, he calls America the greatest country in the world because open discourse and freedom let it work through its disagreements, and he urges bright minds everywhere to come. On greatness: no athlete just happens to be great, it is practice when nobody is watching, setbacks, losing, and “plenty of pain and suffering” that elevate craft, character, and resilience. He wears no watch because now is the most important time, and his perfect Saturday (dogs, work, family dinner, a cocktail) is, he says, exactly what every weekend already looks like.

    Notable Quotes

    “And so the fact that this is going to be the end of humanity, it’s complete nonsense. The fact that this is going to destroy half of the American jobs. It’s complete nonsense. And all of the facts, all of the evidence point exactly to the opposite.”

    Jensen Huang, on AI doom predictions from fellow tech leaders

    “AI is not going to destroy all of our jobs. Someone who uses AI is going to take our jobs, and so we have to make sure that we adopt AI, diffuse AI into the industries as quickly as possible.”

    Jensen Huang, on the real employment risk of the AI era

    “Nobody should outsource their alpha. Nobody should outsource their intelligence. No country should.”

    Jensen Huang, agreeing with the Palantir CEO’s warning about handing IP to frontier labs

    “We don’t have to be the frontier. We have to be at the frontier.”

    Jensen Huang, on NVIDIA’s Nemotron open source model strategy

    “The bubble will come someday. It’s just not today.”

    Jensen Huang, on whether the AI build out is a bubble

    “It is made up that there’s going to be a singularity. It’s made up that somehow we’re living in a simulation. These are all made ups.”

    Jensen Huang, on science fiction narratives he says are scaring the public and policymakers

    “The closest thing to true AI is R2-D2 and C-3PO. And who doesn’t want R2-D2 and C-3PO?”

    Jensen Huang, on how to inoculate the public against fear of AI

    “These two companies will be the most successful IPOs in human history.”

    Jensen Huang, predicting the public debuts of OpenAI and Anthropic

    “If your job is the task, then it’s very likely that when that task is automated, your job will be eliminated or changed.”

    Jensen Huang, on which jobs disappear in an industrial revolution

    “Because now is the most important time. I refuse to let Outlook manage my life, and I refuse to let a watch manage my life.”

    Jensen Huang, on why he does not wear a watch

    Watch the full conversation between Jensen Huang and Mike Allen on Axios Behind the Curtain here.

    Related Reading

  • SubQ 1.1 Small Explained: How Subquadratic Sparse Attention Hits 98% Retrieval at 12 Million Tokens With 64.5x Less Compute Than Dense Attention

    Subquadratic, a frontier AI research and infrastructure company, has released the model card and technical report for SubQ 1.1 Small, a long-context language model built on a new attention mechanism the company calls Subquadratic Sparse Attention (SSA). The headline claim is unusual in two directions at once: the model retains 98% single-fact retrieval accuracy at 12 million tokens, roughly twelve times the length it was primarily trained on, while cutting attention compute by 64.5x against dense attention at a 1 million token context. The deeper argument in the report is not really about a single model at all. It is about what happens to the entire retrieval-and-orchestration stack once reasoning over a complete artifact stops being prohibitively expensive.

    TLDR

    SubQ 1.1 Small is a small long-context model that replaces the dense attention of an existing open-weight frontier model with Subquadratic Sparse Attention, a learned, content-dependent sparse attention mechanism that scales linearly in compute and memory rather than quadratically. On retrieval it posts 99.12% on NVIDIA’s 13-task RULER suite at 128K tokens and 100% needle-in-a-haystack accuracy at 1M and 2M tokens, holding at 98% out to 6M and 12M tokens while attending to only 0.13% of token pairs. It keeps competitive general ability, scoring 85.4% on GPQA Diamond and 89.7% pass@4 on LiveCodeBench v6, and reaches 13% on the long-horizon AutomationBench Finance agentic benchmark, close to Opus 4.8 and GPT-5.5 and well ahead of mid and small tiers. The efficiency story is a scaling win rather than a constant-factor one: 64.5x fewer attention FLOPs than dense attention at 1M tokens and 56x faster than FlashAttention-2 on a single attention layer. The report frames cheap long-context compute as a research accelerator that let the team run more than one hundred million-token experiments and find a training recipe (long-context continued pretraining is the strongest lever) rather than guess at one, positions SSA against FlashAttention, DeepSeek’s Lightning Indexer line, state space models like Mamba, and hybrids, invokes Sutton’s Bitter Lesson to argue that RAG, chunking, and agentic scaffolding are partly workarounds for context scarcity, and was independently verified by Appen. Deployment is starting with design partners now, with a 2M to 12M token lineup planned by year end.

    Thoughts

    The most interesting move in this report is the framing, not the benchmark. Subquadratic plants its flag on Richard Sutton’s Bitter Lesson and argues that much of the modern AI stack, the retrieval pipelines, the chunkers, the re-rankers, the agentic orchestration, is scaffolding built around a single computational constraint: dense attention costs grow with the square of context length. If that constraint relaxes, a lot of hand-engineered machinery that exists to feed a model the right fragments at the right moment starts to look like the task-specific pipelines that learned representations eventually displaced. That is a genuinely provocative thesis, and it is the right lens for reading the rest of the document. The company is not selling a longer context window as a feature. It is betting that whole-artifact reasoning is a different shape of capability than retrieval over fragments, and that fragmentation destroys the cross-references a contract or a codebase actually depends on before the model ever sees them.

    The part of the paper most teams will undervalue is the claim that the real payoff of efficient attention is not cheaper inference but cheaper experimentation. A dense long-context training campaign is expensive enough that most groups get a handful of attempts and are forced to guess at the recipe. Subquadratic says SSA let them run more than a hundred experiments across six model generations with per-step iteration under a minute at million-token context, which is how they discovered that long-context continued pretraining, not clever post-training, was the dominant lever. If that holds, algorithmic efficiency becomes a first-class scaling variable alongside parameters and data, because capability becomes responsive to iteration velocity rather than raw compute alone. It reframes efficiency from a deployment line item into a research multiplier, and that is a more durable advantage than any single benchmark number.

