A $20 Billion Gap! OpenAI Revenue Falls Far Short of Expectations, AI Sector Hit Hard, Oracle Drops Over 5%, Applied Optoelectronics Plunges 13%

OpenAI's annualized revenue is far below the figures previously circulating in the market, triggering a large-scale sell-off of AI concept stocks. The Nasdaq Index ended the day down more than 1%, as investors' expectations of sustained strong demand for AI are facing a severe test.

On October 8th, The Financial Times reported that OpenAI disclosed in its latest investor filing that the annualized revenue as of the end of September is "close to $5 billion," significantly lower than the widely quoted market expectation of $7 billion at the end of last month.

In response to the annualized revenue figures revealed by The Financial Times, CNBC subsequently verified and confirmed the accuracy of the data.

The report points out that the magnitude of this gap "is very likely to dampen market optimism about growth in AI demand." After the news was released, U.S. technology stocks tumbled sharply that day, with the Nasdaq Index closing down 1.4%.

The AI sector generally plummeted. Nvidia fell about 3% on the day, Oracle dropped more than 5.5%, while AMD, Broadcom, Intel, and Super Micro Computer all declined between 4% and 6%.

Data center and optical interconnect stocks plunged. Cloud service stocks Nebius and CoreWeave both fell more than 7%, while optical interconnect stock Applied Optoelectronics plummeted 13%, and Coherent was down 7%.

The blow to the market from this revenue gap stems not only from the numbers themselves but also shakes the core narrative supporting massive AI capital expenditures, questioning whether end-user demand for AI is truly that robust. More importantly, this divergent expectation arises at a time when Token costs continue to decline.

On October 8th, media cited sources stating that the large discrepancy in revenue figures was due to OpenAI and rival Anthropic using different methodologies when calculating annualized revenue.

Anthropic includes revenue generated via cloud partners such as Amazon AWS and Google Cloud in its annualized revenue statistics—that is, the "total revenue" metric; while OpenAI’s annualized revenue figure only counts income directly generated by itself, excluding sales through partner channels.

According to reports, OpenAI investors tried to "standardize" OpenAI's figures to enable lateral comparison with Anthropic, which led to the previously discussed $7 billion figure.

However, after OpenAI’s latest official investor briefing clarified the situation, the market realized that the widely cited previous figure was highly misleading.

The report notes, OpenAI's latest documents show that by the end of July, its annualized revenue was about $3 billion, not the previously reported $4 billion; however, the company also demonstrated strong growth, with overall operating income for the third quarter up 77% year-on-year, and enterprise business up 107% year-on-year.

OpenAI itself has declined to comment publicly. As a private company, it is not obliged to disclose financial data regularly.

This revision of the revenue data comes during OpenAI's most critical valuation-building window, making the timing particularly sensitive.

OpenAI is currently in talks with investors for a new round of financing, which could set the company's valuation at about $140 billion. Previous reports have stated the company intends to raise about $30 billion, though terms have not been finalized.

As recently as March, OpenAI completed a historic fundraising round of $122 billion. The company’s CFO, Sarah Friar, said last week that the company is "well-capitalized."

However, OpenAI is expected to accumulate about $28 billion in losses by 2030, while its valuation stands at $85.2 billion. How to prove its growth trajectory to investors has become a core issue for the company.

OpenAI secretly submitted an IPO application to regulators in June this year, and the market widely expects it will list officially in 2027. CEO Sam Altman stated in September that "now is not a wise time to go public," partly due to ongoing controversies over AI safety.

The company recently announced it had halted plans to release its GPT-6.1 Astra model, as the model did not meet internal safety standards.

In sharp contrast to OpenAI, Anthropic disclosed in August that by the end of July, its annualized revenue run rate had reached $6.5 billion. At that time, the market expected its valuation to be as high as $200 billion, and the company was actively engaged with potential investors in preparation for an IPO.

However, independent financial research firm New Constructs released a report this Tuesday, directly calling Anthropic’s planned listing "the most ridiculous IPO of 2026," estimating the company's actual value at only about $15 billion.

According to its prospectus, Anthropic’s revenue for 2025 is $460 million, with a net loss of up to $4.2 billion for the same period.

OpenAI and Anthropic’s revenue figures have come under intense scrutiny, reflecting the market’s deep concerns about the commercialization process for the entire large AI model sector.

The Financial Times points out that the annualized revenue figures for these two companies are regarded as the most important single indicators of global AI demand, directly underpinning a vast amount of AI infrastructure investment decisions and public market valuation logic.

When such a core data metric experiences this level of discrepancy, the pressure for market repricing will inevitably spread throughout the entire industry chain.