Anthropic, the company behind Claude, is exploring a new round of financing that would value it at more than 900,000 million dollars. If it closes, it would surpass OpenAI as the world’s most valuable AI startup. Altman’s company set its needle at 862 million last month.
The figure more than doubles the 350,000 million it had in February. In just two months.
Why is it important. The valuation no longer reflects Anthropic’s sales. It responds to a bet on what the company can become in five years or a decade: a provider of something resembling an essential service.
Anthropic bills Claude for subscriptions and accesses to its API. That business exists, grows quickly and has reasonable margins. But it does not by itself justify a valuation that is close to that of Samsung, the Korean megalodon that manufactures everything from the chips we carry in our pockets to the ships that cross the ocean.
The context. What the market is buying with Anthropic, and as often happens in the stock market, is not the present, but a hypothesis: that a very small handful of laboratories will control the foundational layer on which the software of the next decade will be built. And that Anthropic will be one of those few. And it will do so in a very profitable way.
The logic, on the other hand, is the same that led to overvaluing telecos during the bubble dotcom or to the electric companies at the beginning of electrification. Whoever owns the basic infrastructure sets the rules.
- Google has already committed 10 billion to the previous valuation, with another 30 billion conditional on objectives.
- Amazon has put in 5 billion and plans to inject 20,000 more.
- An IPO could come before the end of the year, around October.
Between the lines. That Google and Amazon, two of the largest cloud companies in the world along with Microsoft, finance a company that also sells through them says a lot about how they understand the moment. They are ensuring supply, it is not just an investment in a supplier.
It is the difference between buying shares in an oil company and buying a field. Anthropic is, for these hyperscalersa deposit.
Yes, but. The hypothesis has its cracks. The models are commoditizing faster than it seemed a year ago. The technical difference between Claude, ChatGPT and Gemini It is measured in nuances, not in generational leaps.
If foundational AI ends up being a commodity (something like electricity or water coming out of the tap), current valuations are unsustainable. If it ends up being an infrastructure with network effects and high barriers to entry (something like an operating system), they may even fall short. The market is paying for the second hypothesis. Time will tell.
The money trail. Anthropic recently announced, with restrained fanfare, Mythosa model capable of detecting and exploiting vulnerabilities in critical software. The company deemed it “too dangerous” to release and has only given it to a closed group of companies for internal testing. Even so, it has been accessed by unauthorized users.
That is exactly the reason why some investors pay these figures: such a model is not sold but granted. And whoever decides to whom it is granted has regulatory power de facto that not even a Samsung, at least outside of South Korea, has ever had.
The big question. What happens if the bet goes wrong? A valuation of 900 billion means that Anthropic has to generate, at some reasonable point, revenues in the order of tens of billions a year with very high margins. It is possible. But it was also important for Cisco to maintain its 2000 valuation, and it has needed 26 years to tie.
The difference is that this time the buyers of the bet are the companies themselves that depend on the result. This reduces the risk of a sharp correction. And he postpones it.
In Xataka | There is a thing called “Ornn price index”, it is out of control and it is bad news for everyone
Featured image | Xataka
GIPHY App Key not set. Please check settings