    The generalization result deserves scrutiny precisely because it is so clean. A model trained overwhelmingly at 1M tokens, with a sliver at 2M and nothing beyond, holds 98% retrieval at 12M. The proposed explanation is that SSA routes attention by content relevance rather than fixed positional pattern, so there may simply be no obvious length boundary once the routing behavior is learned. That is plausible and the report is careful to say the 12M result emerged rather than being designed for. But single-needle NIAH is a deliberately clean probe with one target and a binary answer. The far harder RULER suite is only reported at 128K, the longest standardized length in the original benchmark, so the multi-hop, aggregation, and distractor-heavy capability that whole-artifact reasoning actually requires has public numbers at 128K, not at 12M. The honest read is that precise retrieval generalizes spectacularly and composite reasoning at extreme length is still an open question the report does not over-claim on.

    What lends the report credibility is how much counter-evidence it volunteers. It walks through MiniMax abandoning its hybrid M1 architecture and returning to full attention for M2 after efficient variants showed multi-hop reasoning deficits at scale. It admits that earlier SubQ checkpoints improved retrieval while regressing on knowledge benchmarks, forcing dedicated capability-balancing work. It describes catching a case where the MRCR benchmark moved up while the model felt worse in real workflow spot-checks, and switching its development signal to RULER as a result. That last point is a quietly important methodological argument: benchmark score and deployment behavior diverged enough to change checkpoint selection, which is a warning every team shipping long-context models should internalize. A vendor confident enough to show where its own metrics misled it is more trustworthy than one that only shows the wins.

    A few caveats keep the enthusiasm grounded. AutomationBench Finance at 13% is genuinely strong relative to peers, but it is a low absolute score across the board, including for GPT-5.5 at 18% and Opus 4.8 at 16%, so this is early evidence of agentic transfer rather than proof of a finished agent. The efficiency comparisons isolate a single attention layer rather than full end-to-end model throughput, which is the right way to expose the scaling shape but not the same as a wall-clock serving benchmark. The model is built from an unnamed donor open-weight frontier model, so some of its general-knowledge and coding strength is inherited rather than created here. And the most aggressive claims about the future, a 2M to 12M lineup and much higher sparsity, are roadmap, not released artifacts. None of that undercuts the core result. It just means the right posture is to treat SubQ 1.1 Small as a strong proof of concept for an architecture that, if it scales as advertised, could quietly remove a layer of the AI stack that everyone currently takes for granted.

    Key Takeaways

    • SubQ 1.1 Small is a long-context language model from Subquadratic AI, built on a new attention mechanism called Subquadratic Sparse Attention (SSA), released June 16, 2026 alongside a model card and technical report.
    • SSA is a learned, content-dependent sparse attention mechanism that scales linearly in both compute and memory with sequence length, rather than quadratically like dense attention.
    • The central result is context-length generalization: the model was trained primarily at 1M tokens, with some training at 2M and none beyond, yet retrieval held far past the training window.
    • Needle-in-a-haystack accuracy is 100% at 1M and 2M tokens and 98% at both 6M and 12M tokens, roughly twelve times the primary training length.
    • At 12M tokens the model attends to only 0.13% of token pairs, close to a 1,000x reduction in attention relationships, while still retrieving accurately.
    • On NVIDIA’s 13-task RULER benchmark at 128K tokens, SubQ 1.1 Small scores 99.12%, with the remaining errors concentrated in aggregation-style tasks rather than retrieval.
    • RULER tests beyond single-fact lookup: single-key and multi-key retrieval, common-word and frequent-word extraction, and multi-hop variable tracing across positions.
    • At 1M tokens, SSA requires 64.5x fewer attention FLOPs than dense attention (3.9 PFLOP versus 252 PFLOP per attention layer).
    • On a single attention layer, SSA runs 56x faster than FlashAttention-2 at 1M tokens (966 ms versus 54,164 ms on an H100), reaching parity near 16K tokens and pulling away as context grows.
    • The efficiency gain is a scaling-law win, not a constant-factor speedup: the advantage over dense attention grows as context length increases.
    • On general knowledge, SubQ 1.1 Small scores 85.4% on GPQA Diamond (pass@1), below GPT-5.5 (93.2) and Opus 4.8 (92), near Sonnet 4.6 and GPT-5.4-mini (87.5), and above GPT-5.4-nano (81.7) and Haiku 4.5 (67.2).
    • On coding, it reaches 89.7% pass@4 on LiveCodeBench v6, close to the absolute frontier (GPT-5.5 92, Opus 4.8 92.2) and ahead of the smaller tiers.
    • On AutomationBench Finance, a long-horizon agentic benchmark, it scores 13%, close to Opus 4.8 (16%) and GPT-5.5 (18%) and ahead of Sonnet 4.6 (8%), Haiku 4.5 (3%), and GPT-5.4-mini (0%). Absolute scores are low across all models.
    • The model was not trained from scratch. The team converted an existing open-weight frontier model by replacing dense attention with SSA, then built long-context ability through staged context extension and continued pretraining.
    • Context was extended in stages (262K, 512K, 1M, 2M) using YaRN positional scaling, with long-context continued pretraining performed between extension stages on naturally long data: books, long documents, and repository-scale code.
    • Roughly one trillion tokens of continued pretraining were performed, most of it at the 1M-token stage.
    • Long-context continued pretraining was the most consistent predictor of long-context retrieval gains across the experiments, more so than post-training tweaks.
    • The team ran more than one hundred long-context experiments across six major model generations, which the report argues is only possible because SSA made million-token iteration cheap (under a minute per step).
    • Capability balance was a recurring challenge: gains in long-context retrieval often regressed short-context knowledge and reasoning unless training was explicitly managed for both.
    • Benchmark scores and real deployment behavior diverged. The MRCR benchmark moved up while qualitative workflow spot-checks got worse, so the team switched its primary development signal to RULER.
    • The report frames RAG, chunking, summarization, and agentic orchestration as scaffolding built around context scarcity, drawing an analogy to Sutton’s Bitter Lesson, where hand-engineered mechanisms get displaced by larger-scale learning.
    • SSA is positioned against FlashAttention (a memory optimization that does not change quadratic compute), fixed-pattern sparse attention, DeepSeek’s learned sparse line, state space models, and hybrid architectures.
    • DeepSeek’s Lightning Indexer (used in DSA and CSA) is the closest published comparison. Its quadratic scoring overtakes the sparse attention it feeds around 52,000 tokens, reaching roughly 16x the attention cost at 1M and 190x at 12M.
    • State space models like Mamba achieve linear cost through a compressed fixed-size state, but that compression is lossy and weakens exact retrieval, which is why production efficient models are usually hybrids with some dense attention layers retained.
    • MiniMax is cited as a cautionary case: it moved from a hybrid M1 to a full-attention M2 after hybrids showed multi-hop reasoning deficits at scale and less mature supporting infrastructure.
    • The benchmark results were independently verified by Appen, a third-party evaluation firm.
    • The named use cases are financial analysis and due diligence, legal and contract work, and software engineering (architecture-level reasoning, cross-file refactoring, dependency tracing, planning, review, and long-horizon memory).
    • Sparsity settings were deliberately conservative, tuned for maximum context length rather than maximum sparsity. Limited experiments at 4x the sparsity reported positive early results.
    • The training infrastructure used a memory-scaling ladder: single node, intra-node sequence parallelism, CPU offload, multi-node sequence parallelism, nested offloading, and Ring Attention for the longest contexts.
    • Beyond about 8M tokens, BF16 numerical underflow and stability became practical constraints on evaluation.
    • The technical report is authored by Saul Ramirez, Alex Whedon, Ashmal Vayani, and Phong Vo of Subquadratic AI.
    • Deployment is starting with a first cohort of design partners, with broader rollout through the quarter and a general model lineup ranging from 2M to 12M tokens by the end of the year.
    • The company’s framing line is “Efficiency is intelligence,” and its broader thesis is that the point is not bigger context windows for their own sake but reasoning directly over complete artifacts with less surrounding scaffolding.

