Cambridge Heavy Industries has been quietly building a desktop fusion device

You can now buy tritium, a critical fusion fuel component, from a startup before other fusion reactors are ready to produce it themselves.

Cambridge Heavy Industries, a stealth startup operating near MIT, has developed a desktop fusion device. CEO Yun Kai explained the company's compact, modular reactor allows for rapid experimentation, with 50-80 "shots" per day to feed AI-driven simulations for design optimization. The device uses a novel inertial confinement approach with two main lasers, minimizing the number of lasers required for ignition. This design, inspired by research from MIT's Keith Nelson lab, achieves high pressures with relatively low laser power. Cambridge Heavy Industries recently secured $6.2 million in seed funding led by Azolla Ventures. The company plans to scale up its technology for a power plant in approximately a decade, but in the interim, it aims to produce tritium, a rare and expensive isotope essential for other fusion power plants.

The development of a desktop fusion device by Cambridge Heavy Industries signifies a potential paradigm shift in fusion energy research, moving from massive, facility-scale experiments to more accessible, iterative prototypes. This approach leverages AI for rapid design optimization and allows for a higher frequency of experimental trials, accelerating the learning curve for fusion technologies. The successful seed funding round underscores investor confidence in this novel, smaller-scale approach to achieving fusion.

Cambridge Heavy Industries' strategy to initially produce tritium presents a compelling near-term business model within the nascent fusion industry. Tritium's rarity and high cost create a significant market gap, and the startup's ability to supply this essential fuel could provide a revenue stream while its larger-scale power plant technology matures. This could also alleviate supply chain concerns for other fusion ventures, potentially speeding up their development timelines.