Potential Sciences lands UK state and Patrick Collison backing for chemistry AI
The UK startup has raised tens of millions of dollars to build physics-based AI for drug discovery, drawing a US investor connection but announcing no American expansion.
Potential Sciences, a UK startup building AI infrastructure for scientific drug discovery, has secured a seed round in the tens of millions of dollars backed by the UK government's Sovereign AI fund and Stripe chief executive Patrick Collison. The financing, reported on November 5, also brings government-backed computing capacity to a company founded a year earlier.
The company declined to disclose the exact amount raised or its valuation, according to Tech.eu. Twin Path Ventures co-led the round with Sovereign AI, pharmaphorum reported. Other investors include Odyssey, Portfolio Ventures, Blue Wire Capital, Rule30, Nauta Capital and Lightfield Equity, alongside Collison and other angel investors.
For a European company seeking to build scientific AI infrastructure, the financing combines UK public support with a prominent US-based technology executive's personal backing. It is not, however, an announced US market entry: the reports identify no American office, hiring programme, customer agreement or expansion timetable. Collison's participation is described as angel investment, not an investment by Stripe.
A bet on physics rather than the experimental record
Potential Sciences uses AI, machine learning and quantum chemistry simulations to help researchers understand how molecules behave and interact. Its proposed distinction is an AI model focused on fundamental physical laws, rather than relying on previous experimental data.
The company argues that chemistry AI faces a data bottleneck: models learn from a finite body of past experiments or from simulations that lack sufficient accuracy. Its strategy is to generate new data from physics, with calculations efficient enough to support AI development at scale.
Founder and chief executive Hatem Helal, a former Graphcore executive, established Potential Sciences in November 2025. Explaining the approach to pharmaphorum, he said: “We rebuilt quantum chemistry for GPUs so that gold-standard accuracy is affordable at the scale AI needs.”
Potential Sciences says its approach could make the AI calculations needed to design chemicals 100 times cheaper. That is a company claim about computational efficiency, not evidence of a comparable reduction in the overall cost of developing a medicine. The distinction matters: cheaper molecular calculations are the proposed enabling technology, while faster and more reliable drug discovery remains the broader ambition.
Compute alongside capital
Sovereign AI's contribution extends beyond cash. The support includes access to the AI Research Resource, the UK's national suite of advanced supercomputers, including Isambard-AI in Bristol, according to pharmaphorum. That gives the company a computing resource alongside the funding needed to develop its platform.
The investment fits the fund's stated mission to ensure that the UK develops, owns and scales important AI companies. Its other investments include chip startup Olix and AI companies Ineffable Intelligence and Callosum, Tech.eu reported.
Potential Sciences is assembling mathematicians, physicists and chemists trained at institutions including Caltech, Cambridge University and CERN. Its team also includes people who have built science-focused AI systems at Graphcore, Recursion and Dayhoff Labs, according to pharmaphorum. Those backgrounds provide context for its technical ambitions, rather than evidence of commercial partnerships with those organisations.
A transatlantic investor link, not yet an expansion
The competitive proposition is infrastructure: making accurate molecular computation cheaper so researchers can explore chemistry that is currently too expensive to investigate. Potential Sciences is positioning itself around the data and computing foundations of drug discovery, rather than announcing a particular medicine or clinical programme.
For now, the American connection sits in the investor roster. The disclosed strategy centres on building a UK-based chemistry AI capability with public compute support and private capital. Any subsequent US commercial push would be a separate development; this financing does not establish one.