    Detailed Summary

    The problem: whole-artifact reasoning and context scarcity

    The report opens by naming a class of tasks it calls whole-artifact reasoning: problems whose structure requires reasoning across a complete artifact rather than over isolated fragments. A legal agreement may define a term on page 2, qualify it on page 12, carve out an exception on page 46, and amend it in a schedule. A function may be defined in one file, called from forty others, and constrained by invariants encoded in the architecture rather than in comments. A financial review may require connecting filings, earnings reports, contracts, and internal records. In each case the difficulty is not locating a passage, it is reasoning over relationships distributed throughout a large artifact. Most production systems do not do this directly. They rely on retrieval pipelines, chunking, summaries, and agentic workflows that partition information and reconstruct fragments at inference time, because dense attention scales quadratically with context length and makes direct reasoning over large artifacts expensive. Subquadratic argues that much of the modern AI stack is therefore designed to manage context scarcity rather than reason over complete artifacts, and it connects this to Sutton’s Bitter Lesson: sophisticated hand-engineered mechanisms historically get displaced once larger-scale learning becomes practical.

    What SSA is and the three requirements it targets

    Subquadratic Sparse Attention is a content-dependent sparse attention mechanism designed to satisfy three requirements at once, a combination the report argues prior approaches never achieved in a practical long-context system. First, dense-attention-level retrieval and reasoning quality, which requires routing that is content-dependent (determined by the tokens themselves) rather than driven by a fixed positional pattern. Second, subquadratic scaling, where selection, retrieval, and attention are each linear in sequence length so the mechanism is linear end to end, not only within the attention read. Third, full-context training with standard autoregressive generation, so the model can optimize over the entire context during training while keeping efficient token-by-token decoding at inference. The internal mechanism by which SSA achieves this is held back as outside the scope of the report, which focuses instead on the requirements and the experimental program that followed.

    Where SSA sits among prior approaches

    The background section is effectively a taxonomy of long-context modeling. FlashAttention is treated not as a competitor but as the standard dense-attention baseline: it solved the memory problem by never materializing the full attention matrix, but it left the quadratic compute cost untouched, so doubling context still quadruples attention computation. Fixed-pattern sparse attention (sliding-window, strided, as in Longformer, BigBird, and the sliding window in Gemma) scales well but sacrifices content-dependent routing and tends to fail on retrieval benchmarks like RULER. Compression methods like Multi-head Latent Attention reduce KV-cache memory at inference but do not change the quadratic prefill cost. Learned sparse attention, exemplified by DeepSeek’s Native Sparse Attention and its Lightning Indexer, learns where to route but pays a quadratic cost in the indexer itself. State space models and linear attention (Mamba, Mamba-2 and Mamba-3, RetNet, RWKV, gated delta networks) achieve linear cost through a compressed fixed-size state, but that compression is lossy and weak on exact retrieval. Hybrids (Jamba, Kimi Linear, Qwen3 Next, Nemotron) keep a few dense layers to preserve retrieval, which means the quadratic component still dominates at long context. System-level workarounds (RAG, agentic frameworks, recursive language models) move retrieval outside the model entirely. The report’s stated open problem is to combine subquadratic scaling end to end with content-dependent retrieval, arbitrary-position access, and practical ultra-long-context training in one system, which it claims no widely deployed architecture provides and which SSA targets.

