Zenithon raises $10M to build world models for extreme physics
Tech.eu Cate Lawrence ● Covered by 2 sources
Zenithon just raised $10M to build AI models for extreme physics. It wants to test millions of design choices fast for fusion, rockets, and chips.
Based on reporting by Tech.eu, Cate Lawrence — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Zenithon, a UK startup, has pulled in $10 million to build what it says are the first world models for extreme physics. The pitch is blunt: if you can model the messy real world well enough, you can speed up the kind of engineering that’s been slowed by simulations that take days and experiments that take just as long.
The company is aiming at fusion reactors, rockets, and semiconductor manufacturing. Those are not exactly toy problems. Zenithon’s idea is to train AI systems to learn how complex physical systems behave, then use them to predict how different designs will perform. The company says its models could explore a million design points in the time it takes to run one simulation, while also learning from real-world experiments.
Backers include BACKED, Lunar, Seraphim, MMC, and SOSV, plus founders and hyperscaler directors. The company was co-founded by Alex Higginbottom and Abetharan Antony, and it says the team includes researchers tied to foundational work in machine learning for physics, with people drawn from labs around the world.
Zenithon also has some heavyweight advisers. They include Charlie Songhurst, a board director at Meta and former head of strategy at Microsoft; Dan Brunner, who co-founded Commonwealth Fusion Systems and is a former MIT research scientist; and Zongyi Li, an assistant professor at NYU. The funding is going into hiring and compute, with a plan to ship new models every three months and grow in San Francisco and elsewhere in the US.
My take — AI-written commentary, not fact-checked reporting
This is the kind of AI pitch that deserves attention because it targets physics, not PowerPoint. The hype cycle loves chatbots; the real money may be in tools that help engineers stop guessing at expensive, slow problems. If Zenithon can make that million-design-points claim real, nobody will care that the buzzword is “world models.”
Read more about this at: Tech.eu
