AstraZeneca announced on October 15, 2026, that it will pay up to $1.5 billion to insitro, a South San Francisco machine-learning drug discovery company founded and led by Daphne Koller, in a partnership that will use AI models trained on lab-generated cellular data to search for new treatments for MASH, a serious liver disease, and cardiomyopathies, a group of heart-muscle disorders.
A Widening Bet on AI-Designed Medicines
The deal lands in the middle of a surge of pharma-AI tie-ups. Big drugmakers have increasingly turned to machine-learning startups to speed up the slow, expensive process of finding and validating new drug targets, betting that models trained on genomic and cellular data can spot patterns human researchers miss. AstraZeneca itself spent $13.85 billion on research and development in 2025, up from $8.05 billion in 2021, and the insitro deal is one more attempt to stretch that budget further by outsourcing early-stage discovery work to specialized AI shops.
The Terms of the Deal
Under the agreement, insitro receives $40 million upfront, with the remaining roughly $1.46 billion tied to development and commercial milestones that may or may not ever be reached, plus single-digit royalties on any resulting global sales. AstraZeneca will hold exclusive worldwide rights to any programs that emerge and will take over clinical development once insitro’s AI work identifies and advances a candidate target. The upfront payment represents a small fraction of the headline number, roughly 0.3% of AstraZeneca’s annual R&D spending, with the full $1.5 billion equal to about 11% of that budget if every milestone is hit.
Targeting Liver and Heart Disease
insitro’s Virtual Human platform combines large volumes of high-content biological data, drawn from human patients and from human-derived cellular systems grown in its own lab, to build models that predict which biological targets drive a disease, which patients form coherent treatment segments, and how a candidate molecule might behave clinically. In this deal, AstraZeneca will nominate specific disease targets within MASH and cardiomyopathies, and insitro’s models will attempt to carry those targets from early discovery through to a development candidate ready for AstraZeneca to take into preclinical and clinical testing.
A Pattern Across Big Pharma
The insitro deal was not an isolated move. Two days earlier, Zealand Pharma agreed to pay Generate Biomedicines $30 million upfront against nearly $1.2 billion in potential milestones to design AI-engineered obesity drugs, a strikingly similar structure: a small guaranteed payment against a large contingent prize. Across both deals, the message from large pharmaceutical companies is the same — AI-driven discovery is worth a seat at the table, but not yet worth paying much for in advance. AstraZeneca chief executive Pascal Soriot said in July that the company does not need more business development to hit its targets, a comment that industry observers say does not necessarily contradict signing a deal like this one, since so little of the money is guaranteed.
Reasons for Caution
The enthusiasm comes with a significant asterisk: no drug discovered using AI has yet received full approval from the FDA, despite several years of heavy investment and dozens of AI-derived molecules entering human trials industry-wide. That gap gives skeptics inside and outside pharmaceutical companies reason to treat headline deal values cautiously, since milestone payments depend on candidates clearing discovery, preclinical safety testing, and multiple phases of clinical trials — stages where most drug candidates, AI-derived or not, still fail. Proponents counter that the economics still make sense for AstraZeneca even if only a handful of programs pan out, because the upfront cost is so low relative to the potential payoff and the cost of traditional in-house discovery.
What Comes Next
For now, the partnership is entirely in the discovery phase: insitro’s models need to nominate and validate targets before any molecule reaches a lab bench, let alone a patient. AstraZeneca has not disclosed how much of the $1.46 billion in milestones could become payable before a candidate reaches clinical trials, nor a timeline for when the first MASH or cardiomyopathy program might advance. The deal will nonetheless be watched closely as a test case for whether AI-driven target discovery can meaningfully shorten the years-long road from biological hypothesis to an approved medicine — the same question hanging over a wave of similar partnerships being signed across the industry in 2026.
Photo: 4339272 / PIXABAY via Pixabay
