Everyone is talking about how Anthropic AI lab now designs drugs, hires the AlphaFold Nobel laureate, and ships Claude Science alongside Claude Code. Also, their partnership with Basecamp Research (i.e. EDEN-designed antibiotic peptides 97% active against WHO priority pathogens), integration of Claude Science with Boltz, and so on. There is a real movement.
But despite the obvious impressiveness and importance of this move, the part that Anthropic is addressing is the one that was never the most expensive/complex part of the drug development business, in my opinion.
Designing a promising molecule has been getting cheaper for a decade, as well as target selection. That is not where drug discovery programs fail, generally speaking. Drug programs fail in toxicity and unexpected systemic effects on body, lack of efficacy, in manufacturing, in failing to enroll enough/right patients, in trials that fail because the biology was misjudged or the population was selected in a sloppy trial design (lack of biomarkers, etc.).
As far as I can tell, Anthropic has limited clinical apparatus, no manufacturing, no regulatory track record, and, likely, no answer yet to “what do you do if you find something that works?” Yes, sure, they can design drugs really fast, but then what? How does their model differentiate against later stages, where the role of AI is surprisingly small at the moment?
Claude Science can probably move “target to shortlist in minutes,” and so it compresses the cheapest, fastest, most-solved stage of the pipeline. It is certainly a milestone, certainly impressive, but have they solved drug discovery with it? Probably not yet.
Now, what does it mean for the AI-native biotechs, the likes of Iambic, Recursion, Insilico Medicine, OWKIN, Isomorphic Labs, Cradle, NOETIK, Xaira, SandboxAQ, etc.)?
Well, their moat was never the model alone, I think. It is this:


The refinement I'd add is that "data" and "validation" serve very different roles in that phrase, and the gap between them is where most of the value lies. Proprietary data is necessary and increasingly common. Validated data, calibrated to the biology you're actually drugging, is the rare, capital-intensive part, and it's the only thing that survives the stages Anthropic can't touch: tox, efficacy, patient selection.
Which sharpens @Bengüsu's build-vs-partner point nicely. You can rent a model, and soon everyone will. You cannot rent a validated loop for your specific biology, the thing that tells you whether the data predicts the clinic. That is what a small player with real data and tacit knowledge actually owns, and it's the leverage to set terms before the big labs generate their own.
It also makes the China section the sharper story: speed and enrollment scale are a validation advantage as much as a cost one, more shots and faster readouts, and that compounds exactly where model access doesn't.
Great read. The China angle is gold.