
Bristol Myers fronts new Schrödinger alliance with $55M upfront, expanding precision oncology profile
Bristol Myers Squibb has a new R&D partner, one to which they’re paying a pretty penny to use their discovery platform.
The pharma company is doling out $55 million upfront to Schrödinger $SDGR to work on up to five small molecules, with the potential for $2.7 billion in milestone payments. Schrödinger’s initial targets include HIF-2 alpha and SOS1/KRAS for a type of kidney cancer and KRAS-driven cancers, respectively.
Monday’s news was greeted well by investors, as Schrödinger shares ticked up roughly 10% in early trading.
Within the collaboration’s 8-K, Schrödinger got into the nitty gritty of the milestones. The company can receive up to $585 million per oncology target, with $360 million for R&D and regulatory milestones and $225 million for commercial milestones. For neurology and immunology targets, BMS’s payments could max out at $482 million — $257 million on the R&D and regulatory side, with another $225 million for commercialization milestones.
Schrödinger can also receive royalties ranging from the mid-single digits to low-double digits.
The company’s discovery platform proved the impetus for such a deal, which CEO Ramy Farid stressed to Endpoints News is not one of the many machine learning or AI-based technologies that have taken the biotech world by storm. Rather, Schrödinger’s platform creates physics-based molecular simulations that can predict how molecules and compounds will bind to proteins before even stepping foot in the wet lab.
And since there’s no AI involved, Schrödinger doesn’t have to spend time training their model to recognize and detect a drug’s potential affinity. Farid says the company can simulate a compound’s entire chemical space, or the pool of its potential elemental combinations, which can total up to 10 to the power of 50 different structures.
“For example when you want to build an AI, let’s say to detect cats in photographs, you feed the model a bunch of pictures of cats … physics-based methods do not do that at all, there is no training set,” Farid said, adding later, “This is the reason why AI doesn’t work reliably. That space is so enormous that it’s impossible to generate a training set that can capture all of it.”

Though Schrödinger isn’t close to testing anything in animals, let alone humans, they’ll be working on the two initial targets for the foreseeable future, as the partnership runs for four years. Once the programs in the deal get to the IND-enabling stage, Schrödinger will hand off development to BMS, R&D chief Karen Akinsanya said. Schrödinger will retain unlimited access to its platform.
For BMS, the deal fits into a trend of expanding their immuno-oncology effort through R&D partnerships since its big-money acquisition of Celgene. The company spent $475 million in August for control of Dragonfly’s extended half-life cytokine DF6002, a monovalent IL-12 immunoglobulin Fc fusion protein designed to fire up a targeted immune response.
A week later, BMS bought out a small biotech called Forbius whose pipeline of TGF-beta inhibitors matched their oncology deal profile. The terms of that transaction weren’t disclosed, though Forbius’ unrelated assets were spun out into another company.