How much is OpenAI worth today?

What is OpenAI’s latest official valuation?

OpenAI’s latest official valuation is still $852B. The $1.4T figure is a proposed price that no investor has paid yet.

OpenAI announced on March 31, 2026, that it closed a $122B round at an $852B post-money valuation. The money kept arriving after the announcement: Nvidia and SoftBank reportedly paid their final $10B installments each on October 1, 2026. In August 2026, current and former employees reportedly sold about $7B of shares at the same $852B price, according to Bloomberg and CNBC.

The new number is higher. Bloomberg reported on September 29, 2026, that OpenAI was seeking at least $30B at about $1.4T before the new money. On October 6, it reported that Abu Dhabi’s MGX and other UAE funds had discussed investing up to $10B combined, with BlackRock and existing backers such as Thrive Capital and Andreessen Horowitz also in talks. No lead investor had been named, and the details could still change.

The path is now $157B in October 2024, $300B in March 2025, $500B in an October 2025 secondary sale, and $852B in March 2026. If the new round closes at $1.4T, the value will have risen about 9x in two years.

What this means: the $852B mark is a real price, confirmed twice. But it was set when OpenAI’s run-rate was about half what it is now. The live question is where between $852B and $1.4T fair value sits.

Is OpenAI already worth $1T today?

Probably yes, on revenue alone. But no completed transaction has confirmed it.

The strongest argument is revenue growth. OpenAI said in March 2026 that it made about $2B per month, or about $24B a year. By the end of September 2026, its annualized revenue was close to $50B, according to the Financial Times and Reuters. Earlier reports of nearly $70B counted sales through cloud partners on a different basis. OpenAI now reportedly presents $70B as a year-end target, not an achieved figure.

At ~$50B of run-rate, $1T is about 20x revenue. In March 2026, the same $1T meant about 42x. The headline number has not moved; the revenue under it has doubled.

The second argument is reach. OpenAI reported more than 1B weekly ChatGPT users in mid-August 2026 and more than 1.2B by its DevDay on September 29, 2026, with 2.5M businesses using its products. It also reported more than 50M paying consumer subscribers in March 2026; we found no newer count.

The third argument is new revenue. OpenAI said on August 31, 2026, that ChatGPT Ads reached $1B of annualized revenue in less than 200 days. Ads turn free users, the vast majority of ChatGPT, into a revenue base.

The caution is that run-rate is not revenue. OpenAI’s reported recognized revenue was $5.7B in Q1 2026 and $6.7B in Q2, and the Financial Times reported a forecast of $36B for the full year. Against that forecast, $1T is about 28x 2026 revenue.

So yes, OpenAI can credibly be worth $1T today. But until a round closes or an IPO prices, $1T is a well-supported estimate, not a market fact.

If you want more recent data on this point, see the latest figures in our Frontier AI Labs report ↓.

Could OpenAI be worth less than $700B now?

Very unlikely, unless something breaks.

The first reason is that real money keeps paying $852B. The March round was funded through October 1, 2026, and employees sold about $7B of stock at that price in August, according to reports.

The second reason is revenue. At ~$50B of run-rate, $700B would be about 14x revenue. That is below Nvidia, which traded at about 17x sales on October 20, 2026, for a company whose run-rate roughly doubled in six months.

The third reason is peer pricing. Anthropic closed a round at $965B in May 2026, and its IPO is reportedly being discussed at up to $2T, with some investor estimates closer to $1.5T. A sub-$700B OpenAI would need a strong company-specific reason to trade at well under half of its closest rival.

Bear signals do exist. OpenAI’s computing-capacity chief, Sachin Katti, was reported on October 15, 2026, to be leaving in November, amid concerns about Stargate execution. At least 47 major US data-center projects were reportedly stalled or canceled in 2026. At least eight US states sued OpenAI over alleged ChatGPT harms in October 2026. And tech and chip stocks reportedly dipped in early October 2026 as OpenAI’s revised revenue figures circulated, a sign of how much the market leans on OpenAI’s numbers.

