Isomorphic Labs Bets $2.1 Billion on AI Engine to Redesign Drug Discovery

Isomorphic Labs is building a general-purpose AI system for drug design rather than focusing on one disease or therapeutic target. The London-based company says its models are designed to learn relationships across proteins, chemicals and biological systems, with the goal of producing medicines for many disease areas. Max Jaderberg, the company’s president, described that ambition in a Bloomberg interview as an engine that can be applied repeatedly across targets and drug modalities.

The company grew out of Google DeepMind and draws on work behind AlphaFold, the system that predicted protein structures at unprecedented scale. Its current program extends that foundation toward drug discovery, combining predictive models with generative systems that propose and refine molecular designs. Jaderberg has led that work since Isomorphic Labs was founded in 2021, according to company and Broad Institute biographies.

AI is changing the earliest stages of drug discovery by helping researchers model biological interactions and search chemical space more efficiently. Isomorphic Labs describes its technology as a unified drug-design engine capable of predicting interactions and generating candidate molecules, while NVIDIA’s account frames the company’s approach around biology as an information-processing system.

AlphaFold 3 is a key part of that lineage. The model expanded structural prediction beyond individual proteins to interactions involving proteins, nucleic acids, small molecules and other biological components. That matters because drug development depends on understanding how a candidate binds to a target, how it behaves in a wider biological system, and whether its chemical properties support development. The company says the resulting tools let researchers work on historically difficult targets at greater speed and accuracy, though laboratory experiments remain essential.

Isomorphic Labs raised $2.1 billion in Series B financing to expand its platform, grow global operations and move therapeutic candidates toward clinical development. The financing gives the company resources to build research infrastructure, hire across machine learning and drug development, and support the expensive transition from computational designs to tested medicines, according to BioPharm International.

That transition is where the drug-discovery claim meets the constraints of biology. AI can compress parts of target identification, molecular design and prioritization, but toxicology, manufacturing, clinical trials and regulatory review still require time and evidence. IEEE Spectrum noted that billions have already flowed into AI drug discovery while relatively few AI-designed medicines have reached patients. Isomorphic Labs’ financing therefore marks a test of execution: the company must show that its engine produces candidates that survive experiments and clinical development, not merely attractive predictions.

A reusable drug-design engine would change the economics of discovery by spreading model development across many programs. Instead of rebuilding a separate workflow for every target, researchers could apply shared representations of biology and chemistry, then adapt them to a particular disease or molecule. Jaderberg told Sequoia that the company’s objective is a general engine usable across disease areas and modalities.

That design also explains Isomorphic Labs’ emphasis on scientific agents. The company’s vision includes AI systems that help researchers formulate hypotheses, design experiments and interpret results, while human specialists set priorities and assess evidence. NVIDIA described the effort as building models that learn from the wider universe of protein and chemical interactions. The promise is breadth, but breadth brings a demanding validation problem: a system must work across unfamiliar targets, not only on benchmark examples or well-characterized biology.

Isomorphic Labs is part of a wider race involving technology companies, biotechnology firms and pharmaceutical partners. Its DeepMind heritage gives it access to leading AI research, while the funding round provides an unusually large financial base for an AI-native drug-discovery company. The company’s London headquarters and ties to Google DeepMind also place it among Europe’s most prominent efforts to turn frontier machine learning into biomedical products.

Jaderberg has said AI will eventually be used to design all medicines, with the open question being how quickly that happens, Bloomberg reported. That forecast is broader than the company’s present evidence. Success will be measured by validated targets, quality candidates, clinical outcomes and approved therapies. The next milestones are therefore practical: expansion of the platform, disclosure of pipeline progress, partnerships or internal programs, and proof that computational speed translates into safer and more effective medicines.