The London-based start-up aims to replace simulations lasting several days with AI models capable of exploring a million design variants, starting with nuclear fusion.
Zenithon AI, a London-based start-up, announced on September 30, 2026 a first $10 million raise led by the Backed Fund, with the participation of Seraphim, Lunar Ventures, MMC, SOSV and leaders from major cloud players. The company was founded in July 2025 by Alex Higginbottom, who left his doctorate in AI at the universities of Cambridge and Manchester, and Abetharan Antony, Doctor of Imperial College.
Accelerate reactor and rocket design
Designing a fusion reactor, rocket engine or semiconductor equipment involves simulating heat, fluids or magnetic fields under extreme conditions, with calculations that sometimes last for days or weeks. Zenithon trains models that learn the behavior of these physical systems, and claims to be able to evaluate one million configurations over time for a single classical simulation, while learning directly experimental results.
The company starts with fusion, where slow simulations slow down the optimization of reactors, before aiming at aerospace, hypersonics and flea plants. The funds must increase the team from 11 to 17 people and support business development in San Francisco and the United States.
Read the full article on the original source
Go to the source →