OpenAI's annualized revenue came in far below figures previously circulating in the market, triggering a large-scale selloff in AI-related stocks and sending the Nasdaq down more than 1% on the day, as investors' conviction about persistently strong AI demand faces a severe test.

On October 8, the Financial Times reported that OpenAI's latest investor documents show its annualized revenue as of the end of September was "nearly $50 billion," well below the $70 billion expectation widely cited in the market at the end of the previous month.

Following the Financial Times disclosure of the annualized revenue figure, CNBC subsequently sought verification from relevant parties and confirmed the accuracy of the data.

The report noted that the magnitude of this shortfall "is likely to dent market optimism about AI demand growth." After the news was published, U.S. tech stocks fell sharply that day, with the Nasdaq closing down 1.4%.

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

Data center stocks and optical interconnect stocks were hit hard. Cloud services stocks Nebius and CoreWeave each fell more than 7%, optical interconnect stock Applied Optoelectronics plunged 13%, and Coherent dropped 7%.

The blow to the market from this revenue discrepancy lies not only in the numbers themselves, but in the fact that it undermines the core logic underpinning the narrative of massive AI capital expenditure鈥攚hether end-user demand for AI is truly as robust as believed. More importantly, this expectation gap emerged against the backdrop of continuously declining token costs.

Numbers diverge due to differing accounting standards

On October 8, media citing people familiar with the matter reported that the huge discrepancy in revenue figures stems from OpenAI and competitor Anthropic using different methodologies when calculating annualized revenue.

Anthropic includes revenue generated through cloud partners such as Amazon AWS and Google Cloud in its annualized revenue calculation鈥攖he "gross revenue" approach鈥攚hile OpenAI's annualized revenue figure only counts revenue generated directly by itself, excluding sales through partner channels.

According to reports, OpenAI's investors, seeking to make an apples-to-apples comparison between the two companies, attempted to "restate" OpenAI's data to match the broader methodology, which gave rise to the previously circulating figure of approximately $70 billion.

However, after OpenAI's latest formal investor briefing clarified the matter, the market only then realized that the previously widely cited figure was significantly misleading.

The report noted that OpenAI's latest documents show the company's annualized revenue at the end of July was approximately $30 billion, rather than the roughly $40 billion previously reported; however, the company simultaneously demonstrated strong growth momentum, with overall operating revenue growing 77% year-over-year in the third quarter and enterprise business growing 107% year-over-year.

OpenAI itself declined to comment, and as a private company, it is not obligated to regularly disclose financial data publicly.

Valuation pressure surges, IPO outlook clouds over

This revenue data correction comes at a particularly sensitive time鈥攄uring OpenAI's most critical window for building market valuation.

OpenAI is currently in talks with investors for a new funding round that could value the company at approximately $1.4 trillion, and there have been prior reports that the company plans to raise about $30 billion, though the terms have not yet been finalized.

As recently as March of this year, OpenAI completed a historic funding round of $122 billion, and company CFO Sarah Friar said last week that the company is "well capitalized."

However, OpenAI is projected to accumulate losses of approximately $280 billion by 2030, while its valuation stands at $852 billion. How to prove its growth trajectory to investors is a core challenge the company must confront.

OpenAI confidentially filed for an IPO with regulators in June of this year, and it is widely expected to go public in 2027. CEO Sam Altman said in September that "now is not an opportune time to go public," citing ongoing controversies over AI safety as one reason.

The company recently announced it is pausing the release of its GPT-6.1 Astra model because it failed to meet internal safety standards.

Anthropic also under pressure as AI large-model IPO narrative cools

In sharp contrast to OpenAI, Anthropic previously disclosed in August that its annualized revenue run rate had reached $65 billion as of the end of July, with outside valuation expectations at the time reaching as high as $2 trillion. The company is actively engaging with potential investors in preparation for an IPO.

However, independent financial research firm New Constructs released a report on Tuesday directly calling Anthropic's proposed listing "the most absurd IPO of 2026," arguing that the company's actual value is only about $150 billion.

According to its prospectus, Anthropic's 2025 revenue was $4.6 billion, with a net loss of $42 billion over the same period.

The revenue data of both OpenAI and Anthropic being placed under the magnifying glass in succession reflects deep-seated market doubts about the commercialization progress of the entire large-model sector.

The Financial Times noted that the annualized revenue figures of these two companies are regarded as the single most important indicator for measuring global AI demand, directly supporting a large number of AI infrastructure investment decisions and the valuation logic of the public equity market.

When core data shows a discrepancy of this magnitude, the pressure for market repricing will inevitably transmit through the entire industry chain.