Claude Found an Enzyme. Now the Biotech Race Gets Interesting

Some of biotech's biggest breakthroughs started with scientists noticing something unusual in nature. Now, an AI has done something similar.

Anthropic set Claude loose on a database of roughly 1.9 billion protein clusters, hunting for reverse transcriptases - enzymes that copy RNA into DNA. The agents pulled out more than 200,000 of them. Around 950 Claude agents searched through the data for unusual patterns.

Twenty-one hours later, one agent noticed something strange: an unusual enzyme sitting next to a repeated sequence.

Further research revealed a previously uncharacterised biological system, which Anthropic calls array-associated reverse transcriptase (ART). The DNA repeats have more similarities to the structures seen in CRISPR.

There is a catch, though: scientists do not yet know what ART actually does.¹

So, this is not the discovery of a new drug, or even a new gene-editing technology.

It is potentially something much more fundamental: AI finding a piece of biology that humans had overlooked.

And This Is Where The Money Gets Interesting

Drug discovery is brutally expensive.

Pharma companies can spend years and billions of dollars moving from biological research to an approved medicine. A huge amount of that process involves searching for promising targets, molecules and mechanisms - with most candidates eventually failing.

If AI can improve the very first stage of that process, the economics could change.

Imagine a scientist searching 200,000 biological sequences manually.

Now imagine hundreds of AI agents doing it simultaneously, filtering candidates and producing hypotheses for scientists to test.

The value isn't necessarily that AI replaces the scientist.

It's that AI could massively increase the amount of biology a scientist can investigate.

A previously unknown biological mechanism can now quickly become a research platform, a therapeutic target, or even intellectual property that attracts licensing deals or acquisitions.

Today's biological discovery can become tomorrow's biotech asset.

Where AI Meets Biotech Economics

This is particularly interesting as pharmaceutical companies are now facing pressure to replenish their pipelines.

Patent expiries are expected to put hundreds of billions of dollars of pharmaceutical revenue at risk over the coming years, increasing the importance of finding new medicines in order to maintain revenue. At the same time, the economics of traditional R&D remain difficult, with investors scrutinising whether huge research budgets are producing sufficient returns.²

That creates a new, obvious incentive: find more valuable biology before your competitors do.

And Big Pharma is already moving in this direction.

Novo Nordisk recently partnered with Anthropic to use Claude to accelerate medicine discovery and development.³ Anthropic has also established its own biology laboratory, suggesting it sees scientific discovery as more than simply another use case for a chatbot.⁴

That could create a new competitive advantage.

The pharmaceutical companies of the future may not just compete on who has the biggest R&D budget.

They may compete on who can search the biological universe most efficiently.

Discovery Is Only The First Step

There is plenty of hype to resist.

ART's function is still unknown, and Anthropic's findings are currently presented as early research. Human scientists are still required to carry out the laboratory experiments needed to investigate the discovery.

There is therefore no guarantee ART becomes commercially useful.

But that almost misses the point. The important demonstration is the process.

AI searched an enormous biological dataset, identified an anomaly, investigated it, checked the existing literature, and produced a hypothesis that humans could take into the laboratory.

That is a potentially valuable change in the cost of finding things.

And in biotech, lowering the cost of finding the next valuable biological discovery could be worth far more than the discovery of any single enzyme.

Final Thoughts

The real question isn't whether AI can help make better drugs.

It's whether AI can find the biology that becomes the next billion-dollar drug.

If it can, the biggest opportunity may not sit with the companies developing today's medicines, but with those discovering what becomes tomorrow's medicine.