UK-based AI startup Zenithon has raised $10 million in funding. The company is building AI models designed to understand how complex physical systems work and help speed up the development of advanced technologies.

The funding comes from BACKED, Lunar, Seraphim, MMC, and SOSV, along with company founders and directors from major cloud technology companies.

Zenithon is developing what it calls “world models.” These AI models can learn how physical systems behave and quickly predict what could happen with different designs. The technology is being developed for areas such as fusion reactors, rockets, and semiconductor manufacturing.

Zenithon says its AI models could test one million possible designs in the time normally needed to simulate just one. The models can also learn directly from real-world experiments.

The company believes this could help engineers explore new designs, improve performance, and make machines work more efficiently over time.

Zenithon was co-founded by Alex Higginbottom and Abetharan Antony. The team includes researchers who have worked on major advances in machine learning for physics, as well as experts from leading research labs around the world.

The company’s advisers include Charlie Songhurst, a Meta board director and former Microsoft strategy executive; Dan Brunner, co-founder and former CTO of Commonwealth Fusion Systems; and Zongyi Li, an assistant professor at NYU.

Zenithon plans to use the funding for hiring and computing resources. The company aims to release new AI models every three months while growing its team in San Francisco and across the US.