Anthropic Says It Will Begin Developing Its Own Drugs

Anthropic, the AI company behind Claude, announced it will begin developing drugs of its own — an ambitious move that pushes the maker of one of the world’s best-known AI assistants directly into pharmaceutical research. The disclosure came Tuesday in San Francisco, alongside the launch of the company’s newest application, Claude Science.

What Anthropic announced

Already a dominant force in technology and a household name for everyday AI users, Anthropic is now testing whether it can make drugs too. On Tuesday the company said it is going to try. It remains unclear whether Anthropic intends to carry drug candidates all the way to commercialization, but multiple executives stressed how important it is for the company to gain hands-on experience using its own products to tackle real scientific problems.

Eric Kauderer-Abrams, Anthropic’s head of life sciences, framed the decision as an answer to a question the company had been asking itself: what should it be doing beyond training models and building products? Developing drugs, and using its AI tools to do so, is one answer. For anyone watching the intersection of artificial intelligence and biotech, it’s a significant statement of intent.

Claude Science and the strategy behind it

The drug-development push was revealed at an event to launch Claude Science, a new application aimed at researchers and the scientific community. The logic connecting the two is straightforward: rather than only selling tools to pharmaceutical companies, Anthropic wants to use those tools itself on genuine research challenges. That hands-on experience, executives argued, is the best way to understand where AI actually helps in the lab — and where it falls short.

This is a familiar pattern in the AI industry, where companies increasingly try to prove their models’ value by applying them to hard, real-world domains rather than benchmarks alone. Drug discovery is among the most demanding of those domains, involving biology, chemistry, and years of validation that no model can shortcut. Readers who track science and research will recognize how high the bar is.

The move also positions Anthropic within a fast-growing intersection of AI and the life sciences, where the promise is that models can help sift enormous volumes of biological data, surface patterns, and speed up early research. The reality is more sobering: turning a computational insight into an approved medicine still requires laboratory work, clinical trials, and regulatory review that unfold over years. By developing drugs itself rather than only selling software to pharmaceutical firms, Anthropic is choosing to confront those constraints directly instead of at arm’s length — a stance that will make any eventual results, positive or negative, unusually instructive.

Why it matters

An AI company deciding to develop its own drugs blurs the line between software vendor and research organization. If Anthropic can use AI to accelerate any part of the discovery process, it would strengthen the case that large language models can contribute to serious science rather than just consumer productivity. If it struggles, that too would be instructive — a real-world check on the frequent claims that AI is poised to transform medicine.

For now, key questions stay open. Anthropic hasn’t said which diseases or targets it will pursue, how far it intends to take any candidates, or whether commercialization is a goal at all. What’s clear is that the company sees direct participation in research — not just tool-building — as central to proving what its AI can do.

Frequently asked questions

Is Anthropic going to sell drugs?

It’s unclear. The company said it will begin developing drugs, but has not confirmed whether it intends to bring any candidates to commercialization.

What is Claude Science?

A new Anthropic application, launched alongside the drug-development announcement, aimed at supporting researchers and scientific work.

Why is an AI company developing drugs?

Executives said the goal is hands-on experience using Anthropic’s own products to solve real scientific problems, rather than only training models and selling tools.