insitro proves fundable candidate pipeline

The real bottleneck in AI drug discovery is not model accuracy, it is whether a model can generate molecules and targets that survive pharma diligence and then survive biology. insitro has shown enough with Bristol Myers Squibb to expand from target discovery into molecule design, which matters because partners only keep paying when early work turns into assets they want to advance, not just interesting predictions.1 2 3

• insitro already has a concrete proof point with Bristol Myers Squibb. The collaboration began in 2020 with $50M upfront and more than $2.1B in potential value, then expanded in 2026 after new target nominations, showing that a large pharma partner was willing to keep funding work after seeing preclinical output.2 3
• Recursion is the clearest scaled rival because it pairs AI models with wet lab data, chemistry, and clinical programs. After combining with Exscientia in November 2024, it described an end to end platform with more than 10 clinical and preclinical programs, which makes it easier to prove commercial value through actual drug progress rather than model benchmarks alone.4 5 6
• Isomorphic Labs and Xaira show the two other ways this market is being won. Isomorphic has major pharma partnerships with Novartis and Eli Lilly, while Xaira raised $1B to build around protein and antibody design. But both still need to show that their platforms can reliably turn computation into funded candidates and then into human data.7 8 9

The next phase of competition shifts from who has the best model to who can stack enough evidence across target selection, chemistry, preclinical data, and early clinical readouts to keep partner capital flowing. The winners will be the companies that make pharma buyers feel they are funding a future medicine, not an AI experiment.1 3 5 7 9