    Training: conversion, staged context extension, and continued pretraining

    Rather than training from scratch, the team converted an existing open-weight frontier model that supported a 262K-token context by replacing its dense attention with SSA. They then extended the context window in stages (262K to 512K to 1M to 2M) using YaRN to rescale positional representations, performing long-context continued pretraining between extension stages rather than jumping straight to the final length. The training mixture emphasized naturally long data such as books, long documents, and repository-scale code, packed to the target length with document separators and without masking cross-document attention boundaries. Most continued-pretraining tokens were trained at the 1M-token stage, with roughly one trillion tokens total. Post-training played a separate role: shaping how the long-context capability was expressed while preserving reasoning, coding, and instruction following. The team explored sample-level loss aggregation to keep a few extremely long examples from dominating gradient updates, and staged the post-training corpus across synthetic retrieval tasks, long-context reasoning, coding, educational material, and general instruction following, alternating capability-building phases with recovery phases.

    Results: retrieval, knowledge, coding, and agentic tasks

    On retrieval, SubQ 1.1 Small scores 99.12% on the 13-task RULER average at 128K, with errors concentrated in aggregation-style tasks like common-word and frequent-word extraction. On needle-in-a-haystack, evaluated on 50 held-out UUID samples per length, it scores 100% at 1M and 2M (within the training window) and 98% at 6M and 12M (held out), attending to only 0.13% of token pairs at 12M. On knowledge, GPQA Diamond pass@1 is 85.4%, landing between the small and mid frontier tiers and confirming that long-context optimization need not sacrifice reasoning, a result the report credits to its capability-balancing stages after earlier checkpoints showed retrieval gains coming at the cost of knowledge. On coding, LiveCodeBench v6 pass@4 is 89.7%, and the report notes coding data played a dual role, also improving non-code long-context retrieval because code is dense with the cross-position dependencies that train general routing. On long-horizon agentic work, AutomationBench Finance is 13%, where agents must discover the right endpoints among roughly 500 across 47 applications, make interdependent API calls, follow layered business rules, and ignore seeded distractors, graded on binary end-state correctness with no partial credit.

    Efficiency and the DeepSeek comparison

    Efficiency is measured on one attention layer against a dense baseline on the same backbone. Per-forward-pass attention FLOPs scale from a 2.1x reduction at 32K to 8x at 128K, 31.5x at 512K, and 64.5x at 1M tokens (3.9 PFLOP for SSA versus 252 PFLOP for dense). Measured against FlashAttention-2 in isolation, SSA reaches parity near 16K tokens and pulls away to 56x at 1M, where it runs in 966 ms versus 54,164 ms on an H100. The report devotes a discussion section to DeepSeek’s sparse attention line as the closest published comparison. DeepSeek’s Lightning Indexer is a learned selector, but it is a full-attention distilled transformer, so it scales quadratically: in a V3.2-style configuration the indexer is cheaper than the sparse attention it feeds only below about 52,000 tokens, then overtakes it, reaching roughly 16x the attention cost at 1M tokens and 190x at 12M. SSA targets that same selection role with a selector the report says is dramatically cheaper and linear throughout, and notes SSA could conceptually replace the selector over either uncompressed or compressed representations.

    Efficiency as a research accelerator and the evaluation lessons

    A recurring theme is that the most valuable effect of cheap long-context compute was on the research loop, not just inference. Where a dense campaign would allow a handful of attempts, SSA enabled more than a hundred experiments across six model generations with per-step iteration under a minute at million-token context. That throughput is what surfaced the finding that long-context continued pretraining is the strongest lever, and it leads the authors to argue that algorithmic efficiency should be treated as a first-class scaling variable alongside model and dataset size. The report is unusually candid about evaluation pitfalls. It describes how the MRCR benchmark diverged from deployment behavior, with MRCR-optimized checkpoints often feeling worse on repository-scale code reasoning, multi-document synthesis, and contract analysis, which pushed the team to rely on RULER and a fixed set of qualitative workflow spot-checks as development signals. It also cites MiniMax returning from a hybrid M1 to a full-attention M2 as evidence that reducing asymptotic cost is not sufficient on its own if retrieval quality, reasoning at scale, and system maturity are not preserved at the same time.

    Implications, availability, and what comes next

    The report’s deployment argument is that the most important enterprise implication of long-context models is not larger windows but the ability to reason directly over complete or more-complete artifacts, moving retrieval, re-ranking, and orchestration logic into the model where the task is naturally whole-artifact rather than naturally decomposable. It is careful not to declare retrieval obsolete: for corpora larger than any plausible context window, fast-changing knowledge, and genuinely multi-stage workflows, RAG and orchestration remain the right tools. The narrower claim is that the class of scaffolding that exists only to compensate for context limits gets smaller as efficient long-context models extend the reachable window. The benchmark results were independently verified by Appen. Subquadratic is deploying SubQ 1.1 Small with a first cohort of design partners now, with broader rollout through the quarter and a general lineup spanning 2M to 12M tokens planned by the end of the year, and it flags much higher sparsity as future work.

    Notable Quotes

    “Much of the modern AI stack is therefore designed to manage context scarcity rather than reason over complete artifacts directly.”

    SubQ-1.1-Small Technical Report, framing retrieval and orchestration as workarounds for an architectural limit

    “The hybrid has moved the line, but not changed its shape.”

    SubQ-1.1-Small Technical Report, on why hybrid models keep their quadratic component at long context

    “A routing mechanism intended to make long context affordable becomes the dominant long-context cost, reintroducing quadratic scaling after providing scalar compute savings.”

    SubQ-1.1-Small Technical Report, on DeepSeek’s Lightning Indexer overtaking the attention it feeds

    “If the cost of long-context experiments is too high, teams are forced to guess at the recipe. If the cost falls far enough, they can search for it.”

