Anthropic: $350 миллиардов за «безопасный» ИИ и право догнать OpenAI

If you thought the financial madness in the world of artificial intelligence had reached its peak after OpenAI's recent round, Anthropic has bad news for you. Or good news, if you're sitting on piles of venture capital. According to information from internal sources being relayed by Chinese media, the company of the Amodei siblings plans to close a $20 billion deal as early as next week.

But the most interesting part here isn't even the amount of funding, but the resulting valuation. We're talking about $350 billion. To put that in perspective: it makes the startup more expensive than most companies on the S&P 500 list and puts it in the same league as tech giants that have spent decades building their empires.

Anthropic's story has always been about a "different path." They left OpenAI in 2021, slamming the door loudly over disagreements on safety issues. Back then, Dario and Daniela Amodei claimed that commercialization was killing ethics.

Now the irony of the situation is that their "ethical" approach requires checks that Sam Altman never even dreamed of. It seems that the concept of "Constitutional AI" is a very expensive privilege that investors are willing to pay any amount of money for, just to have an alternative to the ubiquitous GPT. Why do they need so much money right now?

The answer is simple and mundane: computational power. We've entered an era where every new iteration of a large language model requires exponentially more H100 and Blackwell chips. Anthropic doesn't just need to maintain the current Claude 3.

5 Sonnet, which is already breathing down the necks of market leaders, but to prepare a full leap to Claude 4 or even Claude 5. In this business, you either spend billions on electricity and silicon, or you turn into a historical footnote about how someone tried to make AI safe but couldn't pay the cloud bills. It's important to understand the context of this round.

Anthropic has long been the "favorite child" of Amazon and Google, who used the company as a counterweight to the Microsoft-OpenAI alliance. But a $350 billion valuation is no longer just a strategic partnership, it's a bid for dominance. If these figures are confirmed, then the venture capital market will finally turn into a game for superpowers and corporate-states.

Regular funds have nothing more to do here, they simply can't afford the entry ticket. On the other hand, such a valuation raises a reasonable question about a "bubble." When a startup whose revenue is not yet counted in tens of billions is valued at a third of a trillion, it looks like a bet that AI will replace half the white-collar workers on the planet in the next couple of years.

Anthropic will have to prove that their models are not just "smarter and safer," but capable of generating profit commensurate with their appetite for capital. So far, we're only seeing an endless arms race, where the winner is the one with the longest credit line at the bank. What does this mean for us?

Most likely, Claude will become even more powerful, integration with Amazon services will deepen, and competition will force OpenAI to release their innovations faster. But behind this facade of progress lies a dangerous concentration of resources. When two or three companies control a technology worth trillions of dollars, the notion of "open" or "accessible" AI becomes increasingly illusory.

We are witnessing the birth of a new elite, and Anthropic has just bought itself the most expensive seat in it. Bottom line: Anthropic has finally stopped being a humble research startup. With a $350 billion valuation, the company must either change the world or become the most expensive financial crash in human history.

Will they have the safety to hold such a mass of money?