What this means: a sub-$700B value would need a failed fundraising, a sharp revenue slowdown, or a broad AI selloff. None of those has happened. $700B looks too low.

What does OpenAI’s ~$50B revenue say it is worth?

Revenue multiples now put OpenAI’s fair value roughly between $850B and $1.4T.

Using the close-to-$50B annualized revenue reported for September 2026, the math is straightforward.

Revenue multiple	Implied OpenAI valuation
14x revenue	$700B
17x revenue	$850B
20x revenue	$1.00T
22x revenue	$1.10T
24x revenue	$1.20T
26x revenue	$1.30T
28x revenue	$1.40T

This table shows how far the picture has moved. In March 2026, the $852B mark sat at about 35x annualized revenue. In October 2026, it sits at about 17x. Even the proposed $1.4T, at about 28x, is cheaper than the $852B mark looked in March.

Peers point the same way. Anthropic’s $965B round implies less than 15x its reported July run-rate of more than $65B. A $2T Anthropic IPO would mean about 30x its Q3 2026 revenue annualized. OpenAI at 20x to 26x sits between the two.

The weak point is revenue quality. OpenAI’s reported Q1 2026 gross margin was about 39%, up from about 33% a year earlier but far from software levels. And the run-rate flatters the picture: recognized revenue was $6.7B in Q2 2026.

What this means: revenue alone supports a center around $1.0T to $1.2T. Getting to $1.4T requires investors to believe the reported $70B year-end target.

If you want more recent data on this point, see the latest figures in our Frontier AI Labs report ↓.

What would OpenAI need to justify a $1.4T valuation?

OpenAI needs its current growth to hold through year-end and its margins to keep improving.

At ~$50B of run-rate, $1.4T is about 28x revenue. That is a premium multiple, but not an extreme one next to CrowdStrike at about 43x or Palantir at about 76x in October 2026.

There are three ways OpenAI could get there.

First, revenue could hit the target. OpenAI has reportedly told investors it expects a $70B run-rate by December 2026. That is a target, not an achieved figure. At $70B, $1.4T would be only 20x revenue, and the price would look fair.

Second, the mix could improve. Enterprise was more than 40% of revenue in March 2026, with parity with consumer targeted by year-end. Ads reached a $1B run-rate by August. In June 2026, Codex produced 64% of the output tokens across Codex and ChatGPT in OpenAI’s enterprise data. More enterprise and agent revenue would make sales less dependent on consumer subscriptions.

Third, the market could keep treating OpenAI as scarce. That argument has weakened in 2026. Anthropic’s planned IPO and SpaceX’s roughly $2.0T market value give investors other ways to own frontier AI.

The hardest part is cost. The WSJ reported that OpenAI has discussed roughly $600B to $750B of compute spending through 2030. The Financial Times reported a projection of about $280B of cumulative negative free cash flow over the same period.

What this means: $1.4T is a bet on the next 12 months, not a price for today’s business. The December 2026 run-rate will be the clearest test.

When will OpenAI go public, and at what valuation?

Not before 2027, and a rational IPO range on today’s numbers would be about $1.0T to $1.3T.

The timing has moved further away. OpenAI said on June 8, 2026, that it had confidentially submitted a draft S-1, but that “it may be a while.” In a Fortune interview published on September 12, 2026, Sam Altman ruled out a 2026 IPO and called this “an ill-advised moment” to list, citing the AI safety debate. Reuters reported on October 19, 2026, that a listing is expected in 2027, though that is not guaranteed. No offer price or bank syndicate has been found.

That choice has a price logic. A private round at $1.4T lets OpenAI raise money without publishing audited margins, losses, and compute commitments. Anthropic is showing what those disclosures look like.

