Valo Health vs XtalPi (2026)

Valo Health and XtalPi both pair AI models with laboratories to design small molecules for drug companies, but XtalPi sells the laboratory as well as the discovery, while Valo sells neither software nor equipment and instead partners programs it starts from human data. Valo finds targets in more than 17 million de identified patient records, some with linked biobank samples, using AI, knowledge graphs, causal inference and statistical genetics, and then runs a closed loop in which predictive chemistry models choose what its laboratory makes next. Partners such as Novo Nordisk, Merck KGaA and Lundbeck take programs into clinical development, and Valo keeps royalties. XtalPi, listed in Hong Kong, runs more than 200 domain models trained on its own compound and reaction data alongside robotic laboratories that synthesize and test around the clock, and it began with physics based crystal structure prediction. It sells two lines: discovery services that take a partner's program from target to preclinical candidate, and laboratory automation hardware and software, which brought in more revenue in 2024. Its loop can make and test 3,000 to 4,000 new molecules in two to three months, and programs from it include PEP08, now in a Phase I solid tumor trial for PharmaEngine. A company that wants robotic chemistry capacity, automation systems for its own labs or crystal form work fits XtalPi. A company that wants a partner to find the target in human data and carry the chemistry to a candidate fits Valo.