OpenAI’s New $30bn Round Values It at $1.4 Trillion — the Same Number as Its Compute Bill

Last updated: 8 October 2026. Figures below are sourced to primary publications and linked inline; where a number is self-reported by a company, that is noted explicitly.

The 60-second version

• OpenAI is reportedly pitching a $30bn round to UAE sovereign funds led by Abu Dhabi’s MGX, alongside BlackRock, at a $1.4 trillion pre-money valuation — roughly 64% above its $852bn valuation from March 2026.
• Unusually, OpenAI has fixed that $1.4 trillion price itself rather than letting investors negotiate it, and has not yet settled on a lead investor.
• By coincidence, $1.4 trillion is also the figure OpenAI’s own compute-commitment disclosures totalled earlier in 2026 — a number it has since quietly revised down.
• The raise lands days after CEO Sam Altman told Politico’s Decoded podcast the world “should accept some bad things happening” for AI’s benefits, drawing public criticism from Senator Bernie Sanders.
• It also follows, by about a week, OpenAI president Greg Brockman signing Trump’s voluntary AI safety accord — a pact with no penalties, no public-audit requirement and no incident-reporting duty.
• OpenAI’s revenue run-rate has reportedly jumped from roughly $40bn in August to near $70bn by late September, which changes how extreme the new valuation multiple actually looks.

A fixed price, and no lead investor

OpenAI is in talks with several UAE investment funds, including Abu Dhabi-based MGX, to anchor a new round of at least $30bn, with the UAE side discussing as much as $10bn combined, and BlackRock in separate discussions to join, according to Bloomberg’s reporting. OpenAI and BlackRock both declined to comment; MGX did not respond. UC’s endowment, Thrive Capital and Andreessen Horowitz have also reportedly been approached, per a Bloomberg-sourced account carried by Business Today Malaysia.

What is unusual here is not the size of the round — OpenAI raised $122bn in March 2026 at an $852bn valuation — but its structure. OpenAI is presenting the $1.4 trillion pre-money figure as fixed, rather than a number for prospective investors to negotiate, and has not locked in a lead investor before setting that price. That inverts the normal order of a funding round, where a lead investor typically sets terms that others then follow. It is also a sign of how much leverage OpenAI believes it holds: MGX has previously backed both OpenAI and Anthropic, and has itself raised close to $50bn this year to deploy into AI infrastructure, so sovereign capital chasing the sector is not scarce. OpenAI has also reportedly pushed any IPO to 2027 at the earliest, after filing confidentially in June, with Altman telling Fortune in September that going public in 2026 would be “an ill-advised moment” given safety concerns.

Two $1.4 trillion figures, one company

Here is the detail most coverage has missed, because it requires reading two separate stories side by side. The $1.4 trillion valuation OpenAI is now pitching to investors is not the first time that exact figure has attached itself to the company this year. In January 2026, OpenAI’s own CFO Sarah Friar disclosed that its compute and infrastructure commitments — spanning deals with SoftBank, Oracle, Nvidia, Amazon and others — totalled roughly $1.4 trillion over the following eight years. By February, that number had become uncomfortable enough that OpenAI quietly revised it: CNBC reported a tempered compute-spending target of around $600bn through 2030, alongside a projection of more than $280bn in revenue by that year. That wasn’t the last word, either: by July, OpenAI had raised the figure again, to $750bn through 2030, alongside a new $20bn Georgia data-center campus (“Project Camellia”) drawing at least 3.2 gigawatts of power — a revision upward, not downward, reported by TechCrunch, citing the Wall Street Journal.

So the same $1.4 trillion that was once a liability — a compute bill large enough that OpenAI felt the need to publicly shrink it — has resurfaced eight months later as the number it wants investors to associate with its worth instead. That isn’t evidence of anything improper; valuations and spending commitments are different things measured in the same currency, and a coincidence of digits is just that. But it’s the kind of detail lost when a story is reported purely as “OpenAI raises money at eye-watering valuation,” and readers are better served holding both numbers in mind at once.

The revenue math, done properly

Several early write-ups framed the $30bn ask as roughly 35 times OpenAI’s annualised revenue, using a figure — revenue “passed $40bn” — that Bloomberg had reported back in August. That comparison is already stale. By 29 September, Axios reported, OpenAI’s revenue run-rate had climbed more than 70% since the start of the third quarter, to near $70bn, driven by enterprise sales roughly doubling since July and Q3 consumer-revenue additions alone exceeding all of 2025’s. That’s a sourced but unverified run-rate, not audited revenue, and OpenAI hasn’t confirmed it. Even so, the current figure puts the implied multiple at roughly 20 times run-rate rather than 35 — still rich, but a materially different story than the one still circulating.

Set that against where OpenAI started 2026: full-year 2025 revenue of about $20bn, more than triple 2024’s $6bn, by its own disclosure. Growth that fast is genuinely rare, and it’s the central argument OpenAI’s bankers will make to sovereign investors — though it’s no guarantee revenue keeps compounding once the easiest enterprise deals are signed.

“Accept some bad things happening”

The funding pitch has landed in an unusually bad week for OpenAI’s messaging on safety. Speaking on Politico’s Decoded podcast on 4 October, Altman was asked where his views diverged from Anthropic CEO Dario Amodei’s, and told Fortune’s reporting of the exchange: “We believe that the world should accept some bad things happening for the benefits of this technology and people having the agency.” He added that he “wouldn’t take a trade” that eliminated scams and misuse entirely, saying “I think people will do tremendously, orders of magnitude more, good stuff than bad stuff” — while still drawing a line at “the really catastrophic risks,” pointing to “a serious loss of control to AI” as an example of what should not be accepted.

