Isomorphic Labs: AI Drug Design Engine Approaches Clinical Trials

What does it take to move an industry from trial-and-error to computation-first?

In 2021, Demis Hassabis did something unusual. After leading DeepMind to its most celebrated breakthrough — AlphaFold, the AI system that solved protein structure prediction — he spun out a separate company. It would take the same deep learning toolkit and aim it directly at drug discovery. Not as a side project within Google. A standalone biotech.

Five years and $2.7 billion in funding later, the experiment is about to face its first real test.

The Engine Beyond AlphaFold

AlphaFold 3, released in 2024, could predict the three-dimensional structure of proteins, DNA, RNA, and small molecules with unprecedented accuracy. It was a breakthrough for structural biology. But structure alone doesn't design a drug. Knowing what a protein looks like doesn't tell you which molecule will bind to it, how tightly, or whether the resulting complex will be stable in a human body.

It built IsoDDE — the Isomorphic Drug Design Engine — to bridge that gap. It is a unified AI system that predicts protein-ligand structures, antibody-antigen interfaces, binding affinities, and identifies novel binding pockets from amino acid sequences alone. On the Runs N' Poses benchmark, designed to test performance on structures far from training data, IsoDDE achieved a 50% success rate on the hardest cases. AlphaFold 3 scored 23.3%.

Hassabis has framed the engine as a repeatable method rather than a single-model trick. "Now that we have shown our approach is fundamentally sound," he said after the Series B, "our focus is on scaling our technology to its full potential."

$2.1 Billion on the Table

The scale of the Series B is itself a signal. Thrive Capital led the round, joined by Alphabet, GV, MGX, Temasek, CapitalG, and the UK Sovereign AI Fund. At $2.1 billion, it is the second-largest biotech fundraise in history, behind only Altos Labs' $3 billion emergence in 2022. Combined with the $600 million Series A in March 2025, Isomorphic has raised more capital before its first clinical data point than most biotechs raise in a lifetime.

The money goes toward three things: scaling IsoDDE for biopharma partners, advancing an internal pipeline in oncology and immunology, and hiring across London, Cambridge (Massachusetts), and Lausanne. The company already has active partnerships with Novartis, Eli Lilly, and Johnson & Johnson — each structured with upfront payments and contingent milestones that, combined, exceed $3 billion in potential value.

This funding round is a massive vote of confidence from a diverse group of top-tier international investors in our AI-first approach to drug design and development.— Sir Demis Hassabis, Founder and CEO, Isomorphic Labs

Ruth Porat, president and chief investment officer at Alphabet and Google, said Isomorphic has "already made extraordinary progress in harnessing AI to accelerate drug discovery." The question is whether that progress translates into drugs that work in human patients.

The Clinical Trial Reckoning

Isomorphic currently has 17 drug programs underway. In January 2026, at the World Economic Forum in Davos, Hassabis confirmed that the first AI-designed candidate — an oncology drug — is scheduled to enter Phase 1 trials before the end of 2026. This is a revision from an earlier target of end-2025, a delay that reflects the gap between benchmark performance and clinical readiness.

Phase 1 tests safety and dosing in a small patient cohort, not efficacy. The result that matters commercially is whether the compound advances to Phase 2 and, eventually, proves more effective than molecules discovered through standard screening. AI-designed drugs have reached the clinic before — Insilico Medicine's fibrosis drug is in Phase II — but no company has yet shown that AI-guided discovery improves Phase I-to-II progression rates at scale.

As the biotech analyst Ben Zercher of PitchBook put it, "Rounds like these give tech capital a place to go in biotech, while the rest of the industry stays focused on getting drugs to patients." Isomorphic's funding is large enough to carry it through a multi-year trial arc. Whether AI-directed design actually compresses that timeline is the data point the entire pharmaceutical industry is watching.

The Convergence Behind the Hype

Isomorphic sits at the intersection of two trends that have each attracted enormous capital independently. The first is the maturation of generative AI for molecular science — diffusion models that can design novel proteins, transformers that predict binding affinities, reinforcement learning loops that optimize candidate molecules in silico. The second is the biotech industry's willingness to fund platform companies before they have products, a pattern that accelerated after the pandemic and has produced a string of billion-dollar AI-biotech raises.

As we wrote in July, Insitro is another AI-native platform using machine learning to reshape pharmaceutical R&D. The difference with Isomorphic is the depth of its technical foundation: where most AI-biotech startups build on top of publicly available models, Isomorphic develops its own from the ground up, with direct lineage to the AlphaFold work that won the 2024 Nobel Prize in Chemistry. The difference with Isomorphic is the depth of its technical foundation: where most AI-biotech startups build on top of publicly available models, Isomorphic develops its own from the ground up, with direct lineage to the AlphaFold work that won the 2024 Nobel Prize in Chemistry.

Whether that lineage produces better clinical outcomes is an open question. But the company has placed a specific, testable bet: that a unified AI engine can discover molecules that human researchers, working with conventional tools, would not find. By the end of 2026, we will start to see the answer.

Sources