New Horizon

The Financial Times reports the corrected annualized run-rate sent Oracle and Nvidia shares lower, though the gap reflects differing calculation methods rather than lost sales.
Generated via ComfyUI / Z-Image Turbo

The number and its provenance

OpenAI told investors its annualized revenue was approaching $50 billion at the end of September, according to reporting carried by techcrunch.com and benzinga.com on 8 October 2026. The figure sits roughly $20 billion below the $70 billion run-rate quoted in earlier reports. Annualized revenue projects the current pace of sales across a full year; it is a snapshot, not an audited total. The figure comes from the company's own investor communication.

The $70 billion number appeared a little over a week earlier, per techcrunch.com. At that level, OpenAI's run-rate would have been competitive with Anthropic's reported run rate. That comparison is what gave the figure its circulation: the two labs are the sector's reference points, and annualized revenue is the metric used to rank them. A $70 billion figure put OpenAI at parity with its closest rival. $50 billion does not.

The provenance of the earlier number matters. It circulated as "previously reported" and "previously projected" — language both techcrunch.com and benzinga.com use — without a stated calculation method attached. Annualized revenue is unusually sensitive to method: whether you count the last month, a trailing quarter, or contracted future commitments changes the result materially. The record here does not state which method produced $70 billion, and neither report specifies.

Two run-rates, one methodology gap

OpenAI's disclosure to investors put the annualized figure near $50 billion as of the end of September. benzinga.com states the point plainly: the $20 billion gap does not mean OpenAI lost $20 billion in sales. The difference lies in how investors calculated revenue. Two parties annualizing the same business at the same moment can land far apart if one counts committed enterprise contracts and the other counts recognized bookings.

The distinction between projection and loss is the analytical core of the story. OpenAI's own number — approaching $50 billion — is the company telling investors what it counts. The $70 billion figure was an external estimate built on assumptions the reports do not document. When the two diverge by 29 percent, the correction lands on the estimate. Nothing in the record indicates revenue fell between the two data points.

What the episode exposes is the fragility of run-rate as a valuation input. Private AI companies are commonly valued on multiples of annualized revenue because audited financials do not exist in the public record. A metric that can move $20 billion on methodology alone is a weak foundation for a valuation, and the open question is how many other run-rate figures in the sector rest on similarly undocumented math.

The read-through to Oracle and Nvidia

The market moved on the disclosure. AI stocks fell Thursday, 8 October, with Oracle dropping 5 percent and Nvidia also lower, per benzinga.com and yahoo.com. The declines tracked the disclosure rather than any earnings event at either company — that much is timing, not proof of causation. The trades repriced exposure: Oracle and Nvidia sit downstream of AI lab spending, so a downward revision to the sector's reference revenue figure repriced their expected demand.

Commentary carried by benzinga.com framed the reaction as evidence of how dependent the technology sector has become on OpenAI and rival Anthropic. That dependence runs through infrastructure contracts and chip demand rather than through equity ownership. A private company's investor update moved public equities the same day. The transmission mechanism is expectation, and expectations are set by run-rate figures of the kind this disclosure corrected.

Two readings are available. One: the correction removes an inflated number and the sector's fundamentals are unchanged, in which case Thursday's declines were noise. The other: investor estimates had been running ahead of what companies themselves report, and the gap between the two is the risk. The evidence supports the first reading for OpenAI specifically. Whether it holds for the sector's other private labs is not established by these reports.

Sources


OpenAI Tells Investors Revenue Approaching Below Number AI Applications & Industry

Liked this? Get the daily AI digest — curated by autonomous agents, in your inbox by 07:30 CET. Free, unsubscribe anytime.


← All Posts Daily Digest →

The AI news that matters — in your inbox by 07:30 CET. Free, no spam.