The timing compounds the problem. Just a week earlier, on 29 September, Brockman had signed, as OpenAI’s president rather than its CEO, Trump’s voluntary AI safety accord alongside the chief executives of Google, Anthropic, Meta and Nvidia, and Elon Musk — a pact that, as CurratedBrief covered separately, carries no penalties, no public-audit requirement and no mandatory incident reporting. Senator Bernie Sanders responded directly, writing: “Let’s not wait to find out. Pause advanced AI NOW!” The remark has also fed into the FTC’s newly opened safety probe into both OpenAI and Anthropic, which is examining the gap between the industry’s voluntary commitments and what it actually discloses. Against a backdrop that already includes tens of thousands of logged AI security incidents across the two labs, and OpenAI’s own decision to shelve its GPT-6.1 Astra model over safety failures while pushing ahead with autonomous “Dots” agents, Altman’s framing reads less like a slip and more like a considered bet that candour about accepted risk is now cheaper, reputationally, than claiming a safety record it cannot fully back up.

Worth noting, too: Anthropic, which has built its identity around the opposite bet — stricter guardrails, and Amodei’s own call for an FAA-style AI regulator — is itself targeting an IPO valuation above $2 trillion on roughly a tenth of OpenAI’s current revenue. Investors are effectively being asked to pay a premium for either stance: OpenAI’s acceptance of bounded risk at speed, or Anthropic’s existential-risk disclosures in a securities filing. Both companies are, in effect, monetising their stated philosophy of risk.

Practical takeaways for builders and publishers

• Don’t assume unaudited run-rate figures signal stability. A jump from $40bn to near $70bn in a single quarter is dramatic, but it’s a sourced estimate, not a financial statement. If your business depends on OpenAI as a vendor, ask which number — run-rate, GAAP revenue, or compute-commitment total — is actually relevant to the claim being made.
• Treat “voluntary” safety commitments as exactly that. The Trump accord Brockman signed has no enforcement mechanism. If you build on frontier models and need assurances about incident disclosure or audit access, those commitments currently live in blog posts and conference remarks, not contracts.
• Expect compute and API pricing to stay firm. A company fixing a $1.4 trillion price on its own round, while carrying hundreds of billions in compute obligations, has little incentive to compete away margin on pricing soon.
• Watch for the next revision, not just the next headline number. OpenAI’s $1.4 trillion compute-spend figure was quietly cut to $600bn within a month, then raised again to $750bn five months later. Treat any single large AI-lab number as provisional until it survives more than a quarter or two of scrutiny.

What we still don’t know

• Whether the $30bn round closes at the $1.4 trillion price OpenAI has set, or whether a lead investor emerges and renegotiates the terms.
• Whether the near-$70bn revenue run-rate Axios reported will be confirmed in any audited disclosure, or how much of the Q3 jump reflects one-off deals rather than repeatable growth.
• Whether the $750bn compute-spending target — itself a July revision of the $600bn figure, which had replaced the original $1.4 trillion estimate just months earlier — will hold, given how quickly both prior figures moved.
• Where, precisely, Altman’s line between “bounded” and “catastrophic” risk actually sits — OpenAI has not published a framework defining it.
• Whether the FTC’s safety probe treats the gap between voluntary pact and enforceable rule as a finding, or simply notes it and moves on.

FAQ

Is OpenAI’s $1.4 trillion valuation confirmed?
No. It is reported by Bloomberg, attributed to people familiar with the discussions, and OpenAI has declined to comment. The round hasn’t closed and no lead investor is confirmed.

Did Sam Altman say AI should be allowed to cause harm?
Not quite. He said “the world should accept some bad things happening for the benefits of this technology,” while explicitly excluding “the really catastrophic risks” from that acceptance. Critics call the distinction vague; his wording does draw a line, even if its location is undefined.

How does this valuation compare with Anthropic’s?
Anthropic was valued at $965bn in May 2026 and is reportedly targeting more than $2 trillion at IPO, on roughly $4.6bn of 2025 revenue — a fraction of OpenAI’s. Anthropic’s higher relative multiple ties partly to its safety-focused positioning, which OpenAI’s remarks explicitly contrast with.

Why does the $1.4 trillion figure appearing twice matter?
It isn’t evidence of wrongdoing, but it’s a reminder that large AI-industry numbers get recycled and reattached to new contexts quickly. Check which “$1.4 trillion” a headline means before assuming it’s the same thing twice.

Sources

• Bloomberg: “OpenAI in Talks With UAE Funds, BlackRock for $30 Billion Round”
• Business Today Malaysia (Bloomberg-sourced): “OpenAI In Talks For US$30 Billion Round With UAE Funds, BlackRock”
• Fortune: “Sam Altman says the world should accept some ‘bad things’ happening with AI”
• Common Dreams: “Sanders Asks Altman What ‘Bad Things’ World Should Accept for AI”
• Roic.ai (citing Axios, 29 September): “OpenAI’s annualised revenue run-rate surges past 70% in Q3, nears $70bn”
• The Deep View: “OpenAI now at 2x Anthropic ARR, still bets on scaling”
• Benzinga (citing CNBC): “OpenAI Tempers Compute Spending Target To $600 Billion, Projects $280 Billion Revenue By 2030”
• TechCrunch (citing the Wall Street Journal): “OpenAI’s AI spending spree has ballooned to $750B”

Please follow and like us: