A year ago, ChatGPT held 86.7% of generative AI web traffic. By January 2026, that share had fallen to 64.5%. Over the same period, Google’s Gemini climbed from 5.7% to 21.5%. Anthropic crossed $30 billion in annualized revenue in April 2026, overtaking OpenAI, which sits at approximately $25 billion, while spending roughly a quarter of what OpenAI allocates to model training. The ground Anthropic has taken is concentrated in enterprise and coding, the same segments OpenAI had positioned as core growth drivers.

The consequences of that competitive shift are now visible in OpenAI’s internal numbers. The company missed multiple monthly revenue targets in early 2026 and fell short of its goal to reach one billion weekly active ChatGPT users by the end of 2025. Subscriber losses were recorded during the same period. These are not isolated misses. They are the direct result of two competitors gaining sustained traction in the segments OpenAI was built to dominate.

The revenue shortfall would be a contained problem for most companies. For OpenAI, it collides with a capital structure that was sized for a faster-growing business. The company has committed approximately $600 billion in compute spending through 2030, including a reported $300 billion five-year infrastructure agreement with Oracle and a $250 billion commitment on Microsoft Azure through 2032. CFO Sarah Friar has told colleagues internally that if revenue does not accelerate, the company may be unable to honor its future compute agreements. The board has begun scrutinizing existing data center deals and has applied new oversight to CEO Sam Altman’s push to accumulate additional computing capacity.

The organizational dimension adds further weight. Friar does not report to Altman. She reports to Fidji Simo, OpenAI’s CEO of Applications, an arrangement in place since August 2025. The Information reported in early April 2026 that Altman had excluded Friar from conversations with investors and from meetings concerning key financial decisions, including server procurement. Friar has also told colleagues she believes OpenAI is not organizationally ready for the Q4 2026 IPO Altman has committed to, preferring a 2027 listing instead.

Altman and Friar responded to the Wall Street Journal’s April 28 report with a joint statement issued to Reuters:

The statement addressed alignment on compute acquisition. The Journal’s reporting had focused on whether existing revenue is sufficient to fund commitments already made. Those are two distinct questions, and the statement answered only one of them.

Markets responded to the distinction. Oracle fell 7.7%. CoreWeave, which holds an $11.9 billion infrastructure contract with OpenAI, dropped 7.4%. SoftBank, which has committed $60 billion to OpenAI, declined nearly 10% in Tokyo trading. Nvidia, Broadcom, and AMD each fell between 2 and 6%.

OpenAI closed a $122 billion funding round at a post-money valuation of $852 billion as recently as late March 2026. The company has pointed to GPT-5.5, which performed strongly across recent industry benchmarks, and to Codex, which has been gaining users in the developer segment, as evidence that its technical position remains competitive. The restructured Microsoft partnership, announced the same week as the Journal report, ends Microsoft’s exclusive distribution rights and opens OpenAI’s products to Amazon Web Services and Google Cloud customers, expanding its direct sales reach.

The question the numbers have not yet resolved is whether that expanded reach and product performance will generate revenue at a pace sufficient to meet obligations that were sized for a faster-growing company.