    SubQ-1.1-Small Technical Report, on efficient attention as a research accelerator

    “Fragmentation systematically destroys those relationships before the model ever sees them.”

    SubQ-1.1-Small Technical Report, on why chunking hurts whole-artifact reasoning

    “Holding the whole artifact in context changes the shape of the task rather than only the speed of it.”

    SubQ-1.1-Small Technical Report, on the difference between bigger windows and direct reasoning

    “The value of SSA is therefore not only that it makes long-context inference cheaper. It makes long-context experimentation cheaper.”

    SubQ-1.1-Small Technical Report, conclusion

    Read the full SubQ 1.1 Small technical report and model card here.

    Related Reading

    • Subquadratic (subq.ai) the company behind SubQ 1.1 Small and the Subquadratic Sparse Attention architecture, where you can join the waitlist.
    • The Bitter Lesson by Richard Sutton the short essay whose argument the report leans on, that hand-engineered mechanisms lose to general methods that scale with computation.
    • Attention Is All You Need the original Transformer paper that introduced the dense attention whose quadratic cost SSA is built to remove.
    • RULER (arXiv) NVIDIA’s long-context benchmark that the report uses as its primary retrieval signal, and that fixed-pattern sparse methods historically struggle with.
    • Retrieval-augmented generation (Wikipedia) background on the RAG approach that the report frames as scaffolding around context scarcity rather than a permanent fixture.

  • Jensen Huang at Stanford CS153 Frontier Systems on Co-Design, Agentic Computing, Vera Rubin, Open Models, and the Million-X Decade That Reshaped AI Infrastructure

    https://www.youtube.com/watch?v=tsQB0n0YV3k

    NVIDIA CEO Jensen Huang returned to Stanford for the CS153 Frontier Systems class (the room nicknamed itself “AI Coachella”) to lay out, in raw form, how he thinks about the computer being reinvented for the first time in over sixty years. Across roughly seventy minutes of student questions he walks through the codesign philosophy that gave NVIDIA a million-x decade, the architectural through-line from Hopper to Grace Blackwell to Vera Rubin to Feynman, the case for open source foundation models, the realities of tokens per watt and MFU, energy demand running a thousand times higher, the China and export-control debate, and his own biggest strategic mistakes. Watch the full conversation on YouTube.

    TLDW

    Huang argues every layer of computing has changed: the programming model, the system architecture, the deployment pattern, the economics. Co-design across CPUs, GPUs, networking, storage, switches and compilers gave NVIDIA roughly a million-x speed-up over ten years versus the ten-x Moore’s Law era, and that headroom is what let researchers say “just train on the whole internet.” Hopper was built for pre-training, Grace Blackwell NVLink72 for inference and reasoning (50x over Hopper in two years), Vera Rubin is built for agents that load long memory, call tools and need a low-latency single-threaded CPU bolted directly to the GPU, and Feynman extends that to swarms of agents that spawn sub-agents. Open weights matter because safety, sovereignty (230-plus languages no one else will fund) and domain models for biology, autonomy, robotics and climate need a foundation that NVIDIA is willing to seed. Compute is not really the scarce resource (Huang says place the order and the chips ship), the broken thing is institutional budgeting that can’t put a billion dollars into a shared university supercomputer. Energy demand is heading a thousand times higher and this is finally the moment market forces alone will fund sustainable generation. On geopolitics he rejects the GPUs-as-atomic-bombs framing and warns America will end up like its telecom industry if it cedes two thirds of the world. On career he advises seeking suffering on purpose. On strategy he says observe, reason from first principles, build a mental model, work backwards, minimize opportunity cost, maximize optionality.

