HomeFull speed ahead. Keep track of advancements in every frontier

TheTechno Optimist.

Isomorphic Labs—Drug design engine and pipelineIsomorphic Labs—

Drug design engine and pipeline

by Isomorphic LabsGB

ResearchStage 1 of 5

Own and partner drugs in lab stage; first human trials expected by end of 2026.

Updated 10 Feb 2026Checked 25 Sep0 updates this week

Milestones

Next · First AI-designed drug in clinical trials
  1. Lilly and Novartis partnershipsJan 2024Complete.
  2. AlphaFold 3 published with DeepMindMay 2024Complete.
  3. First AI-designed drug in clinical trialsTarget Dec 2026+12 mooriginally Dec 2025Current milestone.

Most important updates

  • 10 Feb 2026
  • Jan 2024

Upcoming

  1. Dec 2026First AI-designed drug in clinical trials+12 mooriginally Dec 2025in about 7 weeks (next)

Current obstacles

  • Not yet tested in peopleNo Isomorphic drug has been tested in humans, so it's unproven whether AI improves success rates.
  • Too little binding dataMeasured data on how well drugs bind is far scarcer than shape data, which limits accuracy.

Physics limits

  • Too little good data about how drugs bindOnly about 230,000 molecular structures have been solved by experiment, and far fewer have reliable binding measurements. Models trained on that are weakest on exactly the novel chemistry drugmakers want.
  • Binding a target is not the same as a safe, working drugWhether a molecule is absorbed, reaches the right tissue, avoids toxicity and actually changes the disease depends on whole-body biology that no current model predicts. About 90% of drugs entering trials fail.
  • Molecules move; static snapshots miss muchProteins flex and water shifts around a drug, and binding strength depends on those motions. A 1.4 kcal/mol error means a 10-fold error in binding, and today's predictions often miss by that much.

How it works

3 parts
Mounting a protein crystal for X-ray analysis: the kind of lab measurement that checks AI-designed drugs and trains the models
Mounting a protein crystal for X-ray analysis: the kind of lab measurement that checks AI-designed drugs and trains the modelsPhoto: North Sullivan Photography, CSIRO · CC BY 3.0 (opens commons.wikimedia.org)
Target

Model the target

Predict the 3D shape of the disease protein, including the shapes a drug could grab.

Design

Invent candidates

AI proposes molecules shaped to fit, filtered by how well they should bind and behave in the body.

Validate

Test in the lab

The best are made and tested; results teach the AI for the next round.

Update log

2 updates

Tue 10 Feb

  • Minor: BlogSoftware

Jan 2024

  • Major: BlogPartnership

About Isomorphic Labs

The team behind Drug design engine and pipeline

Isomorphic Labs

AI drug design, AlphaFold 3

Alphabet company spun out of Google DeepMind in 2021, led by Demis Hassabis. It co-built AlphaFold 3, which predicts how proteins, DNA and drugs fit together, and uses it to design drugs with Lilly, Novartis and for itself. First trials expected by end of 2026.

  • Raised its first outside money, $600M led by Thrive Capital, in Mar 2025, then a reported $2.1B in May 2026.
  • The goal of AI-designed drugs in trials by end of 2025 slipped; now expected by end of 2026.
  • Its Lilly and Novartis deals (Jan 2024) could pay up to about $3B if goals are hit.

isomorphiclabs.com (opens isomorphiclabs.com)Company profile →

Founded
20215 yrs
Headquarters
United Kingdom
Status
Subsidiaryof Alphabet
Valuation
Alphabet-owned
Raised
$2.7B2 rounds
Last round
Series B · $2.1BMay 2026
Works in
BiotechAI drug discoveryAI
Coverage
2 programs · 7 updateslatest 1 May 2026checked 25 Sep
Lead investors
Thrive Capital
Customers
Eli LillyNovartis (opens novartis.com)
Partners
Google DeepMind (opens deepmind.google)
Investors
Thrive Capital (opens thrivecap.com)
People
Demis Hassabis (opens isomorphiclabs.com)Founder and CEO
Primer