There are thousands of diseases that science understands well at the molecular level. Researchers know which protein is causing the problem. The obstacle is that the protein has no obvious place where a drug can latch on. These targets are called "undruggable," and they account for a large share of conditions where patients still have no good treatment options.
Isomorphic Labs was founded in 2021 as a direct spinoff from Google DeepMind, built on the same AI research that won the Nobel Prize in Chemistry for solving a decades-old puzzle about how proteins fold into their shapes. That Nobel work was a significant scientific achievement, but it was not enough on its own to design drugs. Knowing a protein's shape is step one. Finding where a drug can bind, predicting how tightly it binds, and modeling how the drug behaves in the rest of the body are entirely different problems.
Isomorphic's new engine, called IsoDDE, attempts to tackle all of those steps together. One key capability is identifying "cryptic pockets," places on a protein's surface that look completely flat and inaccessible in normal conditions but open up when exactly the right molecule is nearby. As a validation test, the team used IsoDDE to predict the location of a hidden pocket on a cancer-related protein called cereblon. A Nature paper published in January 2026 had just disclosed that pocket for the first time ever. IsoDDE found it using only the protein's sequence as input, with no prior knowledge of what the Nature paper would reveal.
This matters because the standard critique of AI in drug discovery is that the systems work well on familiar territory and fall apart when the biology is genuinely new. A tool that only confirms what scientists already know has limited value. A tool that reliably finds things no one had seen before is a different story.
The funding picture reflects serious institutional conviction. <br>The company's $2.1 billion Series B, raised in May 2026, was led by Thrive Capital and included Alphabet, Singapore's Temasek, and the UK Sovereign AI Fund. Combined with a $600 million Series A from March 2025, total outside capital stands at roughly $2.7 billion. The Novartis collaboration was expanded in February 2025 after just over a year, which is a meaningful signal: Novartis saw enough early progress to add more programs. The Eli Lilly deal, signed in January 2024, carries potential milestone payments of up to $1.7 billion.
The honest context is that no AI-discovered drug has received regulatory approval anywhere in the world as of mid-2026. The pipeline is real but early. Isomorphic has said it plans to bring oncology candidates into clinical trials, but no named candidates or trial dates have been publicly confirmed. Clinical trials alone typically take seven to ten years on top of discovery work, and roughly nine out of ten drug candidates that enter human trials never reach approval.
What AI can genuinely accelerate is the discovery phase, which historically takes three to six years and produces many dead ends. AI systems can scan far more potential drug candidates in far less time, and Isomorphic claims its engine can compress that search while also expanding it into territory traditional chemistry could not reach.
For anyone outside the pharmaceutical sector, the practical takeaway is this: the companies supplying the ingredients, logistics, manufacturing, and services to the pharma industry should expect AI tools to start changing what gets developed and how quickly. Drug targets that were abandoned for years because they seemed unreachable are back on the table. Programs that might have taken a decade to reach clinical testing could arrive faster. That does not change what you do tomorrow, but it does change the medium-term pipeline of new treatments, and with it the commercial opportunities and pressures across healthcare supply chains globally.