Anthropic's $965 Billion Valuation Rests on a Revenue Curve That's 5 Months Old
Anthropic closed a $65 billion Series H at $965 billion, surpassing OpenAI for the first time, on an annualized revenue run rate that grew from $9 billion in December to $47 billion in May. Extrapolate the implied growth rate forward and Anthropic would generate more revenue than Apple by mid-2027. Meanwhile, Anthropic, OpenAI, and SpaceX are preparing to list simultaneously, and index providers are rewriting rules to accommodate them.
Forty-seven billion dollars. That is Anthropic's annualized revenue run rate as of May 2026, disclosed alongside a $65 billion Series H that values the company at $965 billion post-money, surpassing OpenAI's $852 billion for the first time. In December, the run rate was $9 billion; by February it had climbed to $14 billion, and by April it stood at $30 billion. Five months, five-fold.
The math behind $965 billion rests on a revenue curve younger than most sourdough starters, and if you extrapolate it forward using the growth rates Anthropic has actually demonstrated, you arrive at a number that cannot exist in physical reality. That is the first problem. The second: an accounting difference almost every headline has ignored means Anthropic may not actually be larger than OpenAI. The third: Anthropic, OpenAI, and SpaceX are all preparing to list on public markets this year, and combined, they could absorb more capital than every US venture-backed IPO of the past decade.
The Curve That Must Break
Here is the original math. Anthropic's monthly revenue trajectory, reconstructed from disclosed data points:
| Month | Monthly Revenue | Annualized | MoM Growth |
|---|---|---|---|
| Dec 2025 | ~$750M | $9B | — |
| Feb 2026 | ~$1.17B | $14B | ~24% |
| Apr 2026 | ~$2.5B | $30B | ~46% |
| May 2026 | ~$3.9B | $47B | ~56% |
The acceleration is the story. Revenue is not just growing. The growth rate itself is increasing, producing a convex curve that, extended, goes vertical. At a conservative 40% monthly growth rate — below the most recent observed rate — Anthropic's annualized revenue reaches $66 billion in June, $396 billion by December 2026, and $2.4 trillion by June 2027.
That last number — $2.4 trillion annualized by mid-2027 if the current monthly growth rate merely persists at 40% — would exceed Apple's entire 2025 revenue of $391 billion, a revenue base that Apple needed forty-eight years of operations to build.
The curve must break — it always does — and the question that determines whether $965 billion is cheap or absurd is where it breaks and at what level it stabilizes. If growth decelerates to 10% monthly — still aggressive by any normal standard but a brutal deceleration from current rates — the annualized run rate hits $83 billion in six months and $148 billion in twelve. HSBC analyst Stephen Bersey projects $241 billion by 2030, which implies roughly 4.5% monthly growth sustained over four years. That is ambitious but not insane, and it is a radically different trajectory than the one the current curve projects if you simply draw a line through the data.
Every dollar between the $47 billion trajectory and whatever terminal rate materializes is being priced into the valuation today, all at once, before the curve has told anyone where it flattens.
The Accounting Trick Nobody's Catching
Reuters Breakingviews flagged a distinction that most coverage has buried: Anthropic reports gross revenue, the full amount customers pay before distribution partners take their cut. OpenAI reports net revenue, after Microsoft's revenue share.
This distinction matters enormously, because if Anthropic's distribution partners take a standard 30% cloud reseller margin, net run rate drops from $47 billion to roughly $33 billion, and the revenue multiple shifts from 20.5x to 29.3x. More critically, the comparison dominating every headline — Anthropic's $47 billion versus OpenAI's $13.1 billion — overstates the gap.
Adjust for that. OpenAI's Q1 2026 net revenue: $5.7 billion. Anthropic's Q1 actual revenue: $4.8 billion gross, or roughly $3.4 billion net. On an apples-to-apples net-revenue basis, OpenAI may still be the larger company, and the "$47 billion vs. $13.1 billion" framing that has dominated every headline is not a lie — both numbers are real — but it compares gross oranges to net apples, and the juice tastes different.
Three IPOs, One Market
Anthropic is not the only near-trillion-dollar private company approaching public markets this year.
SpaceX, valued at approximately $1.5 trillion after absorbing xAI, is expected to list around June 12 in what would be the largest IPO in history, potentially raising $75–80 billion and doubling Saudi Aramco's record. OpenAI, at $852 billion after raising $122 billion, has an IPO filing expected within weeks, with a listing as early as September. Combined, the three companies represent roughly $3.3 trillion in private market capitalization.
That is not a typo — $3.3 trillion in private-market capitalization approaching public exchanges in a single calendar year.
PitchBook estimates their combined listings could absorb as much capital as all US venture-backed IPOs of the past decade, and Goldman Sachs partner John Flood noted on May 22 that mutual funds are already hoarding cash in anticipation of the largest capital absorption event public markets have ever faced. FTSE Russell rewrote its index rules to allow fast-track inclusion within five days of listing — a process that previously took weeks. Nasdaq did the same in March, and the S&P launched a public consultation on similar changes the following month.
