Anthropic just announced something that goes way beyond chatbots and coding tools.
At an event called “The Briefing: AI for Science,” the company unveiled Claude Science, a workbench that pulls scattered research tools and data into one place and helps scientists generate figures and visuals.
But the bigger headline came right after: Anthropic plans to develop its own drugs.
That’s a major shift. Anthropic already sells AI models to biotech and pharma companies. Now it wants to become one of those companies itself.
What Anthropic Actually Announced
Eric Kauderer-Abrams, who leads Anthropic’s life sciences work, said the company will zero in on “neglected” diseases – conditions that don’t attract much research funding or attention from big drugmakers.
Beyond that detail, Anthropic hasn’t shared much.
The company hasn’t said which diseases it’s targeting first.
It also hasn’t explained what happens if its research turns up a promising drug candidate, or whether it plans to team up with other companies for lab testing, clinical trials, or manufacturing. Anthropic didn’t respond to requests for more details either.
Why This Move Stands Out
Every major AI company wants a piece of the science and pharma market.
OpenAI, Amazon, and Google all have their own life sciences tools. What makes Anthropic different is the direction it’s heading: instead of just building software for drugmakers, it wants to become one, competing with some of the very companies it sells to.
This puts Anthropic in a crowded field.
AI-first drug companies like Insilico and Google DeepMind’s Isomorphic Labs are already racing to find treatments using machine learning.
Traditional pharma giants are building or buying their own AI tools too. Anthropic is now stepping into that same arena, but from an unusual angle – as an AI lab first and a drug developer second.
AI’s Role in Drug Discovery Is Bigger Than People Think
Here’s something worth understanding before getting too excited or too skeptical: “AI drug discovery” doesn’t mean one specific thing.
Namshik Han, a professor at the University of Cambridge and cofounder of AI biotech startup CardiaTec, called it a broad term that covers nearly every stage of the process.
AI helps find new chemical compounds, improve existing ones, analyze research data, and even support clinical trials and manufacturing.
Matthew Todd, a drug discovery professor at University College London, made a similar point.
He described AI as a catchall phrase because it’s already woven into so many parts of pharmaceutical research.
Big names like AstraZeneca, Novo Nordisk, and GSK are all running their own AI initiatives. So in one sense, Anthropic isn’t doing anything totally new – it’s joining a trend that’s already well underway.
What AI Can and Can’t Do Right Now
AI genuinely speeds things up.
It can suggest new molecules that might interact with parts of the body already linked to a disease, and it can help researchers “road test” ideas faster than old-school trial and error.
Given Anthropic’s expertise in large language models, it’s likely the company will lean on generative AI to scan huge amounts of chemical and biological data, looking for connections humans might miss or take years to find.
But here’s the catch: no AI-designed drug has ever made it through clinical trials and FDA approval to reach pharmacy shelves.
Not one.
Some AI-assisted candidates have entered trials, but it’s genuinely hard to say how much credit AI deserves, or whether those drugs actually work better than ones developed the traditional way.
Why Drugs Still Take Years, No Matter How Smart the AI Is
Frank von Delft, a structural chemical biology professor at the University of Oxford, put it plainly: AI models are impressive, but they haven’t come close to replacing real-world experiments.
A drug still has to prove it’s effective and safe in actual living systems. It also needs to be practical to manufacture, store, and deliver.
None of that happens inside a computer.
That means Anthropic will need serious money, skilled scientists, and time if it wants to turn any AI-generated idea into an actual medicine. Von Delft said flatly that the company is “going to have to spend a lot on experiments” if it’s serious about this.
There’s another obstacle too: missing data.
Todd and Han both pointed out that high-quality experimental data – like how specific chemicals behave inside the human body – simply doesn’t exist for a lot of biology yet.
Even well-studied diseases still have major gaps in our understanding. AI can only work with what’s already known, and in medicine, a lot remains unknown.
Is Anthropic Actually Building the Team for This?
It looks that way. Over the past year, Anthropic has been hiring biologists and setting up its own wet labs – physical spaces where real chemical and biological experiments happen, not just computer simulations.
Han said Anthropic has been actively recruiting in this space, and some of his academic colleagues have already been approached.
He believes the company has successfully pulled talent away from both Big Pharma and top research universities, though he didn’t name specific people.
How Long Until Any of This Pays Off?
Don’t expect quick results.
Clinical trials alone typically take years, and Todd noted there’s always a significant lag time when testing new medicine – you need real experiments to prove something is safe, and that simply can’t be rushed.
Realistically, any payoff from Anthropic’s drug development efforts is likely the better part of a decade away, if it happens at all.
That’s not a knock against Anthropic specifically.
It’s just how drug development works, whether a tech company or a traditional pharma giant is running the show.
AI can narrow the search and speed up early research, but medicine still has to prove itself the old-fashioned way – through slow, careful testing in the real world, on real biology, over real time.
