Isomorphic Labs Enters Human Trials With AI-Designed Drugs, Testing Industry’s $3 Billion Bet
DeepMind spinoff's 2026 clinical trials mark first major test of whether machine learning can genuinely reduce drug failure rates beyond early-stage screening.
Isomorphic Labs, the DeepMind spinoff founded by Nobel laureate Demis Hassabis, will enter human clinical trials in 2026 with its first AI-designed drug candidates—the highest-profile test yet of whether machine learning can translate promising early-stage results into sustained efficacy across full pharmaceutical development pipelines.
The company’s February 2026 release of IsoDDE, a proprietary drug design engine that more than doubles AlphaFold 3’s performance in predicting protein-ligand binding structures, positions AI-discovered molecules at a critical inflection point. Early data shows AI-designed drugs achieving 80-90% Phase I success rates versus 40-65% for traditional methods, per a ScienceDirect analysis of trials completed through December 2023. The question is whether this advantage persists through later development stages, where only 10% of all drug candidates ultimately secure regulatory approval.
From Prediction to Production
Isomorphic’s clinical trial entry represents a maturation from theoretical capability to real-world pharmaceutical application. The company, which now employs 348 people, has secured partnerships with Eli Lilly and Novartis worth nearly $3 billion in potential milestone payments, according to Clinical Trials Arena. Multiple drug candidates remain in preclinical development, with the first Phase I trials prioritizing oncology and immunological diseases.
The $600 million Series A round led by Thrive Capital in March 2025—the company’s first external funding since its 2021 founding—accelerated development of IsoDDE and staffing for clinical operations. “The next big milestone is actually going out to Clinical Trials, starting to put these things into human beings,” Colin Murdoch, Isomorphic’s president, told Fortune in July 2025. “We’re staffing up now. We’re getting very close.”
“This funding will further turbocharge the development of our next-generation AI drug design engine, help us advance our own programmes into clinical development, and is a significant step forward towards our mission of one day solving all disease with the help of AI.”
— Demis Hassabis, CEO of Isomorphic Labs
The IsoDDE Advantage
IsoDDE’s technical capabilities distinguish it from earlier AI drug discovery tools. The engine identifies unknown docking sites on proteins in seconds with laboratory-grade precision while estimating binding strength at a fraction of traditional costs, WinBuzzer reported in February 2026. This represents more than incremental improvement over AlphaFold 3—the protein structure prediction system that earned Hassabis and John Jumper the 2024 Nobel Prize in Chemistry.
The performance gains matter because binding prediction accuracy directly correlates with clinical success probability. Traditional drug discovery screens millions of compounds through iterative cycles of synthesis and testing, a process that typically requires 10-15 years and costs exceeding $2 billion per approved drug. AI-accelerated discovery compresses early-stage screening from years to months, but the critical question is whether computational optimization genuinely identifies more effective therapeutic candidates or simply accelerates selection of molecules that face the same failure rates in human testing.
| Phase | Traditional | AI-Accelerated |
|---|---|---|
| Target Identification | 2-3 years | 3-6 months |
| Lead Optimization | 3-5 years | 6-12 months |
| Preclinical Testing | 1-2 years | 12-18 months |
| Clinical Trials (Phase I-III) | 6-7 years | 6-7 years (unproven acceleration) |
The Clinical Reality Check
The field of AI drug discovery has expanded from three drugs in clinical trials in 2016 to 75 molecules by June 2024, according to Drug Discovery Trends. Yet no AI-discovered drug has achieved full FDA approval, leaving the technology’s ultimate value proposition unproven. Isomorphic’s trials—backed by partnerships with two of the pharmaceutical industry’s largest players—represent the most significant test of whether AI can genuinely improve the 10% overall approval rate for drugs entering clinical development.
If AI-designed drugs maintain their 80-90% Phase I success rates through Phase II and III trials, the technology could raise overall approval rates to 9-18%, fundamentally reshaping pharmaceutical economics. But Phase I trials, which primarily test safety in small cohorts of healthy volunteers, represent the lowest bar in drug development. Phase II efficacy trials and Phase III large-scale studies eliminate most candidates, and AI’s predictive advantage may not translate to these later stages where biological complexity increases.
Isomorphic Labs operates as an independent subsidiary of Alphabet, leveraging DeepMind’s computational infrastructure while maintaining separate leadership and commercial partnerships. Hassabis serves as CEO of both organizations. The company’s business model combines proprietary drug development with fee-based partnerships, allowing pharmaceutical clients to access IsoDDE while Isomorphic retains economics on internally developed candidates.
What to Watch
The first Phase I readouts, expected within 12-18 months of trial initiation, will provide initial safety signals but limited efficacy data. More significant milestones include Phase II trial launches (likely 2027-2028) and comparative analysis of how Isomorphic’s candidates perform relative to traditionally discovered drugs targeting the same biological pathways. Investors and pharmaceutical partners will monitor whether AI-optimized molecules show differentiated safety profiles, target selectivity, or pharmacokinetic properties beyond what binding prediction alone would suggest. The economics of Isomorphic’s partnerships—structured around milestone payments tied to clinical progression—create strong incentives for rapid advancement but also pressure to demonstrate sustained performance through later development stages where the majority of drug candidates fail.