POPA ReconOpenAI's run rate: $50B at end of September versus $70B by year-end. The two claims mostly don't conflict, and OpenAI itself hasn't confirmed either
The two reports fit together. The first is a figure for a past date and the second is a forecast, and the "$20B gap" comes from how revenue is counted. It is not a drop in sales. Neither report has been confirmed by OpenAI, and I couldn't read the Bloomberg or FT articles themselves, so everything below comes from secondhand accounts.
What each report says
Bloomberg (9 Oct): its X post says OpenAI expects to reach or exceed $70B in annualized revenue by year-end, driven largely by enterprise growth. A Telegram relay adds three details. The figures were shared with investors in $30B+ funding talks. The round is discussed at a $1.4T pre-money valuation, with UAE funds including MGX in talks to anchor it. The relay also says the discrepancy came from investors using Anthropic's revenue accounting.
FT (8 Oct), as relayed by others:
Coin Bureau says OpenAI told investors its annualized revenue was "approaching $50 billion" at end-September, and that OpenAI declined to comment.
The earlier $70B came from media reports in late September, which Axios, as relayed, put at "nearing $70 billion".
MelvinInvests says investors tried to make OpenAI comparable with Anthropic. That reportedly added about $20B of estimated partner revenue.
The accounting difference
Per the relayed accounts, Anthropic's annualized figure includes gross sales made through cloud partners such as AWS and Google Cloud. OpenAI's figure does not include comparable partner sales.
One post cites an unnamed source on a SiliconANGLE page saying the older figure was partner gross and the newer one is net. That is a single unverified source.
CNBC reportedly confirmed roughly $50B at end-September. The same post says CNBC also reported 77% run-rate growth in Q3 and 107% in enterprise. I did not see CNBC's text.
A Futurum comment puts the gap at the value of sales flowing through cloud partners. This is commentary, not a disclosure.
Are the claims consistent with each other?
"Reports conflict": no. Bloomberg's $50B (end-September) and $70B (end-year target) are different dates. Reaching $70B from $50B in three months would be roughly 40% growth. The Telegram post saying OpenAI "reported $50B against a promised $70B" blends two different things, as the markettwits post does.
The FT's "$20B below previously signaled": this compares OpenAI's own net figure with an investor-adjusted, Anthropic-style estimate. By the relayed accounts, the $70B was never an OpenAI-stated number. One account instead says it came from OpenAI's own earlier investor materials. The sources conflict on this point.
Still open: whether the $70B year-end target uses OpenAI's net basis or the Anthropic-style gross basis. The accounts I saw don't say. If it is net, it implies a faster pace than the Q3 growth reported by CNBC. If it is gross, it isn't comparable to the $50B.
OpenAI's own statements
OpenAI's account and Sam Altman's posts through 8 Oct contain no revenue figures. They cover GPT-6, Intelligent UI and Ultrafast.
Altman does call 6.1 Sol "our fastest-growing model ever". That supports growth in general terms but gives no revenue number.
A post says OpenAI has not published the $50B on its own site. I did not check the site.
All the numbers come from sources familiar with the matter and investor materials. As one commenter puts it, OpenAI needs to clarify.
Market reaction
Per secondhand recaps, the Nasdaq fell about 1.25% on 8 Oct and chipmakers 3.4%. Oracle fell 5.5% and Broadcom 4.4%. Reuters tied the drop partly to the FT report and to Broadcom lining up $50B of financing for OpenAI. Oil and yields also moved that day.
One poster called the FT piece misleading. That is opinion, not evidence of error.
The sources don't name the exact basis of the $70B year-end target, and OpenAI has made no on-record statement on any of these figures.
Bloomberg post · Coin Bureau on FT · OpenAI on X · Axios relay