Anthropic's IPO splits the spoils: Broadcom cashes in first while AI for biotech faces its acid test

By The Motley Fool · October 31, 2026.

An analysis by The Motley Fool, written by Prosper Junior Bakiny, argues that Anthropic's upcoming initial public offering (IPO) will create winners and losers beyond the company itself. Citing unspecified reports, the article says the IPO could break the record for the largest in history, set four months ago.

The winner it identifies is Broadcom. According to the article, the company has a five-year agreement worth $161.2 billion, non-cancellable except in the event of non-payment, to supply Anthropic with compute capacity based on TPUs (Tensor Processing Units, custom AI chips that Broadcom co-designs with Alphabet). The article adds that Broadcom expects $115 billion in AI revenue in fiscal 2027, more than its total revenue over the past twelve months, and $230 billion in 2028. These are the company's own forecasts, not results.

The loser is Recursion Pharmaceuticals, a biotech founded in 2013 that uses an AI operating system to screen drug candidates. The author argues that, despite the years that have passed, it has no drugs on the market and, in his view, no phase 3 programs. He also argues that its edge will erode because, according to Anthropic, nine of the ten largest pharmaceutical companies are among its customers that spend $100,000 or more a year and use Claude for drug discovery.

A note of caution is warranted: this is an investment article, not a technical analysis. The article itself discloses that its author holds positions in Recursion and that The Motley Fool holds positions in Broadcom. In addition, in the session before publication (October 30), both stocks closed lower (Broadcom, -5.48%; Recursion, -4.95%), amid broad declines on the Nasdaq.

**Our take.** The underlying thesis is more interesting than the bet. When a model lab raises massive capital, a substantial share flows straight back to those who build the foundations: chips, networks and data centers. A non-cancellable $161.2 billion commitment turns compute capacity into a fixed obligation, and that is only sustainable if Anthropic's revenue grows at the promised pace. The risk lies as much with Broadcom as with the business model itself, and the IPO exposes it to public scrutiny.

Recursion is a textbook example of another dynamic: general-purpose AI erodes vertical AI when the latter has not demonstrated clinical results. But it would be a mistake to read this as a failure of AI in medicine. The metric that matters is not who has the best “operating system”, but how many approved drugs come out of the process. That takes years, and no language model can shorten it by decree, because clinical trials and regulation remain the bottleneck.

Here is the part that does matter to us in the long run. That nine of the ten largest pharmaceutical companies use general-purpose AI suggests the tool is becoming standard infrastructure for drug discovery. If that adoption ends up translating into more viable molecules and better-designed trials, it is a real step toward eradicating disease. In the short term, however, what we will see is consolidation: those who control compute and distribution win, and the startups that sold a promise without a product lose. That transition concentrates power, and it is worth watching.