Drug giants no longer license AI tooling from the outside. Bristol Myers Squibb led a $237M Series E into Pathos AI this week, a round that puts the New York oncology developer at roughly $3B and hands one of its biggest customers a seat on the cap table.
Existing backers came back, and the roster reads like a pharma roll call: AstraZeneca, NEA and General Atlantic. Neither the company nor BMS disclosed how much of the round was primary versus secondary.
Pathos runs its science through Foundry, an in-house platform sitting on about 200 petabytes of cancer data. Researchers hand it hundreds of candidate molecules at a time and the system ranks them by promise, then flags the patient groups most likely to respond. The same software designs the trials, which the company describes as biomarker-driven rather than population-wide.
The first test is JSKN016, an antibody-drug conjugate the company has named as its near-term priority. The format attaches a cancer-seeking antibody to a molecule that poisons what it finds, giving treatment a delivery address instead of a systemic blast.
Why a pharma buyer led the cheque
AI drug discovery firms argue that cheaper molecule selection lifts trial odds. A pharma company leading the round is the strongest endorsement that argument has drawn from inside the industry rather than from a venture fund.
For Pathos, clinical work is slow and costly, and running it under a partner’s gaze narrows the room to change course. Iker Huerga, the chief executive, frames the platform as raising the probability of success rather than trimming spend – a claim JSKN016 will settle.