Anthropic will set the reference point first. Its IPO is reportedly planned for November 2026, after the US midterm elections, with Goldman Sachs and Morgan Stanley reportedly leading. Its draft prospectus reportedly shows more than $600B of noncancelable infrastructure commitments and an $80.7B net loss for the first half of 2026, mostly from financing-related accounting. How public investors price those numbers will shape any OpenAI listing.

On today’s run-rate, a $1.0T IPO would be about 20x revenue: above the $852B mark and easy to defend. A $1.3T IPO would be about 26x. Matching the $1.4T private ask in public markets would likely need the $70B target to be met or a strong reception for Anthropic.

If you want more recent data on this point, see the latest figures in our Frontier AI Labs report ↓.

Is Anthropic now worth more than OpenAI?

On last-priced valuation, yes. On reach, no.

On valuation, Anthropic’s May 28, 2026 Series H raised $65B at a $965B post-money valuation, above OpenAI’s $852B. Its IPO is reportedly being discussed at up to about $2T. OpenAI’s own next mark, about $1.4T, is still only an asking price.

On revenue, Anthropic also appears ahead. It reported $11.6B of Q2 2026 revenue against OpenAI’s reported $6.7B, then told investors in October 2026 that Q3 revenue reached $16.8B, up about 45%. It also reportedly posted an 11% adjusted operating margin and positive free cash flow in Q3. Caveat: OpenAI reportedly books partner sales net of cloud-provider shares, while Anthropic uses a gross approach for some deals. The gap is real but not exactly measurable.

Anthropic’s growth is slowing, though. Q3 growth of about 45% followed about 147% in Q2, and some forecasts for its year-end run-rate have been cut from about $110B toward about $74B.

On coding, Anthropic leads. In a JetBrains survey run from May to July 2026, 39% of professional developers used Claude Code at work, against 16% for Codex. Codex use grew about fivefold from January, but it still trails.

OpenAI keeps a much broader base. ChatGPT had more than 1.2B weekly users by the fall of 2026. Anthropic has not disclosed absolute Claude user numbers. Similarweb estimates ChatGPT had 57% of visits to major AI chatbot sites in August 2026, against 9.6% for Claude, though ChatGPT’s share was 78% a year earlier.

What this means: OpenAI deserves a premium over Anthropic only if investors value consumer reach and ads more than enterprise and coding depth. Right now, they do not. Anthropic’s IPO will be the first public test of that judgment.

How does OpenAI compare with xAI now that SpaceX is public?

xAI no longer has its own price. It is part of SpaceX, which gives investors a listed way to buy a frontier lab.

xAI raised $20B in January 2026. In February 2026, it combined with SpaceX in a deal that reportedly valued xAI at about $250B within a $1.25T total. SpaceX then listed in June 2026 at $135 per share under the ticker SPCX. On October 20, 2026, its market value was about $2.0T, with shares near $148.

That does not make xAI a clean comp for OpenAI. SpaceX’s value mixes rockets, Starlink, and AI. But it shows that public investors will pay trillion-dollar prices for an AI story tied to compute and distribution.

On product, OpenAI is clearly ahead. On Arena’s overall text leaderboard on October 20, 2026, xAI’s best listed Grok model ranked 39th. xAI’s strengths are compute, with Epoch AI estimating about 1.1M H100-equivalents at Colossus 2, and X distribution.

The wider field also cuts OpenAI’s scarcity premium. Moonshot AI reached a private valuation of about $50B in October 2026, according to Bloomberg. Mistral raised €3B at more than €21B on September 8, 2026, and launched Large 4 on October 6. Meta released Muse Wonder on October 12 and the open-weight Muse Spark Open on October 15. Google made Gemini 4 Argon broadly available on October 13.

One Musk risk has eased: his claims against OpenAI were dismissed after a May 2026 trial, though an appeal was reportedly planned.

What this means: OpenAI still looks more financially proven than xAI. But OpenAI is no longer the only trillion-dollar way to own frontier AI.

