Potential Sciences has raised seed funding with backing from the UK government's Sovereign AI fund and Stripe CEO Patrick Collison. The company did not disclose the amount. The UK company builds AI infrastructure for chemistry and drug discovery, using quantum chemistry simulations to show how molecules behave and interact. Dr Hatem Helal, a former Graphcore executive, founded the company in November 2025.
Evertrace spotted Hatem Helal, founder and CEO of Potential Sciences, 351 days before the funding announcement.
Before the round
| Evertrace detected the signal | 19 November 2025 |
|---|---|
| Funding announced | 5 November 2026 |
| Time before the announcement | 351 days |
On 19 November 2025, Evertrace detected the first signal of company formation for Hatem Helal. That was 351 days before Potential Sciences's funding announcement.
The signal showed Hatem Helal as founder of Potential Sciences, almost a year before the company came out of stealth with this round.
What they’re building
AI models for drug discovery learn mostly from past experimental data, and that record is finite. Quantum chemistry simulations can predict how new molecules behave, but the most accurate calculations are slow and expensive.
Potential Sciences is rebuilding quantum chemistry for modern GPUs and training an AI model on the fundamental laws of physics rather than on earlier experiments. The company says this could make the calculations needed to design chemicals about 100 times cheaper, and it wants to become a frontier lab for chemistry.
The round
The seed round includes the UK Sovereign AI fund, Twin Path Ventures, Odyssey, Portfolio Ventures, Blue Wire Capital, rule30, Nauta Capital and Lightfield Equity, plus angel investment from Patrick Collison and others. Potential Sciences declined to disclose the amount or its valuation; coverage describes the raise as being in the tens of millions of dollars. The Sovereign AI fund is also committing compute capacity to the company.
Find your next investment before the round. Explore Evertrace →


