AI in pharma: speed is the headline, economics is the story

I have watched this story since around 2007, when Google put money into 23andMe and the tech world first started buying its way into drug discovery. I was inside pharma finance at the time, and it read as a curiosity. It is now the main event: AlphaFold took the 2024 Nobel Prize in Chemistry, and by published counts there are more than 170 AI-originated drugs in clinical trials.

The sequence: speed, then cost, then barriers

AI does three things to an industry, in order. It increases speed. It lowers cost. And then it removes the barriers to entry that used to protect the incumbents. Once a capability that needed a large team and a decade can be reached by a small team in months, the question stops being "how fast can we go" and becomes "who can now do this that could not before".

That third stage is where the money moves, and it is the stage most strategy decks have not caught up with.

What a finance leader does with that

For a finance leader, this is the whole game, because it changes three things at once.

Who your competitors are. The moat was never really the molecule; it was the cost and time of getting one to trial. Lower both and the competitive set redraws itself: biotechs punching above their headcount, tech entrants with data advantages, and geographies that could never fund the old model.

What a pipeline is worth. I spent years building patient-based forecasting models used in transaction valuations. Every one of those valuations carried assumptions about development timelines, attrition rates and the cost of capital tied up in the journey. Compress the timeline and cut the cost, and the maths of pipeline value, in-licensing deals and probability-of-success weightings shifts under the whole industry at once. Portfolios are being revalued by physics, not by sentiment.

Where the moats actually sit. If discovery speed is abundant, advantage migrates to what stays scarce: proprietary data, regulatory trust and the governance to move fast credibly, manufacturing and supply capability, and the commercial machine that turns approval into revenue. Those are the assets to own and the lines to defend.

The barriers are falling in a known order: discovery first, where the tools are already proven; trials next, as adaptive designs and AI-assisted evidence generation mature under the emerging regulatory frameworks; manufacturing last, where physical capital still protects the incumbents. Strategy, and the capital allocation behind it, should be sequenced the same way.

Sources: AI-discovered drugs in clinical trials (sciencedirect.com/science/article/pii/S135964462400134X); AlphaFold and the 2024 Nobel Prize in Chemistry (nature.com/articles/d41586-024-03214-7); Google's 2007 investment in 23andMe (forbes.com/sites/bizcarson/2019/06/06/23andme-dna-test-anne-wojcicki-prevention-plans-drug-development/).