    Key Takeaways

    • The computing model has been substantially unchanged since the IBM System 360, sixty-plus years ago. Huang’s first computer architecture book was the System 360 manual. AI is the first true reinvention.
    • Old computing was pre-recorded retrieval. New computing is generated, contextually aware and continuous. Cloud was on-demand. Agentic systems run continuously.
    • Codesign is NVIDIA’s central thesis. Inherited from the Hennessy and Patterson RISC era at Stanford, extended across CPUs, GPUs, networking, switches, storage, compilers and frameworks all optimized together.
    • The result of full-stack codesign: roughly 1,000,000x faster compute over ten years, versus a generous 10x to 100x for Moore’s Law in the same period. Dennard scaling effectively ended a decade ago.
    • That million-x speed-up is what unlocked “train on all of the internet” as a realistic AI strategy.
    • After GPT, Huang says it was obvious thinking was next. Reasoning is just generating tokens consumed internally, then using tools is generating tokens consumed externally. Agentic systems followed predictably.
    • Education needs AI baked into the curriculum, not just taught as a subject. Pre-recorded textbooks cannot keep pace with knowledge being generated in real time.
    • Huang says he cannot learn anymore without AI. He has the AI read the paper, then read every related paper, then become a dedicated researcher he can interrogate.
    • Mead and Conway and the first-principles methodology of semiconductor design are still worth learning even though most of the scaling tricks have been exhausted.
    • NVIDIA itself is one of the largest consumers of Anthropic and OpenAI tokens in the world. One hundred percent of NVIDIA engineers are now agentically supported. Huang recommends Claude and similar tools by name and says open-source downloads will not match the integrated product harness.
    • NVIDIA still invests heavily in open foundation models because language and intelligence represent the codification of human knowledge. Five pillars: Nemotron (language), BioNeMo (biology), Alphamayo (autonomous vehicles), Groot (humanoid robotics) and a climate science model (mesoscale multiphysics).
    • Sovereign language models matter. Roughly 230 world languages will never be a top priority for a commercial frontier lab. Nemotron is near-frontier and fully fine-tunable so any country can adapt it.
    • Safety and security require open weights. You cannot defend against or audit a black box. Transparent systems let researchers interrogate models and let defenders deploy swarms.
    • The future of cyber defense is not bigger-model-versus-bigger-model. It is trillions of cheap fast small models like Nemotron Nano surrounding the threat.
    • Domain models fuse language priors with world models. Alphamayo learned to drive safely on a few million miles instead of billions because it can reason like a human about the road.
    • MFU (Model Flops Utilization) is a misleading metric. Huang says he wants low MFU, because that means he over-provisioned every resource and never gets pinned by Amdahl’s law during a spike.
    • The xAI Memphis cluster running at 11 percent MFU is not necessarily a failure mode. In disaggregated prefill plus decode inference you can deliver very high tokens per watt with very low MFU.
    • The right metric is performance, ultimately tokens per watt as a proxy for intelligence per watt, and even that needs adjustment because not all tokens are equal. Coding tokens are worth more than other tokens.
    • Hopper was designed for pre-training. NVIDIA chose to build multi-billion-dollar systems when the largest existing scientific supercomputer cost $350 million, with no proven customer base. It worked.
    • Grace Blackwell NVLink72 was designed for inference, especially the high-memory-bandwidth decode phase. It is the world’s first rack-scale computer and delivered a 50x speed-up over Hopper in two years, against an expected 2x from Moore’s Law.
    • Vera Rubin is designed for agents. Long-term memory wired into storage and into the GPU fabric, working memory, heavy tool use, and Vera, a CPU optimized for low-latency multi-core single-threaded code so a multi-billion-dollar GPU system does not stall waiting on a slow tool call.
    • Feynman is being shaped for swarms of agents with sub-agents and sub-sub-agents, a recursive software topology that demands a new compute pattern.
    • Tokens per watt improved 50x in one generation. Compounding energy efficiency is the lever NVIDIA controls directly.
    • Total compute energy demand is heading roughly a thousand times higher than today, possibly two orders of magnitude beyond that. Huang says he would not be surprised if the estimate is low.
    • For the first time in history, market forces alone are enough to fund solar, nuclear and grid upgrades. Government subsidies are no longer required to make sustainable energy investment rational.
    • Copper interconnect is becoming a bottleneck. Photonics is moving from optional to structural inside racks and across them.
    • Comparing NVIDIA GPUs to atomic bombs, Huang says, is a stupid analogy. A billion people use NVIDIA GPUs. He advocates them to his family. He does not advocate atomic bombs to anyone.
    • If the United States cedes two thirds of the global market to competitors on policy grounds, the American technology industry will end up like American telecommunications, which was policied out of existence.
    • Huang directly rejects AI doom-by-singularity narratives. It is not true that we have no idea how these systems work. It is not true that the technology becomes infinitely powerful in a nanosecond. He calls the rhetoric irresponsible and harmful to the field students are about to enter.
    • On Stanford specifically: if the university president places an order, NVIDIA will deliver the chips. The bottleneck is that no university department has a billion-dollar compute budget because budgeting is fragmented across grants. Stanford’s $40 billion endowment is more than enough to fix that.
    • “It’s Stanford’s fault” is meant as empowerment. If something is your fault, you can solve it.
    • Career advice: do not optimize purely for passion. Most people do not yet know what they love. Pick the job in front of you and do it as well as possible. Even as CEO, Huang says, 90 percent of the work is hard and he suffers through it.
    • Suffering on purpose builds the muscle of resilience. When the company, the team or the family needs you to be tough, that muscle has to already exist.
    • NVIDIA’s first generation of products was technically wrong in nearly every dimension: curved surfaces instead of triangles, no Z-buffer, forward instead of inverse texture mapping, no floating point. The strategic recovery, not the technology, taught Huang the lessons that have lasted decades.
    • The biggest clean strategic mistake Huang names is the move into mobile chips (Tegra). It grew to a billion dollars then went to zero when Qualcomm’s modem dominance shut NVIDIA out of the 3G to 4G transition. The recovery into automotive and robotics (the Thor chip is the great great great grandson of that mobile lineage) was real, but Huang refuses to rationalize the original choice.
    • Forecasting framework: observe, reason from first principles, ask “so what” and “what next” until you have a mental model of the future, place your company inside that model, then work backwards while minimizing opportunity cost and maximizing optionality.
    • Best part of the CEO job: living at the intersection of vision, strategy and execution surrounded by people capable enough to make ambitious visions real. Worst part: the responsibility for everyone who joined the spaceship, especially in the near-death moments NVIDIA had four or five times early on.
    • Underrated insider note: Huang’s first apple pie with cheese, first hot fudge sandwich and first milkshake all happened at Denny’s. The Superbird, the fried chicken and a custom Superbird-style ham and cheese with tomato and mustard are his order.

    Detailed Summary

    Computing reinvented from the ground up

    Huang frames the moment as the first true rewrite of the computer in sixty-plus years. From the IBM System 360 forward, the mental model of writing code, running code, taking a computer to market and reasoning about applications stayed roughly constant. AI changes the programming model itself. Software is no longer a compiled binary running deterministically on a CPU. It is a neural network running on a GPU producing generated, contextual, real-time output. That cascades into how companies are organized, what tools developers use, what the network and storage stack look like, and what an application is even allowed to do. Robo-taxis, he notes, are an application no one would have attempted before deep learning unlocked perception.

    Codesign and the million-x decade

    Codesign is the philosophical center of the talk. Huang traces it to the RISC work of John Hennessy at Stanford, where simpler instruction sets won by being co-designed with the compiler rather than maximally optimized in isolation. NVIDIA extends the principle across every layer simultaneously: GPU architecture, CPU architecture, NVLink and NVSwitch fabrics, photonic interconnects, networking silicon, storage paths, CUDA libraries, frameworks and ultimately the model design. The numbers Huang gives are arresting. Moore’s Law in its prime delivered roughly 100x per decade. By the time Dennard scaling broke, real-world gains had compressed to roughly 10x. NVIDIA’s codesigned stack delivered between 100,000x and 1,000,000x over the same ten-year window. That non-linear speed-up is, in Huang’s telling, the precondition for modern AI: it is what allowed researchers to stop curating training sets and just feed the entire internet to the model.

