The breakthrough wasn’t speed or scale, but a neural architecture that finally respected protein geometry.

AlphaFold didn’t accelerate biology by running faster experiments. It changed the engineering assumptions behind protein structure prediction.