Potential Sciences Launches With Tens of Millions to Build AI Drug Design Models
Updated
Updated · WIRED · Nov 5
Potential Sciences Launches With Tens of Millions to Build AI Drug Design Models
3 articles · Updated · WIRED · Nov 5
Summary
Potential Sciences debuted in the UK with seed funding in the tens of millions of dollars to build AI systems that predict which drug molecules will work before costly lab testing.
The startup plans to generate large synthetic chemical datasets by simulating interactions from atoms to proteins, arguing today’s main public source—the Protein Data Bank—is too small for more powerful models.
SovAI, the UK government’s AI venture fund with roughly $800 million, backed the company and can provide access to national AI supercomputing; Twin Path Ventures also invested.
Researchers said accurate simulation of molecular binding could transform drug development, but warned the field still faces major technical limits, crowded competition, and long safety and efficacy trials even if design improves.
Potential Sciences has not yet settled on a revenue model or whether to open its tools, while acknowledging the same technology could be misused to design dangerous chemicals.