If you want more recent data on this point, see the latest figures in our Frontier AI Labs report ↓.

How does OpenAI compare with public AI stocks?

At $852B, OpenAI is no longer expensive by public AI standards. At $1.4T, it would still be in premium territory.

At ~$50B of run-rate, $852B is about 17x revenue. On October 20, 2026, Nvidia traded at about 17x sales, Snowflake at about 21x, Datadog at about 23x, CrowdStrike at about 43x, and Palantir at about 76x. CoreWeave, which runs capital-heavy AI infrastructure, traded at about 4.5x.

So the last OpenAI price sits at Nvidia’s level, and the proposed $1.4T, at about 28x, would sit between Datadog and CrowdStrike, far below Palantir.

The comparison is imperfect. Public multiples use trailing sales; OpenAI’s uses a run-rate, which flatters it. Nvidia is highly profitable, and OpenAI is not. CoreWeave is the warning: when revenue depends on heavy compute spending, the market pays a low multiple.

What this means: if OpenAI is priced like a high-growth AI platform, $1.0T to $1.3T works. If it is priced like compute-heavy infrastructure, fair value falls toward the $852B mark. Only a Palantir-style premium takes it clearly past $1.4T.

Does OpenAI’s profitability make the valuation fragile?

Yes. Profitability is the main reason not to value OpenAI far above $1.3T today.

Revenue is not the problem. The cost structure is. Frontier AI companies spend heavily on training, inference, chips, data centers, and talent.

The first signal is losses. For Q1 2026, The Information reported about $3.7B of cash burn, an operating loss of about $9.3B including stock-based compensation, and a net loss above $21B, of which about $12.4B came from convertible-security accounting. These are three different measures; the net loss is not cash spent.

The second signal is margins. The reported Q1 2026 gross margin was about 39%, up from about 33% a year earlier. Inference costs were about $3.5B and R&D about $8.6B. Margins are improving, but they are far below software levels.

The third signal is the long-term plan. The Financial Times reported projections of about $350B of revenue in 2030 and about $280B of cumulative negative free cash flow through 2030. No break-even year has been verified. OpenAI reportedly had about $73B of cash at the end of March 2026; a Q3 balance was not found.

The fourth signal is Microsoft. Under an April 27, 2026 amendment, Microsoft’s license became nonexclusive through 2032 and OpenAI can sell across clouds. But OpenAI still pays Microsoft a revenue share until 2030, subject to a cap, reported at around 20%.

The fifth signal is price pressure. Google launched Gemini 4 Argon in October 2026 at $2 per million input tokens and $10 per million output tokens; OpenAI’s GPT-6 Astra is listed at $10 and $50. DeepSeek’s cheapest tier is far lower. If prices keep falling, OpenAI must keep spending while charging less per unit.

The contrast with Anthropic matters here. Anthropic reportedly earned about $1.70 of revenue per $1 of compute in Q3 2026 and posted an 11% adjusted operating margin. OpenAI has not shown a comparable figure.

What this means: profitability does not kill the valuation, but it caps it. Until OpenAI shows a path to positive free cash flow, $1.0T to $1.3T is easier to defend than $1.4T.

If you want more recent data on this point, see the latest figures in our Frontier AI Labs report ↓.

Is OpenAI’s market big enough for a trillion-dollar valuation?

The market is big enough, but OpenAI will not capture the whole AI market.

Gartner forecast worldwide generative AI spending of $644B in 2025, up 76.4% from 2024. But much of that is hardware, devices, and infrastructure. That money flows to Nvidia, cloud providers, and data-center operators, not to OpenAI.

The better signal is enterprise software. Menlo Ventures estimated enterprise generative AI spending at $37B in 2025, up from $11.5B in 2024. In the same research, Menlo estimated that Anthropic took 40% of enterprise LLM API spending in 2025, against 27% for OpenAI and 21% for Google. OpenAI’s share had been 50% in 2023.

