OpenAI told prospective investors its annualized revenue is approaching $50 billion, about $20 billion below the $68 billion figure reported two months earlier. The Financial Times disclosed the lower number on the day the Nasdaq fell 1.25% and the S&P 500 lost 0.5%. For companies buying AI infrastructure and software, the gap matters because revenue strength underpins OpenAI ability to meet multibillion-dollar contracts.
Why the $50 billion figure differs from earlier reports
The earlier $68 billion figure included gross revenue from OpenAI partners to allow a more direct comparison with Anthropic numbers, according to an anonymous source cited in the report. The newer $50 billion figure is based on net revenue and was shared during an investor presentation. The same presentation cited 77% growth in annual revenue run rate in the third quarter and 107% growth in the run rate for the enterprise business. Those two metrics frame the revision as an accounting definition change paired with continued expansion.
The mechanics of the market reaction were straightforward: both indexes opened lower, then declines accelerated after the Financial Times report. Oracle fell 5.5%, Intel declined 5.3% and Nvidia lost 2.9%. CoreWeave dropped 8%, Advanced Micro Devices and Broadcom each slipped 4%, and Super Micro Computer fell almost 5%. Oracle exposure is direct through significant contracts with OpenAI valued in multiple billions of dollars, while the others sit in the compute supply chain around graphics processing units, chips and servers.
The episode lands against a year in which many technology stocks rose on expectations that AI demand will keep growing for several years. OpenAI carries a recent valuation of $852 billion and confidentially filed its prospectus with the Securities and Exchange Commission in June. The company first planned to go public this year, then signaled a 2027 debut, with Chief Executive Sam Altman confirming the delay last month amid debate around AI safety. Anthropic is still linked to a possible 2026 IPO, reportedly as soon as Nov. 9, with a sought $2 trillion valuation.
What the repricing means for AI buyers
For corporate users of AI, the practical effect runs through suppliers rather than model quality. If OpenAI revenue is lower than assumed, customers may face tighter scrutiny of long-term capacity commitments, enterprise pricing and service levels tied to infrastructure spending. Large companies with Oracle, Nvidia-based or CoreWeave-backed deployments feel such shifts first because contract volumes are larger and replacement is slower. Smaller firms buying standard seats or API access can switch vendors faster, but they depend more on price stability and continuity of startup partners.
The limits of the news deserve equal attention. A change from gross to net revenue does not by itself prove weaker demand, and the reported 77% and 107% run-rate growth points still indicate expansion in overall and enterprise use. Anthropic comparison is also unsettled: Bloomberg cited annualized revenue above $65 billion in August, while a leaked prospectus seen by Reuters showed $4.6 billion in 2025 revenue and more than $42 billion in losses. Independent researcher New Constructs called the Anthropic offering the most ridiculous IPO of 2026 and valued it at $150 billion. Buyers should therefore ask vendors how revenue is defined, what covers partner resale, and what backs delivery guarantees.
The marker to watch is whether OpenAI and Anthropic hold their IPO timetables and valuation targets through year-end. A confirmed 2026 Anthropic filing or a firm OpenAI schedule for 2027 would suggest capital markets still accept current AI economics. Postponements, lower targets or new contract renegotiations would signal that customers should diversify suppliers and lock in exit terms.
