AI in Drug Discovery: How Isomorphic Labs, Recursion, and Others Are Reshaping Pharma

AI in Drug Discovery: How Isomorphic Labs, Recursion, and Others Are Reshaping Pharma

Isomorphic Labs moved two AI-designed molecules into Phase I trials this spring. The moves mark a shift from early validation to human testing.

AlphaFold 3 expanded ligand docking accuracy. Companies now run larger virtual screens with tighter binding predictions.

Big Pharma signed multiple new research deals in 2025 and 2026. The contracts give smaller AI firms capital and late-stage support.

Isomorphic Labs Advances Two Candidates

Isomorphic Labs disclosed Phase I starts for programs targeting oncology and immunology. Both molecules came from its AI platform in under 18 months of design work.

The company shared that each program cleared standard preclinical safety packages. Regulators accepted the IND filings on first review.

Partners at Novartis and Lilly funded the early work. Those deals now move into milestone payments tied to trial progress.

The timeline from target selection to first patient dosing sits at roughly 24 months.

Recursion Updates Its Clinical Portfolio

Recursion reported new trial data from an AI-derived fibrosis candidate. The results showed dose-dependent target engagement in a small patient cohort.

The company also added two new programs from its map of cellular phenotypes. One targets rare metabolic disease and the other an inflammatory indication.

Recursion's recent partnership with Roche added $150 million in near-term research funding. The agreement covers three disease areas.

Pipeline updates now appear quarterly on the company website.

AlphaFold 3 Changes Screening Scope

AlphaFold 3 improved joint prediction of proteins and small molecules. Teams report higher hit rates in virtual screens of 10 million compounds.

Early users note the model reduces false positives compared with prior docking tools. Fewer compounds move to physical synthesis.

Academic groups and several startups published validation sets in Nature and Science this year. The papers compare model scores to measured affinities.

The improved accuracy lets smaller teams run screens once limited to large pharma compute budgets.

2025-2026 Partnership Deals

Eli Lilly extended its arrangement with Insitro. The new tranche funds work on metabolic disease targets.

Sanofi signed a multi-target agreement with Exscientia. The deal includes options for co-development after lead selection.

Amgen increased its stake in a smaller AI platform through a new series B. The investment supports an internal pipeline.

Most contracts now tie payments to clinical proof rather than discovery milestones alone.

Realistic Timelines for First Approvals

Analysts place the earliest possible AI-first approval in 2028 or 2029. That assumes clean Phase III data from the current leading programs.

Regulatory agencies have begun to publish draft guidance on AI-designed molecules. The documents focus on validation of the underlying models.

Most companies still plan traditional toxicology packages. Regulators have not yet accepted fully in silico safety claims.

Observers track the first pivotal trial readouts expected in late 2027.

Uncertainties That Remain

It is still unclear whether AI predictions will hold across diverse patient populations. Early trials use narrow inclusion criteria.

Manufacturing routes for some AI-designed structures also need scale-up work. Unusual scaffolds can raise cost or yield issues.

Data sharing between partners remains limited. Many agreements keep training sets proprietary.

Success will depend on consistent trial outcomes rather than platform marketing.

What to Watch Next

Check for first patient data from the Isomorphic Phase I studies in the third quarter of 2026. Positive safety signals would support the 24-month timeline claim.

Watch for additional trial starts from Recursion or Exscientia before year end. New IND filings would show continued pipeline momentum.

Follow FDA or EMA feedback on any AI-specific regulatory filings later this year. Early guidance would shape later programs.

Track partnership announcements from the remaining large pharma groups still without visible AI discovery deals. The next contracts may set new pricing benchmarks.