UK Sovereign AI fund co-leads eight-figure seed round for Potential Sciences

Potential Sciences, a one-year-old Cambridge company building AI models for medicine design, came out of stealth on 5 November 2026 with an eight-figure seed round co-led by the UK government's Sovereign AI fund and Twin Path Ventures¹². The government is also giving the company compute on the AI Research Resource, the national supercomputing suite that includes Isambard-AI in Bristol¹.

What the company is building

Most AI models in drug discovery learn from the results of past laboratory experiments, which are slow and costly to generate. The government's announcement describes this dependence on experimental records as the brake on progress¹. Potential Sciences instead trains its model on the fundamental laws of physics, working out interactions within molecules down to the atomic level². The company argues that chemistry models today are short of data because they all draw on the same finite record of past experiments, or on simulations too crude to trust².

The claimed advantage is cost. Potential Sciences says its approach could make the AI calculations needed to design new drugs 100 times cheaper than at present, which would make work in currently uneconomical fields viable¹². Founder and chief executive Hatem Helal said the company had "rebuilt quantum chemistry for GPUs so that gold-standard accuracy is affordable at the scale AI needs"¹. He added that with code 100 times more efficient, "a seed round buys what used to take a dedicated datacenter"¹.

The team brings together mathematicians, physicists and chemists trained at Caltech, Cambridge University and CERN, including people who built science-focused AI systems at Graphcore, Recursion and Dayhoff Labs². Alongside Sovereign AI and Twin Path Ventures, the investors are Blue Wire Capital, Lightfield Equity, Nauta Capital, Odyssey, Portfolio Ventures and Rule30². Angel backers include Stripe co-founder Patrick Collison and Recursion co-founder and former chief executive Chris Gibson, and AI engineer Andrew Fitzgibbon and entrepreneur James Hawkins are advising the company¹.

What it means for the market

For drug discovery teams, the pitch goes to a recognised constraint: AI design tools are only as good as the training data behind them, and generating new experimental data is expensive. Helal said AI for drug discovery "has learned almost everything the existing experimental record can teach it"¹. If physics-derived data can be produced cheaply enough, the bottleneck shifts from laboratory capacity to computing capacity. AI Minister Kanishka Narayan said that solving the data bottleneck in chemistry could let researchers explore treatments for diseases "currently deemed too rare or expensive to focus on"¹.

The round also shows how the UK state is now acting as an early-stage investor in AI for science. The Sovereign AI fund has up to £500 million behind it². The government said the deal came a week after the Budget, in which the Chancellor confirmed a further £140 million for Sovereign AI: £60 million to support the fund's investments and £80 million for its R&D Procurement Scheme¹. The fund says typical equity investments will be around £1 million to £10 million, and it requires the companies it backs to have a meaningful presence in the UK¹. The package pairs equity with access to national supercomputers¹.

For investors and competing AI-native biotechs, the Recursion links among staff and backers place Potential Sciences in the same field as the platform companies that have built large drug discovery businesses on experimental data, though with an opposite thesis on where the next gains come from¹².

Caveats

The 100-fold cost figure is the company's own claim, and neither source gives independent validation or published benchmarks¹². The exact size of the round is not disclosed beyond being eight figures, and the government does not comment on its individual investment amounts, citing commercial sensitivity¹. Neither source names a drug programme, a disease target or a pharma partner, so the platform is at a pre-pipeline stage, and the government's own description of the benefits is conditional throughout¹.