AI
UK Startup Proposes Using Physics Simulations to Leapfrog Drug Discovery Labs
Potential Sciences wants to replace years of bench work with AI trained on synthetic molecular data, drawing early backing from the British government's sovereign AI fund.
WIRED — AI
A fresh take on accelerating pharmaceutical development is emerging from the UK: instead of relying on traditional laboratory experimentation to design new treatments, Potential Sciences aims to use artificial intelligence trained on vast simulated datasets to identify promising drug molecules before they ever reach a bench.
The startup, which officially began operations in early November, draws on physics-based simulations of atomic and molecular interactions to generate large collections of synthetic data. These datasets would then fuel machine-learning systems capable of predicting which candidate molecules will successfully target disease-related proteins—dramatically narrowing the search space within which scientists work.
A New Approach to an Intractable Problem#
Drug discovery remains notoriously slow and expensive. Founder and CEO Hatem Helal frames the challenge in stark terms: with roughly 10^60 possible ways to arrange atoms into molecules suitable for human administration, the field operates as an enormous search problem. Conventional methods rely on painstaking laboratory work to test whether each potential compound actually does what researchers hope.
By building physics-driven simulations of molecular behavior, Potential Sciences sidesteps this bottleneck. Rather than testing candidates one by one, the company aims to train predictive models on millions of synthetic interactions, allowing algorithms to identify the most promising candidates before experimental validation begins.
"We want molecules that are never going to work in the clinic to fail fast," Helal tells WIRED. The bet is that better computational prediction of how drug candidates interact with human biology will save time, cut costs, and make it economical to pursue treatments for rare diseases previously considered unprofitable.
Backing and Market Context#
The startup has secured seed funding in the tens of millions from SovAI, the UK government's sovereign AI venture capital fund, alongside Twin Path Ventures and other investors. SovAI, established with roughly $800 million to cultivate British companies essential to the global AI supply chain, is also granting Potential Sciences access to the UK's national AI supercomputer—a significant computational advantage for early-stage work.
Potential Sciences joins other AI drug discovery firms—including Basecamp Research and Isomorphic Labs—in receiving SovAI backing. The crowded competitive landscape reflects genuine optimism that machine learning will reshape how medicines are designed, building on breakthroughs already seen in protein structure prediction.