    Education has to fuse first principles with AI tools

    Asked how curriculum should evolve, Huang argues AI must be integrated into the learning process, not just taught about. He recalls Hennessy writing his textbook by hand a chapter a week while Huang was a student, and says pre-recorded textbooks cannot keep up with the rate at which AI generates new knowledge. He describes his own learning workflow: hand the paper to an AI, then have it read the entire surrounding literature, then treat the AI as a dedicated researcher who can be interrogated. At the same time he defends the classics. Mead and Conway are still the foundation. Most modern semiconductor scaling tricks have been exhausted, but knowing where the field came from sharpens judgment when designing what comes next.

    Open source and the five domain pillars

    Huang gives one of the most detailed public accounts of why NVIDIA invests so heavily in open foundation models even while being a top customer of closed labs. He recommends Claude and OpenAI by name for production coding work, and says 100 percent of NVIDIA engineers are now agentically supported. The open-weights case rests on three legs. First, language is the codification of intelligence, and there are at least 230 languages that no commercial lab will ever prioritize. Nemotron is built near frontier and released so any country or community can fine-tune it. Second, the same representation-learning approach has to be replicated in domains where the data is not internet text, so NVIDIA seeded BioNeMo for biology, Alphamayo for autonomy, Groot for humanoid robotics and a climate model for mesoscale multiphysics. The economics of those fields would never produce a foundation model on their own. Third, safety and security require transparency. A black box cannot be defended or audited, and the future of cyber defense is not bigger-model-versus-bigger-model but swarms of cheap fast small models like Nemotron Nano surrounding the threat.

    MFU is the wrong metric, tokens per watt is closer

    A student raises the leaked memo that the xAI Memphis cluster is running at 11 percent Model Flops Utilization. Huang flips the framing. He says he would rather be at low MFU all the time, because that means he over-provisioned flops, memory bandwidth, memory capacity and network capacity. Bottlenecks shift constantly, so over-provisioning across every dimension is what lets the system absorb a spike without getting pinned by Amdahl’s law. In disaggregated inference, where prefill and decode are physically separated and decode is bandwidth-bound rather than flop-bound, NVLink72 can deliver extremely high tokens per watt while reporting very low MFU. Huang argues the right framing is performance, and ultimately tokens per watt as a rough proxy for intelligence per watt, adjusted for the fact that not all tokens are equal. A coding token is worth more than a generic token.

    Hopper, Grace Blackwell NVLink72, Vera Rubin, Feynman

    Huang gives the clearest public framing of NVIDIA’s roadmap as a sequence of architectural answers to evolving compute patterns. Hopper was built for pre-training, at a moment when NVIDIA chose to build multi-billion-dollar machines while the largest scientific supercomputer in the world cost $350 million and the marketplace for such systems was, on paper, zero. Grace Blackwell NVLink72 was the answer to inference and reasoning: a rack-scale computer that ganged 72 GPUs together because decode needs aggregate memory bandwidth far beyond a single chip. The generation-over-generation speed-up was 50x in two years, twenty-five times what Moore’s Law would have delivered. Vera Rubin is being built explicitly for agents. Agents load long-term memory from storage that has to be wired directly into the GPU fabric, they use working memory, they call tools that run on a CPU, and they wait. So the CPU has to be Vera, optimized for low-latency single-threaded code, because the multi-billion-dollar GPU system cannot afford to idle waiting on a slow tool call. Feynman extends the pattern to swarms of agents with sub-agents and sub-sub-agents, a recursive software topology that will demand its own compute pattern.

    Energy demand and the grid

    Huang’s energy projection is one of the most aggressive numbers in the talk. NVIDIA can compound tokens per watt by 50x per generation through codesign, but the total compute demand is heading roughly a thousand times higher, and Huang says he would not be surprised if the real figure is one or two orders of magnitude beyond that. The reason is structural: future computing is generative and continuous, not pre-recorded and on-demand. The good news, he argues, is that this is the best moment in the history of humanity to invest in sustainable generation. Market forces alone are now sufficient to fund solar, nuclear and grid upgrades. Government subsidies are no longer required to make the math work.

    Adversarial countries, export controls and the telecom warning

    This is the segment where Huang is visibly fired up. He attacks the GPUs-as-atomic-bombs framing on its face. NVIDIA GPUs power medical imaging, video games and soy sauce delivery. A billion people use them. He advocates them to his family. The analogy collapses at the first comparison. He attacks the second framing, that American companies should not compete abroad because they will lose anyway, as a self-fulfilling defeat. Competition makes the company better. The third framing, that depriving the rest of the world of general-purpose computing benefits the United States, also fails on first principles: it benefits one or two American companies at the cost of an entire industry. The cautionary parallel is telecommunications. The United States once had a leading position in telecom fundamental technology and policied itself out of it. Huang’s worry, voiced explicitly to a room of CS students, is that they will graduate into a shell of a computer industry if the same path is repeated.

    AI doom and rational optimism

    In the same arc Huang rejects the science-fiction framing of AI as a singularity that arrives suddenly on a Wednesday at 7pm and ends civilization. He calls those claims irresponsible, says they are not true, and points out that the people advancing them are believed by audiences who then make policy on that basis. It is not true that no one understands how these systems work. It is not true that intelligence becomes infinitely powerful instantaneously. It is not true that there is no defense. His framing, which the host echoes as “rational optimism,” is that the goal is to create a future where people care about computers because the technology students are learning is worth mastering.

