Chips & Compute

Sam Altman Warns Chip Suppliers Are All Counting the Same Buyers

Sam Altman says chip suppliers are all assuming OpenAI and Anthropic will buy their full output, warning short-term supply and demand may not match as AI stocks slide.

Sam Altman Just Delivered a Warning to Investors Betting on Endless AI Growth
Sam Altman Just Delivered a Warning to Investors Betting on Endless AI GrowthQuanlecntt2004 / pixabay
By James Calloway5 min read

Why it matters

  • NVIDIA stock is down 28.88% over the past month as Altman flagged demand concentration on November 16.
  • OpenAI has committed to about 12 gigawatts of NVIDIA AI infrastructure through 2030; NVIDIA expects credit-backed AI labs to be roughly a quarter of next year's business.
  • CoreWeave plans $37B-$39B fiscal 2026 capex against a $709M Q3 net loss and $846M net interest expense.
  • IREN's net debt rose to $4.43B from $216M and its shares fell 42.39% in a month.
  • Anthropic's October 26 S-1 says it "may be committed to excess compute that remains underutilized."

NVIDIA stock is down 28.88% over the past month, and OpenAI chief executive Sam Altman just told investors part of the reason why: many AI chip suppliers are planning their futures around the same handful of buyers. Speaking at Semafor's November 16 Silicon Valley & the World event, Altman pointed to a structural weakness in the AI hardware boom. Suppliers across the silicon supply chain, he said, are each assuming the big AI labs will absorb their full output. The math does not obviously work if several of them are counting the same orders twice.

"There are now many people making chips. And many of those people think they're going to sell most of the volume to us and Anthropic, and others. And we have some limited amount of demand and data centers, and everything else. And I think there are many people making chips, all talking to the silicon supply chain, assuming that we're going to buy all of their chips," Altman said.

What exactly is Altman worried about?

Altman was careful to separate the long-term picture from the near-term one. He said underlying data center demand remains solid, and he expects long-term computing demand to keep growing. His concern is timing. He questioned whether supply and demand will line up over shorter periods when several suppliers may be banking the same future orders as firm commitments. In plain words: the total demand may be real, but the supply chain may be building capacity against a demand book that double-counts a few customers.

The remarks landed as AI stocks were already selling off. NVIDIA (NASDAQ:NVDA) has fallen 28.88% over the past month, and the broader infrastructure complex has dragged with it. Altman's comments give the pullback a named mechanism: demand concentration — a small set of buyers appearing simultaneously in every supplier's forecast.

How concentrated is the customer base?

The evidence sits in the suppliers' own statements, and it is remarkably consistent.

In August, NVIDIA's chief executive at the time, Jensen Huang, said "OpenAI has committed to substantial deployments of NVIDIA AI infrastructure through 2030," which amounts to about 12 gigawatts of capacity. NVIDIA expects AI labs backed by its own balance sheet to make up "roughly a quarter of our business next year." The company posted Q2 revenue of $96.22 billion and guided Q3 to $108 billion, plus or minus 2%. A quarter of revenue resting on credit-supported labs is exactly the kind of concentration Altman is describing.

Advanced Micro Devices (NASDAQ:AMD) tells a similar story from the supply side. Its first Helios racks are already running workloads for Anthropic, Meta, Microsoft, OpenAI and Oracle Cloud Infrastructure. CEO Lisa Su said "It is true that demand is higher than supply," and AMD expects 2027 revenue growth above 80%. Growth that steep depends on the same five or six names continuing to buy.

Micron Technology (NASDAQ:MU) has "more than 75% of our shipments committed" for fiscal 2027, according to management. The company also said it lacks "line of sight to when supply and demand balance." Its fiscal Q4 revenue rose 379.3% — a number that reflects memory pricing power driven by AI demand, but also a customer base narrower than the revenue spike suggests. The takeaway is that commitments cover most of the book, yet the timing of balance remains opaque even to the supplier.

Coherent (NYSE:COHR), which supplies optical interconnects, rounds out the picture. CEO Jim Anderson said "Calendar 27 is basically booked out and the customers are booking out calendar 28 now." The company also signed a PhotonLink agreement worth about $20 billion with a major AI company. A single unnamed AI customer anchoring a $20 billion component deal is concentration in its purest form.

Who carries the financing risk?

The risk does not stop at chipmakers. It concentrates in the so-called neoclouds — specialized cloud providers renting AI compute — that finance data center buildouts with debt against future contracts.

CoreWeave (NASDAQ:CRWV) plans fiscal 2026 capital expenditure of $37 billion to $39 billion. In Q3 it posted $846 million in net interest expense and a $709 million net loss. CEO Michael Intrator insisted "We are not a growth-at-all-costs business," but the stock is down 33.12% year to date. Interest expense of that size against a net loss means the model only works if contracted demand converts to revenue on schedule. If Altman is right that short-term supply and demand may mismatch, the neoclouds sit closest to the blast radius.

IREN shows the same stress in sharper relief. Co-CEO Daniel Roberts said investor questions "have undoubtedly moved from how big the contract is to whether it can be delivered, financed, and run at a profit." That sentence is the market's shift in one line: size is no longer the question, execution is. IREN's net debt rose to $4.43 billion from $216 million, and the shares fell 42.39% in a month. A twenty-fold increase in net debt inside a compressed window is the balance-sheet expression of the buildout Altman is cautioning about.

What does the buyer side now admit?

Demand concentration is no longer just an observer's observation — it appears in the buyers' own filings. Anthropic's October 26 S-1 filing states the company "may be committed to excess compute that remains underutilized." That is the mirror image of the suppliers' problem: the labs may have over-committed just as the chipmakers may have over-built. Both sides of the market are now disclosing, in their own words, that the matching of supply and demand is not guaranteed.

The next data point arrives with NVIDIA's earnings report — its first since Huang's ouster. The number to watch is not the revenue figure but the customer list. If NVIDIA names the same leading research labs that AMD, Micron and Coherent cite, Altman's warning stops being a hypothetical and becomes an audited fact. The market will then have to price what happens when one set of buyers underwrites an entire supply chain's growth expectations.

Source: 247wallst.com

Share this article:

More from James Calloway

James Calloway

Show full bio

News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.

722 articles

Related articles

  1. Nvidia Case Puts AI Chip Supply Chains Under New Scrutiny
  2. AMD Is Up 158% While Nvidia Slips. The Case Runs Deeper Than Rival Trouble
  3. Analysts: Nvidia's Wobbling Chip Dominance Hands Samsung an Opening
  4. NVIDIA’s 4% Drop Puts a $108B Earnings Test in Focus
  5. Intel's 28% Crash Wipes $12 Billion Off the US Government's Stake

« Previous articleNext article »