Anthropic is now valued at $380 billion after closing a $30 billion Series G on February 12, 2026, according to the company’s own announcement. That is a staggering number in any market. It is even more striking when you put it next to Anthropic’s $61.5 billion Series E valuation in March 2025 and its $183 billion Series F valuation in September 2025. In roughly twelve months, the company has gone from $61.5 billion to $380 billion. That is not a typo. That is a 6x re-rating.
If you want the cleanest possible summary of the AI market right now, there it is: capital is not just rewarding growth. It is rewarding speed, scale, and strategic control. Anthropic’s announcement says the round was led by GIC and Coatue, with D. E. Shaw Ventures, Dragoneer, Founders Fund, ICONIQ, and MGX co-leading, plus a long list of heavyweight backers. AP’s coverage framed the deal as one of the most valuable private startup financings on record. That tracks.
But the more interesting question is not whether the number is big. Of course it is. The real question is: what exactly is the market buying at this price?
A valuation is a theory, not a fact
Valuations often get discussed like they are measurements. They are not. They are theories about the future, compressed into a single number and wrapped in a wire transfer.
In Anthropic’s case, the theory appears to be something like this:
- enterprise adoption is real and accelerating,
- Claude Code is becoming an actual revenue engine, not just a clever demo,
- frontier model quality remains scarce,
- and the company can keep converting product momentum into durable commercial demand.
Anthropic says its run-rate revenue is now $14 billion, up more than 10x annually in each of the past three years. It also says more than 500 customers now spend over $1 million annually, and that eight of the Fortune 10 are Claude customers. Those are not hobby numbers. That is a serious enterprise footprint.
And the market is clearly pricing that footprint aggressively.
At a rough back-of-the-envelope level, a $380 billion valuation against $14 billion of run-rate revenue implies something like 27x revenue. That is a very rich multiple, even by 2026 AI standards. But if the market believes revenue can keep compounding at this pace, the multiple stops looking outrageous and starts looking like a time-horizon problem. Same math, different religion.
Why the jump matters
The jump from $183 billion in September 2025 to $380 billion in February 2026 is telling on its own. That is about a 2.1x increase in five months. There are not many private companies on earth that can move like that without public markets, and fewer still with business models that are not obviously monetizing a temporary mania.
Anthropic’s own disclosures help explain the enthusiasm. In the same announcement, the company says Claude Code has reached $2.5 billion in run-rate revenue and that weekly active users have doubled since January 1. It also says Claude Code business subscriptions have quadrupled since the start of 2026.
That matters because it shifts the company’s story from “promising model lab” to “distribution plus workflow plus monetization.” Investors are not just buying model intelligence. They are buying workflow capture.
And in AI, workflow capture is the prize. If the model becomes the place where code is written, analysis is done, documents are drafted, and decisions are mediated, then the product stops being a feature. It becomes infrastructure. That is where pricing power lives. That is where switching costs begin to harden. That is where people start saying “platform” and “moat” with a straight face.
The hidden costs of frontier AI
Here is the part that gets less attention during the victory lap: frontier AI is expensive, operationally intense, and strategically fragile.
You do not get to scale to these valuations on software margins alone. You need:
- massive compute access,
- durable cloud relationships,
- excellent security and governance,
- disciplined data rights,
- and enough enterprise trust that large customers will actually standardize on your stack.
Anthropic says the Series G will fund frontier research, product development, and infrastructure expansion. That is a useful reminder that the valuation is partly a bet on capital intensity. The company is not just selling software. It is building and financing a very expensive industrial system around software.
That is why AI diligence is different from generic SaaS diligence. If you are underwriting an AI company, you are not just asking whether the top line is growing. You are asking whether the business can survive the physics of its own cost structure.
Can it keep models competitive without torching economics?
Can it defend its data and IP position?
Can it prove the training data chain is clean enough for enterprise buyers?
Can it document governance, security, and compliance well enough for procurement teams that are getting smarter by the quarter?
If the answer to those questions is “sort of,” the multiple is already ahead of the business.
What investors should be asking now
When we do technology due diligence and valuation work, this is where the conversation gets interesting. The pitch deck usually shows the TAM. The diligence process shows the friction.
For an AI company at this scale, the questions should include:
- How concentrated is revenue in a few large accounts?
- How durable is the usage expansion after the first use case lands?
- What is the model dependency risk if a key provider changes pricing or terms?
- Are privacy, copyright, and data provenance issues documented well enough for enterprise review?
- Does the company have the governance posture to support large regulated customers?
- Is the valuation justified by repeatable economics, or by the market’s current willingness to pay for frontier scarcity?
That last question is the one people tend to avoid, because it is uncomfortable. Also because it is usually the most important.
A company can be exceptional and overvalued at the same time. In fact, that combination is more common than investors like to admit. The market can be right about the company’s importance and wrong about the exact price. Both things can be true. Finance loves nuance until the round closes, and then everyone develops a very confident tone.
What this says about AI markets
Anthropic’s $380 billion mark tells us a few things about the current AI market:
- Capital is still willing to pay up for leaders.
- Enterprise adoption matters more than consumer hype.
- Coding and agentic workflows are becoming core value centers.
- Infrastructure, governance, and distribution are now valuation inputs, not afterthoughts.
- “AI company” is no longer a useful category by itself.
That last point is especially important. Not every company with an LLM wrapper deserves a frontier multiple. Not every “AI platform” has durable economics. Some are real businesses. Some are beautifully formatted aspirations. The market is getting better at telling them apart, but not fast enough to be comfortable.
So what should founders, boards, and investors take from Anthropic’s round?
Not that every AI company should be worth more.
Not that revenue growth excuses everything.
Not that scale solves governance.
The smarter takeaway is simpler: the market is rewarding companies that can turn technical advantage into enterprise habit. If you cannot do that, your valuation will eventually meet gravity.
And gravity, unlike hype, does not need a press release.
That is why AI valuation work cannot stop at the headline number. It has to look at the plumbing: the contracts, the security posture, the data rights, the model dependencies, the customer concentration, and the actual quality of the revenue. Otherwise you are just staring at a very expensive opinion and pretending it is a balance sheet.
At licens.io, that is the kind of question we like to pressure-test. Because in AI markets, the difference between “extraordinary business” and “extraordinary pricing” is where the real diligence starts.
Related posts
FASB Rewrites Software Cost Accounting: ASU 2025-06 and What CFOs Need to Know
FASB’s ASU 2025-06 replaces the old stage-based software capitalization playbook with a single recognition test, forcing CFOs to rethink policy, controls, and valuation.
Read moreMeta's $14B Scale AI Deal: The Biggest AI Acqui-Hire in History
Meta’s $14.3 billion Scale AI transaction is a minority-stake deal that looks a lot like an acqui-hire, with major implications for AI valuation and control.
Read moreGoogle Buys Wiz for $32B: The Largest Cybersecurity Acquisition in History
Google’s $32 billion agreement to acquire Wiz is a landmark cybersecurity deal, and it says plenty about cloud security, antitrust risk, and valuation discipline.
Read more