Pharma partners could commoditize Isomorphic
Isomorphic Labs
This is the core risk in selling AI discovery into large pharma, the customer can treat the partner as a temporary speed layer, then keep the workflow and bargaining power in house. Isomorphic’s announced partners, Lilly and Novartis, already buy broad discovery software from Schrödinger and work with other AI design vendors, so Isomorphic has to prove it improves hit finding, lead optimization, and eventually clinical progress, not just model quality.
-
Schrödinger shows what the incumbent alternative looks like. Pharma teams can license software, run molecule design inside their own org, and expand usage across many scientists over time. That setup naturally teaches the customer to rely less on any single external platform company.
-
Chai Discovery represents the newer multi vendor pressure. It has signed multiple large pharma collaborations, including Bristol Myers Squibb and Novartis, and has been publicly linked to Pfizer and Lilly interest, which means the same buyers can compare vendors side by side and push pricing toward service like economics.
-
Insilico raises the bar by pairing AI discovery with public clinical progress. Its first generative AI drug candidate entered Phase II in 2023, while Isomorphic still had no publicly named clinical candidate or trial timeline as of September 2026. Without human data, partners may reserve cleaner programs for internal teams and externalize harder biology.
The next phase of competition shifts from who predicts structures best to who owns the full loop from model, to assay, to medicinal chemistry, to clinic. If Isomorphic can turn partnerships into named candidates and human data, it keeps leverage. If not, pharma will keep multi homing, internal tools will get stronger, and Isomorphic risks becoming one option in a crowded vendor stack.