    Stanford’s compute problem is Stanford’s fault

    A student presses on the scarcity of compute for independent researchers, startups and universities inside the United States. Huang’s answer is sharp: there is no shortage. Place the order and the chips will arrive. The actual broken thing is institutional. University grants are fragmented across departments. No researcher can raise enough on a single grant to fund a billion-dollar shared cluster, and no one shares. He compares it to showing up at the grocery store demanding a billion dollars of tomatoes today. The solution is planning, aggregation and a campus-scale supercomputer, the way Stanford once built the linear accelerator. The endowment is $40 billion. Pulling a billion off it, contracting cloud capacity and giving every student and researcher AI supercomputer access is, in Huang’s view, obviously doable. When he says “it is Stanford’s fault” the host laughs, but Huang clarifies: if it is your fault you have the power to fix it.

    Career, suffering and resilience

    Asked how a CS student should spend the next few years, Huang pushes back on the standard “follow your passion” advice. Most people do not know what they love yet, because no one knows what they do not know. The bar of demanding joy from every working day is too high. Whatever the job is, do it as well as you can. Even as CEO of NVIDIA he says he genuinely loves about 10 percent of his work. The other 90 percent is hard and he suffers through it. He recommends suffering on purpose, because resilience is a muscle that only builds under load, and when the company, the team or the family needs that muscle, it has to already exist. Earlier in his life that meant cleaning toilets and busing tables at Denny’s. He does it today running a multi-trillion-dollar company.

    The biggest mistakes

    Huang separates technical mistakes from strategic mistakes. NVIDIA’s first generation of products was technically wrong in almost every way: curved surfaces instead of triangles, no Z-buffer, forward instead of inverse texture mapping, no floating point inside. The company wasted two and a half years. But the strategic genius of the recovery, the reading of the market, the conservation of resources and the reapplication of talent, is what taught him strategy. The clean strategic mistake he names is mobile. NVIDIA’s Tegra line grew to a billion dollars of revenue and then collapsed to zero when Qualcomm’s modem dominance locked NVIDIA out of the 3G to 4G transition. Huang explicitly refuses the comforting rationalization that the Tegra effort fed the Thor automotive chip (“Thor is the great great great grandson”). The original decision, he says, was a waste of time. The lesson is to think one or two clicks further about whether a market is structurally winnable before committing the company.

    Forecasting under fog of war

    The final substantive exchange is on forecasting. Huang’s method has four steps. Observe what is actually happening (AlexNet crushing two decades of computer vision research in one shot, GPT producing reasoning by token generation). Reason from first principles about why it works. Ask “so what” and “what next” recursively until a mental model of the future emerges. Place the company inside that future and work backwards. Crucially, expect to be partly wrong. Some outcomes will absolutely happen, some will likely happen, some might happen, and the strategy has to be robust across that distribution. The real cost of any strategic choice is the opportunity cost of the alternatives you did not take, so the discipline is to minimize that cost and maximize optionality while letting the journey itself pay for the journey.

    Thoughts

    The most useful thing in this conversation is the explicit architectural mapping of compute patterns to chip generations. Hopper for pre-training. Grace Blackwell NVLink72 for inference, because decode is bandwidth-bound and a single chip cannot supply it. Vera Rubin for agents, because tool calls stall multi-billion-dollar GPU systems and so the CPU has to be optimized for low-latency single-threaded code. Feynman for swarms. That sequence is not marketing. It is a falsifiable thesis about where the bottleneck moves next, and every other infrastructure company should be measuring themselves against it. If Huang is right that swarms of sub-agents are the next dominant pattern, then the design pressure shifts from raw flops to fabric topology, memory hierarchy and storage-to-GPU latency. That has implications for everyone downstream, including the hyperscalers building competing accelerators.

    The MFU section is the most intellectually generous moment in the talk. The instinct in the AI ops community has been to chase MFU as if it were a virtue. Huang argues, persuasively, that low MFU is consistent with high tokens per watt in a disaggregated inference setup, and that bottlenecks rotate fast enough that over-provisioning every resource is the rational design. That reframing matters because it changes what “scarce” means. Compute is not scarce in the way the discourse treats it. What is scarce is a coherent system designed end-to-end. The xAI 11 percent number, in that frame, is not embarrassing. It is the natural reading of a workload that is mostly decode.

    The Stanford segment is the part most likely to be quoted out of context. “It’s Stanford’s fault” is a deliberately provocative line, but the underlying claim is correct and load-bearing. Compute is not gated by NVIDIA refusing to ship chips. It is gated by the fact that fragmented grant funding cannot aggregate into the billion-dollar order that NVIDIA can fulfill. The implication is that universities and national labs need a structural change in how they pool capital for compute, and that the current model of every researcher buying a handful of cards is genuinely obsolete. Huang’s nudge about pulling a billion off the endowment is concrete enough to be acted on, and other major research universities should read this segment as a direct prompt.

    The geopolitical segment is the highest-stakes one. The telecommunications comparison is correct as a historical pattern, and Huang is one of the very few executives in a position to deliver that warning credibly. The unresolved tension is that the argument applies symmetrically. If American AI dominance is built by selling globally, that includes selling into adversarial states, and the policy question is where the line falls. Huang does not answer that question. He attacks the framing that lets the question be answered badly. That is a meaningful contribution to the discourse even if it does not resolve the underlying tradeoff.

    The career advice section is the part the social-media clips will mishandle. “Seek suffering” reads as macho when extracted. In context it is a specific operational claim about how resilience compounds, and it is paired with the Tegra story where Huang himself paid the price of not thinking one more click ahead. That kind of self-implication is rare in CEO talks, and it is the reason the talk is worth listening to in full rather than only reading the recap.

    Watch the full Stanford CS153 Frontier Systems conversation with Jensen Huang here.