Three months ago, Anthropic was valued at $380B. Today the number on the table is $900B. That is a 2.4x jump in 90 days, and the round closes in two weeks.
Anthropic is the AI lab behind Claude, currently raising a $50B round at a $900B valuation that would make it the most valuable AI startup ever. Bloomberg broke the story on April 29. CNBC confirmed the same day. By May 1, TechCrunch, PYMNTS, and Reuters had all piled on. The question stopped being "will it close" and became "what does the market look like the morning after."
I've been tracking Anthropic's cap table for two years and the velocity here is what I keep getting stuck on. Not the absolute number. The slope.
2.4x in 90 Days
Let me put the curve on the table, because the trajectory is the story.
| Date | Valuation | Step |
|---|---|---|
| Sep 2024 | $180B | — |
| Mar 2025 | $610B | 3.4x |
| Feb 2026 | $380B | 0.6x (correction) |
| May 2026 | $900B | 2.4x |
That February dip is the part most coverage misses. Anthropic actually got marked down during the Q1 correction, then more than doubled in a single quarter. SpaceX's fastest historical climb did not hit 2.4x in a quarter. There is essentially no precedent for this slope at this scale.
The investor pool tells you why. Existing backers like Google, Salesforce, and Spark Capital are reupping, and a wave of new institutionals are clamoring to get an allocation. A $50B round with 48-hour allocation windows is not a fundraise — it is a queue management problem.
The 48-Hour Math
Here is the deal mechanic that has term-sheet veterans raising eyebrows.
Anthropic's board is expected to approve in May. The target close is two weeks after that. Inside that window, investors who get an allocation have 48 hours to commit. Forty-eight hours to decide whether you want a slice of a $900B private company.
In normal markets, a round this size takes months to syndicate. Lawyers redline. LPs ask questions. Diligence drags. A 48-hour window means Anthropic is not negotiating — it is rationing. Demand exceeds the round size by a margin large enough that the company can dictate terms most founders would never get away with.
The structural bet behind that confidence is the revenue curve. Anthropic ended 2025 at roughly $9B ARR. By end of March 2026, they were at $30B. That is 3.3x in four months. Roughly 80% of it is enterprise, with over 1,000 customers spending $1M+ per year. This is not consumer subscription churn — it is committed annual contracts from companies whose procurement teams already justified the spend.
At $30B ARR, the implied multiple is about 30x. High, but not absurd at 200%+ growth. Whether "not absurd" survives the next two earnings cycles is the live question.
The Number That Actually Matters
Compare the two giants on the metric that determines whether $900B holds.
OpenAI ARR: ~$130B -> Valuation $852B (~6.5x)
Anthropic ARR: ~$30B -> Valuation $900B (~30x)
OpenAI has 4x the revenue, and Anthropic is now priced higher. That gap is not irrational on its own — it is a bet on the second derivative. If Anthropic's growth rate persists for another year, the multiples converge. If it decelerates, the gap looks like a pricing error.
The composition of the gap also matters. OpenAI built breadth — hundreds of millions of consumer users, ChatGPT as a brand, an API platform. Anthropic built depth — fewer logos, bigger contracts, a "safe AI" wrapper that lets enterprise procurement teams sign without writing a memo defending the choice. In 2026, the second story is the one investors are paying for.
There is also a structural advantage I keep underweighting. Amazon has committed $25B and 5GW of compute. Google has committed $40B. Both clouds host Claude. Unlike OpenAI's deep dependence on Microsoft, Anthropic plays both sides of the cloud duopoly without locking into either. That optionality is worth a premium, and the term sheet reflects it.
Fortune ran an analysis the same week claiming roughly half of Google's and Amazon's AI-related profits trace back to Anthropic stake appreciation. Read in reverse: big tech AI profitability now depends materially on Anthropic's mark going up. That is a flywheel — or a feedback loop, depending on which side of the cycle you think we are on.
The Pentagon Paradox
The same week the $900B headlines hit, the Pentagon excluded Anthropic from its AI contract shortlist. Most companies would call that a bad week. For Anthropic, the two stories are arguably the same story.
Anthropic has held a cautious posture on military AI since founding. That clashes with Pentagon procurement. But it is exactly the posture that lets a bank, a hospital, or a law firm justify the contract to a risk committee. "This is the AI lab that turned down Pentagon money because it takes safety seriously" is a powerful procurement narrative in regulated verticals — and reports indicate those verticals are Anthropic's fastest-growing segments.
Reddit summarized it as "Pentagon snub, cap table revenge." That is glib, but directionally correct. The snub and the $900B are two faces of the same brand strategy.
Bubble or Breakout?
If $900B feels like a stretch, the instinct is not unfounded. The bear case writes itself: 30x revenue, 200% growth that has to persist, open-source models like Llama and DeepSeek closing the gap on price-per-token every quarter. Pre-2000 dot-com peaks looked structurally similar.
The difference, and it is a real one, is that Anthropic has $30B in ARR with 80% recurring. This is not a dream with no revenue. The bubble, if it is one, is not empty. That does not eliminate risk — it just reframes it from "is there a business" to "is the growth rate priced correctly."
What this means depends on where you sit. If you are a founder fundraising right now, the "AI valuations are still climbing" narrative works in your favor, but investor attention is concentrating at the very top of the stack — your differentiation has to be sharper than it was in 2024. If you are a developer building on Claude, expect API price drops as Anthropic deploys this capital, and revisit your platform dependency before the next major release locks in your stack. If you are an investor staring at a $900B entry, the only metric that matters is whether ARR growth holds for two more quarters; any deceleration breaks the multiple. And if you are inside a big tech AI org, this round is the starting gun for "phase two" — the axis is now which frontier model your cloud can offer with structural advantage, not whether you have one of your own.
The IPO timeline is the next domino. Market chatter places the earliest window at October 2026, with H1 2027 as the realistic case. Dario Amodei has said there is "no rush." That can mean two things: the company needs maturity, or when you can raise $50B privately at $900B, IPO urgency simply evaporates. Both are probably true.
A $900B private valuation isn't a fundraise — it's a bet that the AI revenue curve doesn't bend before the IPO does.
If you had a 48-hour allocation window at $900B and $30B ARR, would you write the check?
Full Korean analysis on spoonai.me.
Sources:
Top comments (0)