AstraZeneca Signs Up to $1.5 Billion AI Target-Discovery Deal With insitro in MASH and Cardiomyopathy

What happened

AstraZeneca has formed a partnership with insitro, an AI-focused drug discovery company, to find new targets in metabolic dysfunction-associated steatohepatitis (MASH) and cardiomyopathies. BioSpace reported the agreement on Oct. 15, saying insitro will receive $40 million upfront and could earn up to $1.46 billion in milestones, plus royalties, for a total potential value of up to $1.5 billion.

Under the deal, insitro will use its Virtual Human AI platform to identify causal disease biology, and AstraZeneca will advance resulting candidates through development.

Deal details

• Upfront: $40 million.
• Milestones: Up to $1.46 billion.
• Royalties: On any resulting products.
• Disease areas: MASH and cardiomyopathies.
• Roles: insitro handles target discovery; AstraZeneca leads development.

insitro has partnered with several large pharma companies, including Bristol Myers Squibb and Eli Lilly, to build a pipeline of AI-generated candidates, according to BioSpace. The company laid off 22% of its staff in May 2025 to extend its cash runway amid what it called macroeconomic uncertainty.

“One of the most critical decisions in R&D is selecting the right target,” said Sharon Barr, executive vice president and head of BioPharmaceuticals R&D at AstraZeneca.

AstraZeneca’s context

BioSpace placed the deal against recent setbacks in AstraZeneca’s programs in both areas:

• MASH: AstraZeneca’s earlier candidate AZD2693, partnered with Ionis, was discontinued in November 2025 after a Phase 2 trial fell short. Its current work includes AZD2389, a Phase 1 anti-fibrotic.
• Cardiomyopathy: Ionis-partnered Wainua failed a Phase 3 trial in transthyretin-mediated amyloid cardiomyopathy. AstraZeneca is also developing AZD4063 in Phase 1 for R14del dilated cardiomyopathy.

Why it matters for the sector

Target selection is one of the main reasons drug programs fail. Many candidates that work well in laboratory models do not change the course of disease in patients because the underlying biology was not causal. AI platforms that integrate human genetic, cellular and clinical data aim to improve the odds by pointing to targets with stronger evidence of a role in disease.

MASH, a progressive liver disease linked to obesity and diabetes, has attracted intense interest after the first approvals for the condition, and large drugmakers are seeking differentiated mechanisms. Cardiomyopathies, which weaken the heart muscle, are similarly the focus of new genetic and precision approaches.

The deal is one of several AI-related pharma partnerships announced in October. BioSpace also reported Zealand Pharma’s agreement worth up to $1.2 billion with Generate Biomedicines for AI-designed obesity and metabolic drugs on Oct. 14. The deal structures share common features: modest upfront payments and milestone potential above $1 billion.

For insitro, the AstraZeneca agreement adds another large pharma partner after its 2025 restructuring, providing non-dilutive funding and validation for its platform as it competes with other AI-enabled discovery companies for partnerships.

For AstraZeneca, outsourcing early target discovery to a specialist allows it to explore new biology in areas where its internal programs have stalled, without committing large sums before targets are validated.

AI target-discovery partnerships typically take years to produce clinical candidates, and their success is measured by how many validated targets move into development. Pharma companies often sign several such agreements in parallel, comparing platforms before committing deeper resources to the ones that deliver.

The upfront payment is small relative to the headline value, so the economic outcome for insitro will depend largely on whether AstraZeneca advances the targets it identifies.

What to watch

• Which targets emerge from the collaboration and when AstraZeneca selects candidates for development.
• Clinical progress of AZD2389 in MASH and AZD4063 in dilated cardiomyopathy.
• insitro’s other partnerships and any new financing.
• Further AI discovery deals in cardiometabolic disease.

Sources: BioSpace

The Two Disease Areas in Brief

MASH

Metabolic dysfunction-associated steatohepatitis, previously known as NASH, is the more serious form of fatty liver disease. It begins with fat building up in the liver, often alongside obesity, type 2 diabetes or high cholesterol. In some people, that fat triggers inflammation and liver cell injury, which can lead to scarring, called fibrosis. Over years, fibrosis can progress to cirrhosis, liver failure or liver cancer. Because MASH often causes few symptoms until it is advanced, many people do not know they have it.

Cardiomyopathies

Cardiomyopathy is a general term for diseases of the heart muscle that make it harder for the heart to pump blood. Common types include dilated cardiomyopathy, in which the heart’s main chamber enlarges and weakens; hypertrophic cardiomyopathy, in which the muscle thickens; and amyloid cardiomyopathy, in which misfolded proteins deposit in the heart. Many forms have genetic causes, which makes them a natural fit for approaches built on human genetic data.

Anyone with concerns about liver or heart health should talk with their physician.

What “Causal Biology” Means in Target Discovery

A biological change can be associated with a disease without causing it. Inflammation markers, for example, may rise in a sick organ because of the damage rather than driving it. Drugs aimed at such bystanders tend to fail. Platforms focused on causal biology try to distinguish drivers from passengers, typically by combining several kinds of evidence:

1. Human genetics. If people born with variants that reduce a gene’s activity have lower disease risk, blocking that gene’s product may be protective. Nature, in effect, has run the experiment.
2. Cellular models. Cells derived from patients, such as stem cells turned into liver or heart cells, can show how a genetic change alters cell behavior.
3. Clinical and imaging data. Machine learning applied to scans, biopsies and health records can reveal patterns linked to disease progression.
4. Integration. Models combine these sources to rank targets by the strength and consistency of the evidence.

Even strongly supported targets can fail for reasons such as safety or the difficulty of designing a suitable drug, which is why milestone payments are tied to later progress.

Why Target Choice Shapes Program Costs

Most of the money in drug development is spent in mid- and late-stage clinical trials. A program built on a weak target can consume years and large budgets before failing in Phase 2 or Phase 3. Improving the odds at the very first decision, which target to pursue, can therefore have an outsized effect on research productivity. That is the logic behind paying for discovery partnerships long before any molecule exists.

Questions Readers Ask

Why did NASH become MASH?

Medical societies adopted new terminology to describe the disease by its metabolic causes and to move away from wording considered imprecise or stigmatizing.

What happens after insitro identifies a target?

AstraZeneca would typically design or screen for molecules against it, run preclinical studies and, if successful, begin clinical trials, with milestones triggered along the way.

Why do companies sign several AI deals at once?

Running parallel partnerships lets a drugmaker compare platforms and spread risk before committing deeper resources.

This article is for general information only and is not medical or investment advice.