On October 8 local time, U.S. stocks experienced an unexpected "earthquake".
The trigger was a report from the UK’s Financial Times. The report stated that OpenAI informed investors that its annualized revenue as of the end of September was “close to $50 billion,” significantly lower than the approximately $70 billion previously reported.
Upon the announcement, the market reacted sharply: Oracle's stock plunged more than 5%, NVIDIA fell about 2.8%, AMD dropped over 4.6%, Microsoft also closed lower, and the Nasdaq Index widened its afternoon loss to 1.4%.
ARR: A "bloat" forecast
To understand this, you first need to understand ARR.
ARR (Annualized Revenue) does not measure the actual cash a company has received, but rather estimates the annual revenue by multiplying the revenue from the most recent month by 12.
For a company like OpenAI, this number is naturally "inflated."
The Financial Times said the confusion originated with Anthropic. Anthropic includes all sales completed through cloud partners in its revenue calculations, while OpenAI uses a net basis, recording only the portion it actually retains. To make the two companies “comparable,” investors actively adjusted OpenAI’s figures upward, first circulating an annualized revenue figure of $40 billion for August, and later, after OpenAI informed investors that revenue growth exceeded 70%, multiplying that high base yielded a figure of approximately $70 billion.
But OpenAI's own figures have consistently been close to $50 billion.
According to CNBC, the $70 billion in ARR includes "revenue sharing from major partners" such as Amazon and Microsoft; $50 billion excludes these partnership transactions.
Tech influencer Tae Kim commented that the bulk of the difference stems from OpenAI’s revenue-sharing payments to Microsoft, reportedly around 20%, continuing until 2030. Some might argue that gross revenue is a better metric for evaluating the performance of AI model businesses, since this revenue-sharing agreement will eventually expire—after which gross and net revenue will converge, making this entirely “not news” to anyone tracking the company.
But right now, the market is only seeing the huge gap between “$50 billion” and “$70 billion”.
The "OpenAI dependency" in the AI sector
A single report caused Oracle to drop more than 5% and exacerbated losses for NVIDIA, AMD, and Microsoft, illustrating how highly leveraged the entire tech sector is on OpenAI and Anthropic.
OpenAI is no longer just a company—it’s the central anchor of the AI capital expenditure narrative. Oracle’s cloud orders, NVIDIA’s GPU demand, AMD’s追赶 story, and Microsoft’s AI monetization expectations are all, to varying degrees, tied to OpenAI’s expansion pace. When this anchor sends even an accounting-level “shrinking” signal, the market will flee first.
CNBC’s “Mad Money” host and former hedge fund manager Jim Cramer said the report contains no new information—OpenAI has simply rephrased how it confirms revenue, and the market will soon dismiss it as a “nothing burger.” Cramer has consistently supported the demand for AI computing power, but he has also repeatedly warned about OpenAI’s financial structure and the “story pricing” risks in the AI sector.
"Serenity, the 'White-Haired Stock God,' believes that AAOI's laser shortage will last three to five years, and Micron's memory components are also in high demand—these are real supply and demand facts. A report may change accounting methods, but it cannot change the reality that computing power, optical modules, and storage remain scarce."
However, the pace of financing around AI continues to accelerate. OpenAI is seeking at least $30 billion in new funding, with a reported valuation as high as $1.4 trillion. Based on an annualized revenue of $50 billion, the price-to-sales ratio is approximately 28x; based on the previous $70 billion figure, it’s about 20x. Broadcom is also arranging over $50 billion in financing to customize chips for OpenAI.
On the prediction market Polymarket, traders assign approximately a 58% probability to OpenAI going public via IPO before the end of 2027 with a first-day market capitalization of at least $1.5 trillion, down from about 80% in September; the probability of it not going public before the end of 2027 has risen to approximately 23%.
The AI story hasn't changed, but the market's patience is no longer what it used to be.
Author: bootly
Twitter: https://twitter.com/BitpushNewsCN
BitPush Telegram community: https://t.me/BitPushCommunity
BitPush TG subscription: https://t.me/bitpush