AstraZeneca and insitro have signed a worldwide exclusive research collaboration to find new drug targets in MASH and cardiomyopathies. insitro announced the AstraZeneca and insitro agreement on 15 October 2026 in South San Francisco: $40 million upfront, up to about $1.46 billion in milestones, and tiered single-digit royalties on any sales.
No medicine comes out of this deal yet. The agreement covers the earliest stage of discovery, finding the biology that drives each disease and turning chosen targets into development candidates. Every figure and description below comes from insitro's own announcement, which in the evidence hierarchy we use is a company statement, not evidence that the approach works.
Three questions matter for a UK reader: what the money is really worth, who does what, and where MASH treatment in Britain stands while this research begins. Our wider view of the year's deals is in our review of AI drug discovery deals in 2026.
What have AstraZeneca and insitro agreed
AstraZeneca pays insitro $40 million upfront for a worldwide exclusive collaboration in MASH and cardiomyopathies. insitro can earn up to about $1.46 billion more in discovery, development and commercial milestones, plus tiered single-digit royalties on global sales. insitro finds targets; AstraZeneca runs clinical development, regulation and commercialisation.
MASH is metabolic dysfunction-associated steatohepatitis, the inflammatory form of fatty liver disease formerly called NASH. Cardiomyopathies are diseases of the heart muscle. The release describes both as common and underdiagnosed, and makes a narrower point worth noting: patients who share one diagnostic label can have quite different underlying biology, which is part of why both conditions are hard to treat.
insitro will apply its Virtual Human platform, which it describes as a causal AI discovery platform, to identify disease biology in both indications. It then takes targets nominated by AstraZeneca from discovery through to development candidate. From that point AstraZeneca leads everything that follows.
The release also names a second insitro platform, TherML, which it says designs medicines against the causal drivers that Virtual Human identifies. It does not say whether TherML is part of the AstraZeneca work. For a reader, the practical meaning is narrower than the language suggests: insitro's job ends at a development candidate, before any human trial.
Why the reported deal value differs
Readers searching for this deal will see 3 different numbers. Bloomberg's headline gives up to $1.86 billion. Fierce Biotech and PharmExec give up to $1.5 billion. Euronext's newswire led with the $40 million.
The release itself gives no total. It states the $40 million upfront and "up to approximately $1.46 billion" in milestones as separate figures, and adds royalties that cannot be valued until a product exists. Added together, the 2 stated figures come to roughly $1.5 billion. We could not find, in the release, the basis for the $1.86 billion figure, so we report it as Bloomberg's and not as a disclosed term.
The distinction matters more than it looks. Milestone totals in discovery deals are maximums, paid only if programmes clear each stage. The certain money today is $40 million.
Royalties work the same way. "Tiered single-digit" means a percentage below 10% of sales, rising in steps as sales grow, and payable only on a product that reaches the market. The release does not say how many targets AstraZeneca may nominate, which is the number that would turn the milestone ceiling into a realistic estimate.
Who leads the work and what each side has said
Daphne Koller, insitro's founder and chief executive, framed the deal around data. "MASH and cardiomyopathies are prevalent, underdiagnosed diseases where high-content data holds immense potential," she said in the release.
For AstraZeneca, Sharon Barr, Executive Vice President and Head of BioPharmaceuticals R&D, put the aim plainly: the company is using AI "to identify the biological drivers of disease", in her words.
Both statements describe intent. Neither is a result, and the release reports none from this collaboration, which is still at its start.
insitro is not new to liver disease. In June 2026 it presented data at the American Diabetes Association's 86th Scientific Sessions on what it calls an AI-discovered MASH candidate, reporting an anti-fibrotic signal beyond liver-fat reduction. That was a conference presentation, not a peer-reviewed publication, and it concerns insitro's own programme rather than anything in the AstraZeneca and insitro collaboration.
The company has been here before with large partners too. insitro's release lists existing collaborations with Bristol Myers Squibb, Lilly and Gilead, and says it has 9 first-in-class programmes of its own. On 6 October 2026 it announced that machine learning models it designed for predicting how small molecules behave in the body were available in Lilly's TuneLab. One of its own programmes, described as genetic regulators of liver fat and fibrosis, is due to enter the clinic in early 2027; the release does not link that programme to the AstraZeneca work, and neither do we.
What this means for MASH patients in the UK
For patients in the UK, nothing changes now. This deal will take years to produce a candidate, if it produces one. Two MASH medicines already hold UK licences: resmetirom (Rezdiffra) and semaglutide (Wegovy), both for adults with moderate to advanced liver fibrosis, according to the charity Liver UK.
The detail matters. Writing on 6 July 2026, Liver UK reported that the MHRA authorised resmetirom in June and approved semaglutide for MASH that week, the latter with a conditional marketing authorisation that depends on further evidence from ongoing studies. Both were then going through NICE appraisal, and the charity was clear that neither would be routinely available on the NHS until that process ends.
Licence and NHS access are different decisions, made by different bodies, and an MHRA authorisation in one indication says nothing about availability in another.
Anyone with a MASH diagnosis should discuss treatment with their GP or hepatologist; nothing in this article is a recommendation for any medicine. Liver UK's explainer sets out what the approvals do and do not mean for patients.
How the deal fits AstraZeneca's other AI bets
AstraZeneca has made this kind of move several times in 2026.
The pattern is the story. Fierce Biotech, reporting the insitro deal, lists an immunology target partnership with Algen Biotechnologies, the acquisition of Modella AI and an agentic AI licensing agreement with Owkin.
Set side by side, the deals divide the work in a recognisable way. Partners such as insitro and Algen bring a platform for finding targets; AstraZeneca keeps the clinical, regulatory and commercial stages, where most of the cost and most of the risk sit. Payment follows that split: modest sums upfront, large sums only if a programme survives.
We have covered the case for and against AI in target discovery before, including how far it actually shortens research timelines. Our AstraZeneca UK listing carries the company's UK details.
What to watch next is specific. The first public signal of progress would be AstraZeneca nominating targets and insitro reaching a development candidate, a milestone the companies may or may not announce. Until then, the reliable facts are the $40 million upfront, the 2 disease areas and the division of labour. Any claim of a total deal value above the $1.5 billion the release supports should be read as an estimate, not a term.
