Bristol Myers Leads a $237 Million Round That Values AI Drug Developer Pathos at $3 Billion

Pathos AI, a New York company that describes itself as an AI and technology company developing its own drugs, said on Friday that it closed a $237 million Series E at a post-money valuation of approximately $3 billion. Bristol Myers Squibb led the round, with existing investors participating. Axios reported that the new valuation is nearly double the $1.6 billion Pathos carried after a $365 million Series D in May 2025.

The financing landed alongside a second, more unusual announcement: a three-way arrangement with Bristol Myers Squibb and Tempus AI aimed at Alzheimer's disease and other neurological conditions. Tempus will assemble a de-identified longitudinal dataset from hundreds of thousands of patients with neurological disorders. Pathos will license that data for $55 million over five years and point its platform at it. A regulatory filing reported by Dow Jones put the combined cash commitments at $192.5 million over five years, including $137.5 million from Bristol Myers directly and the $55 million from Pathos for 500,000 de-identified patient records. The Bristol Myers-Tempus data agreement runs through 2031.

The structure is what makes the deal notable. Bristol Myers is simultaneously the lead equity investor, a platform customer, and the counterparty on a dataset purchase. Under the collaboration it gains non-exclusive access to Pathos's Foundry platform across its wider research portfolio and three-year exclusive rights to use it in neuroscience development. That is a pharmaceutical company buying an option on a software stack while funding the company that owns it.

Foundry is the commercial pitch. Pathos says it runs thousands of specialized AI agents that continuously aggregate clinical data, genomics, biological specimen data and the competitive landscape for a drug program, then surface what it calls the optimal next decision: which sub-population is most likely to respond, when to pivot a trial, and which programs to kill early. Chief executive Iker Huerga framed the goal as building a platform to systematically improve the probability of success in drug development. Those are company claims; Pathos has not published prospective data showing that its decisions beat conventional portfolio management.

The pipeline it is funding now has four clinical-stage assets: JSKN016, a first-in-class TROP2/HER3 bispecific antibody-drug conjugate that the company wants to move into registrational studies; pocenbrodib in metastatic castration-resistant prostate cancer and relapsed or refractory multiple myeloma; AZD4241, an ER PROTAC partnered with AstraZeneca; and DO-2, a next-generation MET kinase inhibitor. Pathos says Foundry helped it identify and acquire those assets quickly through a series of transactions over recent months, which is the operational argument behind the round.

Whether that argument survives contact with clinical readouts is the open question. Drug development fails mostly in Phase II and III for reasons that have little to do with information retrieval: the biology does not behave, the endpoint is wrong, the trial population is too heterogeneous. An agent swarm that synthesizes everything known about a program can sharpen decisions, but it cannot manufacture efficacy. The registrational study for JSKN016 will be the first real test of whether the platform changes outcomes rather than just accelerating deal-making.

For the wider AI-in-biotech market, the round is a pricing signal. Investors have spent two years debating whether AI drug discovery platforms deserve software multiples or biotech multiples. A $3 billion valuation attached to a company with four clinical assets and a named pharma lead investor suggests the market is willing to pay for the combination, at least while the assets are still unread. The next comparable datapoint will come when a Pathos program reports results, not when the next round closes.