Isomorphic Labs vs Valo Health (2026)

Research collaborations with drug companies are how Isomorphic Labs and Valo Health sell their AI, rather than products, and neither has taken a molecule of its own design into the clinic. What each brings to a partner differs: Isomorphic brings a design engine built on the structure of the target, and Valo brings human data that says which target to pursue. Isomorphic's drug design engine, IsoDDE, grew out of AlphaFold 3, which it released with Google DeepMind in 2024. It predicts protein ligand structures, finds binding sites from a protein sequence alone, estimates binding affinity and models antibody antigen interfaces, while design agents generate molecules against a specification. The company reports IsoDDE more than doubling AlphaFold 3's accuracy on the protein ligand cases least like its training data. Novartis, Eli Lilly and Johnson & Johnson are partners, and a $2.1 billion Series B in May 2026 funds the work. Valo's Opal platform works further upstream. It groups de identified patient histories into subtypes, ranks targets with causal inference and statistical genetics, checks them in human data, and then runs small molecule chemistry in a loop with its laboratory. Novo Nordisk's agreement, the one with public terms, covers up to twenty cardiometabolic programs. When the target is known and the hard part is the molecule, whether a small molecule, an antibody or a molecular glue, Isomorphic fits. When choosing the target is the open question and small molecules are the modality, Valo fits.