OpenAI closed the largest private funding round in history on March 31, 2026: $122 billion at an $852 billion valuation. Eight weeks later, Anthropic topped it, closing a $65 billion round at a $965 billion valuation and becoming the most valuable private AI company in the world. Now, four months after that, people familiar with the matter say OpenAI is back at the table asking for a valuation as high as $1.5 trillion.
That is the real story behind the number the AI industry keeps repeating this year. Not that OpenAI raised a record round. That nobody has held the record for long.
The Six-Month Scoreboard
Start with what actually happened, in order. Anthropic raised $30 billion at a $380 billion valuation on February 12. OpenAI answered with a round first reported at $110 billion in late February, which closed on March 31 at $122 billion and an 852billionvaluation,fundedbyAmazon(50 billion, with 35billioncontingentonanIPOoranundefined“AGI”milestone),Nvidia(30 billion) and SoftBank ($30 billion), earmarked for chips, data centers and talent. Anthropic answered back on May 28 with a $65 billion Series H at $965 billion, a deal that itself had been reported two weeks earlier at a $900 billion valuation before closing higher. The round was led by Altimeter, Dragoneer, Greenoaks and Sequoia, with Amazon, Micron, Samsung and SK Hynix also participating. Bloomberg and CNBC both described it as Anthropic overtaking OpenAI as the most valuable startup in the sector.
Since then, neither company has slowed down enough to let the other’s number stand. OpenAI is reportedly fielding offers around $1.2 trillion and pushing for $1.5 trillion, in talks first reported in mid-September and not yet confirmed by the company. Anthropic, meanwhile, filed confidentially for an IPO in June and is said to be targeting a listing valuation north of $2 trillion, based on projected 2028 revenue of $190 billion to $200 billion, when it goes public as early as this fall. Two companies, four valuation jumps, ten months, and if the pattern from May holds, whatever OpenAI’s round closes at is likely to land higher than the number now being reported.
What’s Actually Paying for This
The growth underneath these numbers is real, which is what makes the cycle defensible rather than absurd. OpenAI’s annualized revenue run rate topped $40 billion in August, roughly double where it stood a year earlier. Anthropic’s run rate crossed $47 billion in May, reached about $65 billion by July, and the company is projecting more than $120 billion by year end, growth Anthropic’s own investors have called the fastest of any company in history.
Growth like that is also precisely why so much capital keeps flowing in the same direction. Neither company has actually shown it can convert that revenue into profit at this scale. Both are spending well beyond what their revenue currently supports, on the bet that the growth curve keeps bending upward long enough to catch the spending.
The Ledger That Doesn’t Balance Yet
OpenAI’s spending side illustrates the gap. The company has signed roughly $1.4 trillion in infrastructure commitments and a further $1.15 trillion in hardware and cloud agreements running through 2035. Its own compute-spending guidance has moved twice this year: down to around $600 billion through 2030 in February, then back up to $750 billion in July as the company cited expanding cloud capacity needs. A number that swings by $150 billion in five months is not a forecast. It is a placeholder.
$40 billion in revenue against $1.4 trillion in signed commitments is not a balanced ledger, even accounting for growth. It is a bet that revenue keeps roughly doubling annually for years while a new, larger funding round arrives whenever the gap gets uncomfortable. So far, that bet has paid off on schedule, every few months, for both companies.
The Circularity Nobody Prices In
Look closely at who is writing these checks and the picture gets more complicated. Nvidia put $30 billion into OpenAI while remaining OpenAI’s largest chip supplier. Anthropic’s Series H pulled in strategic money from Amazon, Micron, Samsung and SK Hynix, three of which also sell Anthropic the memory and infrastructure it needs to run its models. In both cases, a meaningful share of the “external” capital fueling these valuations flows from companies that profit again when it’s spent. That doesn’t make the deals improper, strategic investors backing their biggest customers is a familiar pattern in enterprise technology, but it does mean a slice of the growth story on both sides rests on the same dollars changing hands twice.
A Race That’s Also a Fight for Customers
The valuations are being tested in the market too, not just in boardrooms. OpenRouter’s own spending data shows OpenAI overtaking Anthropic in combined developer dollar share for the first time since February 2024, a crossover that happened the week of September 7. Anthropic had held 75 to 80 percent of that spend through most of 2024 and 2025; it had slipped to the high 50s by mid-August before OpenAI’s GPT-5.6 lineup and its Astra model, launched in July and August, pulled the balance the other way. Anthropic answered with its own Claude Fable models aimed at agentic coding work, and Chinese labs including Moonshot AI and DeepSeek are narrowing the performance gap from a different direction, adding price pressure neither U.S. lab wants. Anthropic chief executive Dario Amodei has argued publicly for a more deliberate development pace on safety grounds, a stance critics say conveniently doubles as cover against rivals racing to ship faster.
What the Number Actually Signals
Zoom out and both companies are running the same experiment: raise at a price that assumes near-total dominance of a market that doesn’t fully exist yet, spend the proceeds on infrastructure fast enough to make that dominance real, and raise again before the previous round’s assumptions get tested by anyone outside the deal. Google alone is on pace to spend roughly $205 billion on AI infrastructure this year, and combined AI capital spending across the largest cloud and model providers is tracking toward roughly $700 billion in 2026. That is the scale of the bet the industry has placed on winner-take-most economics in foundation models, and OpenAI and Anthropic are simply the two companies pricing it most aggressively.
The bet might still pay off. Revenue at both companies is growing faster than almost any company in history has managed, and that is not a small thing to dismiss. But a valuation record that changes hands twice in under a year isn’t evidence of a settled market. It’s evidence of one still being priced in real time, deal by deal, with each round setting the floor the next one has to clear. The number that will matter in 2027, when both companies say they intend to be public, isn’t $852 billion, $965 billion or $1.5 trillion. It’s whether either company’s revenue curve closes the gap with its spending commitments before a private investor, or a public one, decides to stop financing the difference.
The post OpenAI Set the AI Valuation Record in March, Anthropic Broke It by May appeared first on DataFLOQ.
