On March 30, 2026, Eli Lilly and Hong Kong-based Insilico Medicine announced a research and licensing agreement worth up to $2.75 billion, one of the largest AI-drug-discovery deals ever struck between a Big Pharma giant and an AI-native biotech. Lilly is paying $115 million upfront for an exclusive worldwide license to a portfolio of novel oral therapeutics that Insilico’s generative-AI engine designed and that are currently sitting in preclinical development across multiple disease areas. The rest of the money is staged behind development, regulatory, and commercial milestones, plus tiered royalties if any of the molecules eventually reach pharmacy shelves.
How Insilico Got Here
Insilico Medicine is not a newcomer chasing hype. Founded in 2014 and built around its Pharma.AI platform, the company has spent the better part of a decade training generative models to do two things traditional drug discovery treats as separate, slow-moving disciplines: find a druggable biological target, and then design a molecule that hits it. Between 2021 and 2024, Insilico says its platform nominated 20 preclinical candidates, with each program taking just 12 to 18 months from start to nomination and requiring the synthesis of only 60 to 200 molecules — a fraction of the thousands typically screened in conventional medicinal chemistry. Conventional drug discovery, by contrast, routinely takes three to six years to reach the same milestone. The Lilly deal builds on a relationship the two companies have quietly maintained since 2023, when smaller collaborations first tested whether Insilico’s outputs could survive contact with a major pharma company’s due-diligence process.
What Lilly Is Actually Buying
Under the agreement, Lilly gets rights to develop, manufacture, and commercialize a slate of oral therapeutics that Insilico’s Pharma.AI engine has already carried through target identification, molecule design, and preclinical candidate nomination. The companies will also run additional joint R&D programs, pairing Insilico’s generative chemistry with Lilly’s clinical development machinery and regulatory experience — the part of the pipeline where AI-native biotechs still lean heavily on partners. Andrew Adams of Lilly framed the tie-up as a way to “explore novel mechanisms and accelerate identification of therapeutic candidates,” while Insilico founder and CEO Alex Zhavoronkov described the platform’s edge as its ability to identify “multi-purpose targets driving multiple diseases simultaneously” using what he called frontier AI technologies.
The Skeptical Read
Not everyone in the industry treats these numbers uncritically. Milestone-heavy biobucks deals routinely get announced at eye-popping topline totals that pharma companies have little obligation to actually pay out — the $115 million upfront is real money, but the remaining $2.6 billion-plus is contingent on candidates clearing clinical trials that, historically, more than 90% of drug candidates fail to do. AI-designed molecules have not yet proven they fail less often than traditionally discovered ones; the field’s first fully AI-discovered-and-designed drug, Insilico’s own rentosertib, only reported Phase IIa results in June 2025. No AI-originated drug has been approved by the FDA as of mid-2026. Skeptics also note that “12 to 18 months to a preclinical candidate” describes the front end of drug development, not the far more expensive and failure-prone clinical stages that determine whether any of this saves money at all.
Why Lilly Is Betting Anyway
Lilly’s counter-argument, echoed across the industry, is that even modest reductions in preclinical timelines compound across a portfolio. If Insilico can reliably shave two to four years off early discovery for even a handful of programs, that is years of patent life and years of reduced R&D burn that Lilly does not have to fund alone. It also lets Lilly diversify its pipeline risk across many AI-nominated candidates rather than betting big on a handful of internally discovered ones — a hedge that has become increasingly attractive as internal R&D productivity across the industry has plateaued.
What Happens Next
The near-term test is not whether Insilico’s AI can nominate candidates quickly — it has already demonstrated that — but whether Lilly can push several of these programs into human trials and get clean safety and efficacy signals. Watch for IND filings tied to this collaboration over the next 12 to 18 months, and for whether Lilly discloses which therapeutic areas the portfolio actually covers, information both companies have so far kept vague. The deal also cements 2026 as the year Big Pharma stopped treating AI drug discovery as an experiment and started treating it as a standard, if still unproven, sourcing channel for its pipeline.
