Simon Kohl is recognized in this year’s AI Power Index for being at the forefront of a transformation in science: the convergence of artificial intelligence and biology. Cole was a co-developer of AlphaFold2, a Nobel Prize-winning breakthrough that solved one of biology’s grand challenges. Cole has now turned his attention from understanding the molecular mechanisms of life to creating them. As co-founder and CEO of Latent Labs, he is advancing a vision to make biology programmable and drugs can be designed with the precision and speed of semiconductor engineering. Kohl’s platform, LatentX, achieves lab hit rates of 91 to 100 percent for macrocycles, an astonishing leap compared to the less than one percent success rate of traditional methods. The system doesn’t just predict what nature has created, it generates what nature can create, while designing molecular sequences and 3D structures in real time. Backed by investors including Google’s Jeff Dean and Cohere’s Aidan Gomez, Latent Labs is applying these capabilities to areas where traditional discovery has long stagnated, such as oncology, autoimmune diseases and rare genetic disorders.
The promise of generative AI in biology is matched by its complexity and responsibility. Cole rejects the assumption that AI will make biology “easy” and argues that the race to create novel biological systems requires new safety and governance frameworks. From DeepMind’s London lab to Latent Labs’ San Francisco wet lab, Cole’s trajectory traces the next frontier of scientific discovery: the rapidly disappearing boundaries between computation and creation.
One assumption about artificial intelligence that you think is completely wrong?
Artificial intelligence will make biology “simple” overnight. After co-developing AlphaFold2, I saw firsthand how artificial intelligence can solve extremely complex problems such as protein folding. But this means that the assumption that we can now calculate the perfect drug is wrong. Biology remains fundamentally messy. Artificial intelligence currently augments our capabilities—in potential laboratories, we are making biology programmable—but it still requires deep scientific intuition to ask the right questions and interpret what the models are telling us.
If you had to pick one moment in the last year when you thought, “Oh, crap, this changes everything for artificial intelligence,” what was that moment?
It wasn’t a single model released over the past few years, but it was when I realized that we could go beyond just predicting biological structures and actually design them from scratch. This is why I left DeepMind at the end of 2022 and founded Latent Labs – I see that we are at a turning point where generative AI can make biology truly programmable. We no longer just understand nature, we can create it with precision.
What are some of the things in AI development that keep you up at night but that most people aren’t talking about?
The gap between our ability to engineer biological systems and our ability to predict their broader consequences is growing. We can now generate novel proteins and biological circuits with unprecedented precision, but biological systems are interconnected in ways we are only beginning to understand. As we provide researchers and companies with these powerful generative tools, we need to develop equally sophisticated frameworks to test for safety, understand off-target effects, and ensure that we are not creating biological complexity beyond our control.
You co-led the Nobel Prize-winning AlphaFold2 project with DeepMind’s protein design team, and now LatentX goes beyond structure prediction to truly design entirely new proteins. Which technological breakthroughs enable the leap from predicting existing structures to creating new ones, and how does this change the drug discovery timeline?
The breakthrough lies in moving from predicting what nature creates to generating what it could create but does not. AlphaFold2 can understand existing structures, but Latent-X can jointly sample sequence and structure simultaneously—designing molecular sequences and 3D shapes in real time while following atomic-level rules. We are creating biology, not just predicting it. The impact is huge: lab hit rates are as high as 91% to 100%, compared with less than 1% for traditional methods. The scientists achieved what previously required testing millions of times on 30 candidates, turning months of experimentation into seconds of calculation.
The web-based LatentX platform allows researchers to design proteins directly in the browser, making this cutting-edge capability available to academic institutions and biotech startups. How do you balance the need to democratize this technology with ensuring its safe and responsible use, especially given potential dual-use impacts?
We envision a future where effective treatments can be designed entirely in computers, much like how space missions or semiconductors are designed today. Our platform gives scientists easy access to lab-validated protein binder designs, whether they are experts or new to artificial intelligence drug design. In our drive to democratize groundbreaking science, we take seriously the impact of dual use – actively engaging in biosafety discussions with regulators and restricting access in line with international sanctions lists. Our comprehensive approach validates everything in the San Francisco wet lab, meaning we understand real-world impacts, not just calculated possibilities. We demonstrate value first while maintaining strong safeguards.
You have achieved state-of-the-art results in lab tests of protein binding and recently raised €47.9 million with backing from prominent AI leaders such as Jeff Dean and Aidan Gomez. What specific therapeutic areas are you targeting first? As more companies enter the AI-driven protein design space, how do you see the competition evolving?
Our model is versatile in nature, enabling de novo generation of macrocycles, miniaturized binders, and antibody formats. We are passionate about the application of traditional discovery to oncology, autoimmune diseases, and rare genetic diseases. Macrocycles are exciting—combining the specificity of biologics with the oral delivery capabilities of small molecules. In head-to-head laboratory comparisons, we have surpassed previous work from large tech companies and academic labs. Our strength is combining our world-leading expertise (from experience building AlphaFold) with wet lab validation and enterprise-scale platform engineering. With the biologics market set to grow to over £1 trillion by 2033, success depends on delivering laboratory-proven results, as well as scalable engineering that meets industry safety requirements.
