Schrodinger, Inc. — AI supply-chain exposure

Schrodinger applies AI to one of the hardest search problems in science: finding molecules that will become drugs. Its platform pairs physics-based simulation with machine learning to predict how candidate compounds behave before anything is synthesised, and it licenses that software to most of the major pharmaceutical companies. In the model it is a GPU integrator — its edge depends on running large physics-and-ML workloads across substantial accelerator fleets, making it a direct downstream consumer of the same compute that trains frontier models.

Its structural hook is a dual business: it sells the software and also co-develops its own drug pipeline, so it captures value both as a tool vendor and as a discoverer. That ties its economics to GPU availability and cost — cheaper, more plentiful compute widens the molecular space it can afford to explore. The model positions it at the applied layer, where AI turns into scientific output rather than where the scarce hardware itself is made.