A three-part series exploring how the biopharma industry is evolving in the face of two global factors: AI and China.
The first wave of AI-bio capital has been priced around the companies developing the models, but I think the model layer is becoming increasingly saturated. While models are no doubt important, the harder questions are what feeds those models: whether the biology is truly predictive, whether the clinical trial infrastructure can absorb more candidates, and whether companies can learn quickly enough from first-in-human human data.
AI and machine learning have been used in drug discovery for a long time, but the new wave of generative AI is now making huge impacts across the broader biopharma industry. However, AI-bio models inherit the biology, the experimental systems, the clinical infrastructure, and the regulatory frame that they operate within. Capital has so far concentrated in the model layer that looks like software; meanwhile, the adjacent layers that determine whether the software actually produces medicines have been under-looked. The companies that provide outsized returns in the medium-to-long term may be the ones that build the infrastructure underpinning the models.
The first piece asks whether the biology is predictive. The second asks whether the clinical system can absorb more candidates. The final piece asks where those learning loops can actually be built, and what that means for European biotech in a world where China has industrialised the cheap-fast clinical loop.
Why Better Biology Beats Bigger Models. Biopharma doesnt need more data, it needs biological fidelity. Models trained on weak experimental systems learn weak proxies for human biology. The question is what the predictive validity of the system generating the training data actually is.
PublishedWhy More Candidates Won't Mean More Drugs. Even if AI generates better candidates, the clinical system cannot automatically absorb them. The bottleneck moves from candidate generation to first-in-human translation. The question is how a company gets from candidate to human readout faster than the market.
PublishedIf China owns the cheap-fast clinical loop, Europe needs a different kind of learning asset. Europe is unlikely to win a brute-force speed-to-first-in-human race. The question is whether a European company is exploiting a structural advantage (e.g. higher-fidelity biology, registry-linked human data, regulator-defined evidence pathways, and US capital legibility) or just running a slower version of a US or Chinese comparable.
Published