Fast-track inclusion means passive index funds managing trillions must buy these stocks within days. The money comes from somewhere, which means selling existing large-cap holdings to make room. Forced selling, mechanically triggered, at scale.
The revenue multiples tell the story of how much risk is embedded in the pricing:
| Company | Market Cap | Revenue | Multiple |
|---|---|---|---|
| Apple | $3.6T | $391B | 9.2x |
| Microsoft | $3.4T | $262B | 13.0x |
| Nvidia | $3.2T | $130B | 24.6x |
| Amazon | $2.4T | $638B | 3.8x |
| SpaceX | $1.5T | $18.7B | 80.2x |
| Anthropic | $965B | $47B (gross) | 20.5x |
| OpenAI | $852B | $13.1B | 65.0x |
The three upcoming IPOs are priced for perfection. Any growth deceleration does not merely disappoint — it reprices.
Strongest Counterargument
The case for Anthropic's valuation is not speculative hand-waving — it is Claude Code, and the total addressable market it unlocks exceeds $500 billion.
Coding automation spans the combined value of developer tools, software services, and the enterprise productivity gains that companies report when deploying AI coding assistants. Organizations measuring 3–10x developer productivity improvements are not spinning a narrative; they are measuring output against baseline. If the productivity gains are real, the revenue is sustainable because the return on investment for enterprises is enormous, and churn stays low when the tool genuinely works.
This is not cryptocurrency speculation or dot-com vaporware. The relevant comparison is early Salesforce or early AWS — products whose growth looked unsustainable until the underlying market turned out to be larger than anyone projected. Amazon traded at what looked like an absurd revenue multiple for an entire decade, and every analyst who called it overvalued watched the company grow into the number while they waited for the correction that never came. Then it grew into it. Anthropic's investors are betting Claude Code compresses that timeline from decades to years, and the $36 billion Apollo/Blackstone debt deal to buy Google TPUs and Amazon's $25 billion cumulative commitment suggest the infrastructure providers share that bet.
That case deserves serious weight, and the sheer scale of infrastructure commitments backing it — $36 billion in debt, $25 billion from Amazon, $100 billion in AWS obligations — suggests the counterparties who know Anthropic's books best are placing the same bet.
What This Analysis Did Not Prove
"Annualized revenue" and "run rate" are self-reported by a private company with no auditor sign-off — these are marketing numbers, not GAAP. The 30% distribution margin used to estimate net revenue is based on standard cloud reseller terms; Anthropic's actual arrangements with Amazon and Google could differ significantly. The revenue breakdown by product — how much comes from Claude Code versus API versus consumer subscriptions — is not public, and the sustainability of the growth curve depends entirely on which product is driving it. SpaceX financials are visible only through limited IPO filing data, and the xAI absorption muddies pre-merger comparisons so thoroughly that even estimating a clean revenue-per-segment figure requires guessing at allocation methodologies that SpaceX has not disclosed. Finally, no model can predict how index rebalancing behaves at this scale because nothing at this scale has happened before.
What You Can Do
If you invest in index funds — roughly 60% of US equity assets are now passively managed — three trillion-dollar listings will force your fund to sell existing holdings to buy new ones. This is not a reason to sell, but it is a reason to expect unusual large-cap volatility around the listing dates: mid-June for SpaceX and potentially September for OpenAI.
If you are evaluating AI companies on revenue, always ask whether the reported figure is gross or net. The difference can exceed 30%, and financial media — Reuters, Bloomberg, the Wall Street Journal — routinely print whichever number the company provides in its press release without pausing to normalize for accounting methodology. When someone says "Anthropic is larger than OpenAI," ask: by which accounting?
If you hold Anthropic API contracts, build applications on Claude, or have enterprise agreements that depend on Anthropic's pricing stability, watch the $100 billion AWS commitment carefully. That deal locks Anthropic into AWS infrastructure for a decade, and any renegotiation of that relationship flows through to your API pricing — because a company spending $36 billion in debt financing on Google TPUs while simultaneously locked into a $100 billion AWS commitment is carrying infrastructure obligations that would make a telecom nervous.
The Bottom Line
Anthropic surpassed OpenAI in valuation on a revenue curve that is five months old, an accounting methodology that flatters the comparison, and an implied growth rate no company in the S&P 500 has ever sustained. The revenue might be real and the growth might continue, and Claude Code might genuinely represent a $500 billion market crystallizing in real time before anyone has built a durable competitive moat around the engineering workflows it is automating. But $965 billion is the price you pay when every one of those assumptions holds simultaneously, and the thing about assumptions stacked in series is that they do not merely add risk — they multiply it. Every last one.