Google DeepMindLab

Track record applying deep learning to materials discovery (GNoME) and physical simulation.

Deep RL and generative design expertise increasingly aimed at autonomous scientific discovery loops.

AlphaFold and AlphaMissense make it the definitional player fusing deep nets with structural/functional biology.

AlphaFold and materials/GNoME work show intent to model catalysts and biomolecular systems at scale.

Track record applying deep learning to materials (GNoME) and scientific operator learning bridges both fields.

AlphaFold made learned protein structure the default; extending to dynamics and density maps is the obvious next front.

GNoME and materials-discovery agents push AI-planned inorganic/nanomaterial search at scale.

Applies deep learning to scientific imaging and physical-system design at scale.

GNoME materials-discovery work directly targets novel inorganic crystals, a natural launchpad into superconductor screening.

Demonstrated ML control of power/cooling and energy systems, and works on hardware-aware learning.

AlphaFold and successor models make protein/matrix structure prediction a direct input to ECM biology.

Track record applying deep learning to materials discovery (GNoME), a natural fit for predicting thermal transport.

GNoME and graph-network work already predict stable inorganic materials at scale, directly bridging AI and materials physics.

GNoME and materials-discovery efforts apply learned models to predict novel stable materials at scale.

Materials and structure inverse-design (e.g., GNoME-style discovery) that feeds AI-driven fabrication targets.

AlphaFold and successors define the structural-prediction substrate any ER-folding model would build on.