Did OpenAI miss a $70 billion target, or did everyone else miscount? On October 8 the Financial Times reported that OpenAI's annualised revenue is approaching $50 billion, about $20 billion below a figure that had circulated for two weeks. The gap is real. Whose mistake it was is the more interesting question.

Where $70 billion came from

Around September 29, reports put OpenAI's run rate near $70 billion. According to the FT, as relayed by TechCrunch and Yahoo Finance, that number was not an OpenAI projection but an investor estimate. Investors trying to compare OpenAI with Anthropic first checked whether OpenAI's sales had reached $40 billion in August, then layered on roughly 70% growth the company had cited. The two companies also count annualised revenue differently: Anthropic includes sales through cloud partners, OpenAI does not. Anthropic's own run rate was $65 billion at the end of July.

Then Bloomberg added a twist on October 9: OpenAI has told investors it expects to reach or exceed $70 billion in annualised revenue by the end of 2026, driven largely by enterprise sales. So the same figure is both a wrong estimate of the present and the company's own year-end target. "Twenty billion below projection" is how it was headlined; "twenty billion below what had been reported" is more precise.

The numbers that are actually on the table

  • Annualised revenue, end of September: approaching $50 billion (FT; Bloomberg sources)
  • Year-end 2026 target: $70 billion or more annualised (Bloomberg, people familiar)
  • 2025 revenue: about $13 billion, with spending "significantly more" (leaked financials cited by TechCrunch)
  • Funding round: $30 billion or more at a $1.4 trillion pre-money valuation, in talks since September, not closed; UAE funds including Abu Dhabi's MGX among the anchors in discussion
  • Previous round: $122 billion in March 2026 at $852 billion post-money

OpenAI declined to comment to Bloomberg and did not respond to TechCrunch. No analyst reaction appears in any of the reports. The only market signal is from October 8, when technology stocks led the S&P 500 lower, the Nasdaq 100 fell 1.4% and a chip-stock index dropped 3.4%; Bloomberg noted the timing without claiming causation. Full-year 2026 revenue, compute commitments and user counts were not disclosed in any of the sourced reports.

So is this bad news?

For a company that made $13 billion last year, a run rate near $50 billion nine months later is not a miss by any ordinary standard. The valuation talks tell the same story: a $1.4 trillion pre-money figure after an $852 billion post-money round in March implies investors still expect sharp growth. What the episode exposes is how little verified information sits under these numbers. Run rates are self-reported, definitions differ between companies, and a two-week-old estimate was treated as a target until someone checked.

The timing also matters. The report landed a week after DevDay, with the GPT-6.1 Sol launch and "Dots" enterprise agents, and while OpenAI's IPO has moved to 2027 at the earliest. Anthropic is expected to file for its own listing soon, which would put audited numbers next to these estimates for the first time.

What a buyer should take from this

If you are choosing an AI vendor, the lesson is not about OpenAI's growth. It is that the public numbers around every frontier lab are estimates layered on estimates, and the gap between them can be $20 billion. Price your contracts and your architecture on what you can verify: the per-token rates in the published price list, the uptime you measure yourself, and the ease of switching models if the terms change. In our client projects we treat provider choice as a configuration, not a commitment, precisely because the ground under these companies shifts this fast. A vendor's valuation headline, up or down, should not move your roadmap.

Sources: TechCrunch, Bloomberg via Yahoo Finance, Yahoo Finance, relaying the Financial Times