AstraZeneca turns to insitro’s AI platform for MASH and cardiomyopathy target discovery
AstraZeneca has struck a broad collaboration with insitro that places artificial intelligence at the center of early drug discovery for metabolic dysfunction-associated steatohepatitis and cardiomyopathies. The agreement carries $40 million upfront and up to $1.46 billion in development and commercial milestones, a scale that signals how heavily large pharmas now value computational target selection before a single pivotal trial is launched.
The partnership is built around insitro’s platform, which combines human and cellular data with AI-enabled target identification to surface causal disease biology. In practical terms, the work is aimed at moving beyond associations and toward mechanisms that can be translated into drug programs. For diseases like MASH and cardiomyopathy, where heterogeneity can blur the path from biomarker to intervention, the promise lies in finding targets tied to disease drivers rather than downstream noise.
Causal biology is the point of the collaboration
MASH and cardiomyopathies sit in a part of medicine where prevalence is high, diagnosis is often delayed, and therapeutic options remain limited relative to need. In that setting, target discovery is not simply an academic exercise. It determines whether a program lands on a mechanism with a plausible chance of changing fibrosis, metabolic injury, or myocardial dysfunction, or whether it spends years chasing a biological correlate that never translates into benefit.
Daphne Koller, insitro’s chief executive, framed the collaboration around that logic, saying the two disease areas are prevalent and underdiagnosed and that high-content data can reveal disease drivers. The phrasing aligns closely with what drug developers want from AI-enabled discovery: not more data for its own sake, but a way to interrogate complex datasets for genetic and cellular signals that point to actionable biology.
AstraZeneca will hold exclusive worldwide rights to advance selected programs emerging from the collaboration. That structure matters because it links target discovery directly to the company’s downstream clinical development and regulatory infrastructure. The value of AI in drug discovery often rises or falls on whether a novel target can be moved efficiently into medicinal chemistry, translational studies, and eventually human testing. A platform that finds a target but cannot support progression adds little to a portfolio under pressure for speed and probability of success.
A strategic extension of AstraZeneca’s AI push
The deal fits into AstraZeneca’s broader effort to embed AI across research and development. The company has already invested in internal capabilities and has paired with Algen Biotechnologies on immunology target discovery, acquired Modella AI, and entered an agentic AI licensing agreement with Owkin. The insitro collaboration adds another layer to that strategy, but in a therapeutically distinct area where liver and cardiac disease biology may offer substantial room for novel target identification.
The move also reflects a wider recalibration across big pharma. AI is no longer being treated solely as a back-end analytics tool or a screening aid. It is increasingly being positioned upstream, where target selection can shape the full economic and clinical trajectory of a program. If the early biology is wrong, every later stage becomes more expensive and less likely to succeed. If the biology is right, the rest of the development path becomes more defensible.
One sentence captures the commercial logic plainly: this is a bet on better entry points into disease.
insitro gains capital and a larger validation signal
For insitro, the collaboration adds another large pharmaceutical partner to a platform already used by several major companies. Gilead entered a deal with the company in 2019, Bristol Myers Squibb followed in 2020, and that BMS relationship has been extended and expanded over time. Eli Lilly partnered with insitro in 2024, and the platform was later made available to select biotechs in 2025. Each new agreement broadens the company’s external validation while also supporting the capital base needed to build its own pipeline.
That pipeline now spans liver fat and fibrosis, obesity, neuroscience, and ophthalmology. The financing profile described in the article suggests a familiar biotech pattern: upfront cash provides operating flexibility, while the larger value sits in back-loaded milestone payments tied to scientific and commercial execution. For insitro, the immediate benefit is not just revenue. It is continued access to resources that can sustain platform development while partner programs move through early discovery.
A useful way to read the deal is as a test of whether AI can make target discovery more empirically grounded in diseases that are biologically complex and clinically consequential. MASH, in particular, has become a major area of drug development because of its broad burden and the difficulty of identifying patients early enough to intervene before irreversible liver damage occurs. Cardiomyopathies present a parallel challenge in the heart, where multiple pathways can lead to structurally and functionally distinct disease states. A discovery platform that can sort among those mechanisms may be more valuable than one that simply increases the volume of candidates.
What this means for the next wave of discovery programs
The collaboration does not disclose specific targets, development timelines, or clinical endpoints. Those details will determine whether the partnership produces first-in-class programs, best-in-class follow-ons, or simply another expensive discovery exercise. For now, the significance lies in the choice of therapeutic areas and in AstraZeneca’s willingness to place substantial capital behind AI-enabled causal inference at the front end of the pipeline.
The downside risk is familiar to anyone watching the sector: strong computational output does not guarantee translational success. Human biology is still full of false leads, and disease networks in MASH and cardiomyopathy are unlikely to yield to a single elegant model. Even so, the collaboration shows where the industry’s center of gravity is shifting. Discovery is moving closer to data-rich phenotypes, genetically anchored hypotheses, and platform-driven decisions about which targets deserve the long march into development.
For AstraZeneca, the deal broadens an already active AI portfolio. For insitro, it brings another heavyweight validation of a business built on linking cellular and human data to target selection. The clinical value will ultimately depend on whether that approach delivers programs robust enough to survive the next stages of development, when mechanistic confidence must give way to human evidence.