Insilico Medicine — AI supply-chain exposure

Insilico Medicine uses generative AI to design drug candidates from scratch — proposing novel molecules rather than screening existing libraries — and has carried several of those AI-originated candidates into clinical trials. It listed in Hong Kong in the largest local biotech offering of its year, giving public markets a pure-play read on generative drug discovery. In the model it is a GPU integrator: its generative pipelines run on large accelerator fleets, placing it downstream of the same compute stack that powers frontier models.

The structural hook is that its output is validated in the clinic, not just in silico — the value case rests on whether AI-designed molecules actually work in patients. That makes it exposed both to GPU cost and to biological risk, a combination few peers carry. The model keeps it at the applied edge, where AI capability is consumed to generate candidates, well away from any control over the scarce silicon underneath.