Nvidia's potential $250B backstop for OpenAI's Ohio data center highlights circular AI deals and growing investor skepticism about massive capex without clear returns.
Nvidia shares fell more than 4% Monday after the Wall Street Journal reported the chipmaker is in talks to provide a roughly $250 billion financial backstop for OpenAI. The guarantee would help the Sam Altman-led startup lease a proposed 10-gigawatt data center in southern Ohio, being developed by SB Energy, a SoftBank subsidiary.
The deal, if it goes through, is another example of the circular financing that has come to define the AI buildout. Nvidia has invested in multiple companies that buy its chips, including a $30 billion stake in OpenAI earlier this year and an investment in Anthropic last year. The company has also backed so-called neoclouds that rent out its hardware.
Nvidia has said these investments support the AI ecosystem and offer attractive returns. But the sheer size of the potential OpenAI backstop – covering just the lease and debt for the facility, not the server racks inside – has investors questioning the sustainability of the model.
The Journal also reported that Nvidia is in talks to finance OpenAI's chip purchases beyond the $250 billion figure. That would tie the two companies even closer, raising concerns about a domino effect should one party fail to meet its obligations.
Investors are not fixated on what the backing is for, only that it is massive and tied to a cash-burning private company. With every commitment, OpenAI will find it harder to raise money without help, traders said. Nvidia's financial position is far stronger, but the interconnectedness of the two is a new risk factor.
On the other side of the trade, the news confirms the insatiable demand for AI compute. OpenAI believes it needs more capacity to serve customers, and it is pursuing unconventional ways to get it. If OpenAI is behaving that way, publicly traded hyperscalers – Meta, Alphabet, Amazon and Microsoft – may continue their own spending sprees, even if their stocks are getting punished for it.
That dynamic was on display last week. Alphabet reported strong second-quarter results but saw its stock fall Thursday on concerns about rising capital expenditures in 2026 and management's indication of another step-up next year. Meta, Microsoft and Amazon report this week.
Nvidia, with an Alpha Score of 74, is rated Moderate by AlphaScala's model. The stock fell 5.03% to $196.44 on Monday, extending a recent pullback. NVDA stock page
The market's reaction to Intel's superb results on Thursday night also underscores the shift. The "easy money" in tech has been made, and investors are now demanding that AI spending translate into earnings growth. Building to meet demand is no longer enough; how much did it add to the bottom line this quarter?
For the hyperscalers, that question is becoming harder to answer. Management teams that talk about the need to spend more without addressing the return on that spending will see their stocks under pressure, several analysts said.
Corning reports Tuesday morning, and Eaton reports Friday. Both are suppliers to AI data centers. Their earnings calls will offer a real-time check on whether the infrastructure buildout is actually generating revenue for the broader supply chain, or just shifting money around the same small circle of AI winners.
Beyond tech, stocks that benefit from lower oil prices and lower interest rates are gaining traction. Honeywell Aerospace and Boeing fall into this camp, as does Home Depot. The idea is that these companies benefit from a calming of Middle East tensions, regardless of what happens with AI spending in the near term.
The bottom line? The Nvidia-OpenAI backstop story reinforces the concerns about AI capex levels that had already been building. All this spending needs to show a strong return to soothe investor concerns; it eventually will, but for now, the market is rotating into opportunities outside the pure AI trade.
Corning reports Tuesday. Eaton reports Friday. The next batch of hyperscaler earnings from Meta, Microsoft and Amazon will arrive later this week.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.