Faster, cheaper, better? Post-buyout deal Celgene jumps into AI alliance with a $25 million bet on speeding discovery work
Celgene clearly isn’t waiting in limbo to see when, or if, the big Bristol-Myers Squibb deal will go through. It’s still executing deals, and the R&D side of the business has just enlisted one of the more prominent AI players to go to work on a trio of new drug projects in oncology and immunology. They’re paying $25 million up front to get the tech party started.
Celgene, a top 15 R&D group worldwide by research budget, tied up with Exscientia in Oxford, UK for the work. We aren’t getting any specifics about the targets, but the company is exploring AI to see how it lives up to the emerging field’s big boast: That they can deliver new drugs for human testing faster and more accurately than the standard industry approach the big players have been using. In Celgene’s case, that would commonly mean going out and doing a discovery deal with a biotech, but the majors also have their own in-house operations.
“This is the largest AI deal so far,” Exscientia CEO Andrew Hopkins tells me, with “very substantial” milestones built in.
The AI player’s boast is that they can take you through concept to hit discovery and lead identification in 12 months or less — which they say they’ve done 4 times now. The first 3 were bispecifics, not easy to do, the CEO adds.
It’s an interesting business model which has attracted considerable attention over the last 2 years. It’s also delivered pacts for Exscientia with a bevy of prominent industry players: Roche, GSK, Sanofi and Evotec.
Like others in the field, Hopkins likes to point to the numbers in an influential study that Steven Paul — now a biotech entrepreneur — did in the spring of 2010 during a lengthy stint at Eli Lilly which underscored the kind of time and money that a pharma giant spent on discovery (How to improve R&D productivity: the pharmaceutical industry’s grand challenge). The capitalized cost of lead optimization alone, he and his team — which included Bernard Munos — calculated, was $414 million. The out of pocket for lead optimization was $146 million.
These numbers have been known to cause some hooting in the biotech world, where it’s not uncommon to get $30 million — or considerably less — in VC money to get through to an IND. Lead optimization in the biotech world happens less expensively. But the majors are playing a different game, which is one reason why most of the pharma giants have been working on AI as a way to improve efficiencies — though many would tell you the jury is out and will remain out until these IND projects turn into approved drugs.
One number that Hopkins was curiously unwilling to part with was the size of his staff. He declined to say how many employees work at the AI company, and a spokesperson followed up with “less than 50.” After this story posted online, another contact came back with “around 50.” So make of that what you will.