A $348 billion markup in under six months puts OpenAI at 50 times its March run rate

OpenAI told investors in March that it was earning $2 billion of revenue a month, with enterprise customers producing more than 40% of it ^[6]. Annualise that and the run rate is $24 billion ^[1]. Roughly $1.2 trillion against $24 billion is 50 times revenue ^[2]. The March post-money of $852 billion against the same figure was 35.5 times ^[3]. The distance between the two marks is $348 billion of paper value in under six months ^[4].

Investors made the approach, according to Reuters ^[3]. That changes who sets the price. The March round showed what this group of buyers can write at once: $122 billion of committed capital, 14.3% of the $852 billion post-money ^[4][5], from Amazon, Nvidia and SoftBank, with Microsoft continuing to participate and $3 billion more raised from other investors through banks ^[5].

Altman said on September 12 that OpenAI would not go public in 2026, citing frontier AI safety ^[7], and the company has joined Anthropic in endorsing tougher rules for more powerful systems ^[8]. Amazon, Nvidia, SoftBank and Microsoft hold the position listed investors cannot buy directly, and their own shareholders own it at one remove ^[5]. Forge Global counts 2.3 years for xAI to pass $100 billion and about 4.5 years for Anthropic, against roughly 16 years for companies founded before 2011 ^[12]. More of the compounding now happens before a listing.

The private marks and the revenue ranking point different ways. DefiLlama's pre-IPO tracker put Anthropic at $1.48 trillion and OpenAI at $903.29 billion on September 14, figures it describes as estimates of pre-IPO value, not the valuations set in the latest funding rounds ^[9]. The tracker's OpenAI number is 6% above the March post-money ^[8], so a $1.2 trillion print would be 33% above the tracker ^[7]. Anthropic sits about 64% above OpenAI on those estimates, and Cryptopolitan calculates the gap narrowing to nearly 23% if OpenAI reprices and Anthropic's estimate holds ^[10]. Cryptopolitan has also reported that Anthropic's annualized revenue run rate passed OpenAI's ^[11].

On the spending side, Gartner expects global outlay on AI platforms and models to reach $64.25 billion, 63.4% more than 2025 ^[14], and Goldman Sachs Research forecasts more than $1 trillion of AI investment worldwide in 2026, $581 billion of it in the United States ^[13]. The $348 billion added to OpenAI's paper value is 5.4 times Gartner's whole category ^[6]; one figure is a stock of value and the other a year of spending. "Enterprise AI budgets are coming under greater scrutiny, with increased focus on usage efficiency, cost control and measurable outcomes," said Gartner's Arunasree Cheparthi ^[15].

Doubled since March, $1.2 trillion is 25 times a $48 billion run rate ^[9], and the markup prices growth already delivered. Flat, and buyers are paying 41% more ^[17] for six months that produced no new revenue figure. McKinsey found 40% of respondents at large companies actively implementing AI agents against 22% at small ones ^[16], and enterprise customers already supply more than 40% of OpenAI's revenue ^[6].

I'd expect any round to price near the number, because the buyers are the ones who asked for the meeting. A close materially below $1.2 trillion would say the inbound demand was thinner than the approach implied. The report does not describe the terms, the preference or the ratchets that can make a headline valuation soft.