OpenAI is locked into $600 billion in data center commitments. Its CFO just warned it might not make the next payment. The market heard it: Oracle fell 7.5%, SoftBank fell 10%, and chip stocks fell 2 5%.
OpenAI has a math problem. Not a technology problem. Not a product problem. A math problem, and its chief financial officer just told people it might not be able to pay its next bill.
According to The Wall Street Journal, OpenAI recently missed its own internal revenue targets for the first time in its history. CFO Sarah Friar has told colleagues she is worried the company might not be able to pay for future computing contracts if revenue growth does not accelerate, The Journal reported. For a company that has raised more than $120 billion in capital and is preparing for a public offering, that is an unusual thing for its finance chief to say.
The reaction was immediate. Oracle, which has a roughly $300 billion five-year partnership to supply computing power to OpenAI, fell 7.5 percent in premarket trading. SoftBank, one of OpenAI's largest investors, dropped about 10 percent. CoreWeave, the cloud provider built specifically to serve AI workloads, fell 7 percent. Nvidia, Broadcom, and AMD each lost between 2 and 5 percent. The Nasdaq futures fell. None of those companies missed their targets. The market was repricing the counterparty.
The $600 billion question
Here is the structural problem that nobody else has connected to the miss. OpenAI is on the hook for roughly $600 billion in future data center spending, a figure that emerges from the infrastructure deals struck at peak AI enthusiasm over the past two years, The Decoder reported. Independent research from Sacra, which tracks AI company financials, puts disclosed cloud capacity commitments at more than $500 billion across multiple providers as of early 2026. That figure is roughly twenty times OpenAI's stated 2026 revenue target of $30 billion.
OpenAI brought in approximately $13 billion in revenue in 2025 while burning through roughly $8 billion in losses, The Decoder reported. The company expects to burn through $25 billion in cash in 2026 against that $30 billion revenue target. The gap between what the company owes and what it earns is not narrow. It is structural.
The irony is that the commitments were made in part because compute was the scarce resource. Locking in data center capacity was the rational move when AI infrastructure was oversubscribed and delivery timelines were uncertain. The problem is that the deals were struck before anyone knew whether the revenue would actually follow.
Anthropic in the gap
OpenAI is not just facing a financial problem. It is facing a competitive one. The company lost ground to Anthropic in both coding and enterprise markets, Reuters reported, missing multiple monthly revenue targets earlier this year as a result. ChatGPT fell short of an internal goal to reach 1 billion weekly active users by the end of 2025, according to Reuters.
Anthropic, by contrast, has been expanding quickly. The company's annualized revenue grew from roughly $1 billion at the start of 2025 to more than $5 billion by August of that year, Panto AI reported, and Claude Code alone reached $2.5 billion in annualized revenue by February 2026, Sacra estimated. Anthropic closed a $30 billion funding round in February at a $380 billion post-money valuation. It is not the one with the liquidity problem.
The IPO problem
OpenAI is trying to go public. It closed a $122 billion funding round in April at an $852 billion post-money valuation, the company announced. But a company preparing for an IPO whose CFO is worried about paying its compute bills is a company whose S-1 will require some explaining. The financial controls that public companies must maintain do not coexist easily with a CFO warning internally about liquidity risk on multi-hundred-billion-dollar infrastructure obligations.
Friar and CEO Sam Altman called the Journal's reporting ridiculous in a joint statement, saying they are totally aligned on buying as much compute as we can and working hard on it together every day. That statement was itself notable: when a CFO and CEO have to jointly deny internal friction over a financial risk, the denial does not entirely quiet the question.
What happens next
Four large cloud providers are on track to spend up to $660 billion on AI data centers in 2026, up 66 percent from 2025, JPMorgan estimated. That number is the bet the broader industry has made on AI infrastructure: that the compute will pay for itself as the technology scales. OpenAI is the largest single anchor tenant in that bet. If it cannot pay its bills on time, the others are exposed too.
The market has already started pricing that exposure. Whether it is a temporary selloff or the beginning of a structural repricing depends on whether OpenAI can actually close the gap between $600 billion in commitments and $30 billion in revenue. Nobody outside the company's financial projections knows the answer. That uncertainty is itself the story.