Potential Sciences exits stealth with an eight-figure round led by the UK Sovereign AI fund to train chemistry models on physics
The UK company says it trains atomistic models on data generated from the laws of physics at lab-measurement accuracy, aiming to make gold-standard chemistry data 100 times cheaper; Patrick Collison is among the angel investors.
Why it matters
Drug-discovery AI is limited by scarce, expensive experimental chemistry data, so the company generates training data by simulation grounded in physics instead. The UK government's Sovereign AI fund is both investor and compute partner, giving access to national supercomputing including the Isambard-AI machine in Bristol. The company is starting with small-molecule lead optimisation, reaction prediction and covalent drug design, working with AI-first labs. The round size is described only as eight figures. Watch for its first foundation models and any published accuracy comparisons with lab measurements.
Key facts
- Eight-figure round led by the UK Sovereign AI Fund and Twin Path Ventures
- Other investors include Nauta, Blue Wire Capital, Lightfield Equity, Odyssey, Portfolio Ventures and Rule30
- Angels include Patrick Collison, Chris Gibson and Tim Sadler
- Aims to make gold-standard chemistry data 100 times cheaper
- Sovereign AI Fund also gives compute access, including the Isambard-AI supercomputer in Bristol