Buyers are also spreading their bets. A September 2026 Stacklok survey of North American enterprises found that about 92% use more than one model provider. McKinsey’s 2025 survey found that nearly two-thirds of organizations had not yet scaled AI across the enterprise. Budgets are growing, but they are being split.

OpenAI’s own plan shows the scale it needs. The Financial Times reported a projection of about $350B of revenue in 2030. That would require OpenAI to win a large share of applications, agents, APIs, enterprise workflows, and advertising.

What this means: the market can support a trillion-dollar OpenAI, but only if OpenAI wins in its own layers. It does not work if most profit flows to chips, clouds, and rival labs.

Does OpenAI still have a moat without the best model?

Yes, but it is a distribution moat, not a technology moat.

OpenAI no longer leads the main model rankings. On October 20, 2026, Google’s Gemini 4 Argon led Arena’s overall text leaderboard on a preliminary basis, followed by Meta’s Muse Wonder and Anthropic’s Claude Opus 5.5. Anthropic led the web-development and agent leaderboards, with OpenAI’s GPT-6 Astra second. Artificial Analysis ranked Anthropic first and put GPT-6 Astra level with Gemini 4 Argon.

The first moat is distribution. ChatGPT passed 1.2B weekly users in the fall of 2026. No other AI lab has matched that reach. But the share is eroding: Similarweb estimates ChatGPT’s share of visits to major chatbot sites fell from 78% to 57% in the 12 months to August 2026.

The second moat is monetizing that audience. Ads reached a $1B run-rate in under 200 days. OpenAI reported 2.5M businesses and 35M weekly users of ChatGPT Work and Codex at its September 2026 DevDay.

The third moat is integration. OpenAI is tying ChatGPT, Codex, APIs, memory, search, ads, and partner deals such as its October 2026 Atlassian expansion into one system. If that system becomes the default interface for work, the moat holds.

The counter-moat is switching. Open-weight models from DeepSeek, Moonshot, Z.ai, Meta, and Reflection AI keep getting closer, and most enterprises already use several providers.

What this means: OpenAI’s moat is real but narrower than its user count suggests. The 2026 moat is distribution plus ads plus enterprise embedding plus compute. It is a valuable business moat, but it no longer rests on having the best model.

So how much is OpenAI worth today?

OpenAI is probably worth $950B to $1.3T today, with a center around $1.1T.

The lower end starts from the $852B price, which now looks stale. Since March 2026, OpenAI’s run-rate has roughly doubled to close to $50B, ChatGPT has passed 1.2B weekly users, and ads have become a $1B run-rate business. At $950B, OpenAI would trade at about 19x run-rate revenue, close to Nvidia.

The upper end is capped by profitability and by the gap between run-rate and recognized revenue. At $1.3T, OpenAI would trade at about 26x run-rate and about 36x the $36B of 2026 revenue it reportedly forecasts.

Below $700B looks too bearish. It would mean about 14x run-rate and a deep discount to Anthropic, with no clear reason.

$1.4T and above is not yet proven. That is what OpenAI is reportedly asking private investors for, but the round has not closed yet. Anthropic’s planned November IPO will give the first public signal of how far investors will go.

So the clean answer is this: OpenAI is worth around $1.1T today, with a range of $950B to $1.3T. If the $1.4T round closes and Anthropic’s IPO prices strongly, fair value moves toward $1.4T. If public investors punish AI burn, it could fall back toward $852B.

So, our conclusion: $852B is the last completed price, but it no longer describes OpenAI’s business. The evidence points to a company worth somewhat more than $1T. But OpenAI is no longer the most valuable lab on last-priced marks, and it has not yet earned a confirmed $1.4T.

To get there, it needs one more proof point: not a bigger fundraising target, but evidence that ~$50B of run-rate can turn into durable, high-margin cash flow.