A Startup Built a Tabletop Machine to Chase Fusion Energy

Cambridge Heavy Industries has come out of stealth in Cambridge, Massachusetts, with a tabletop laser fusion machine and $6.2 million in seed funding led by Azolla Ventures.
Co-founder and CEO Yun Kai gave TechCrunch an exclusive look inside the startup's small lab near MIT, where the company has been working quietly. TechCrunch
The machine is small and built in modules. It uses lasers to squeeze a tiny bit of fuel to start fusion, while using as few lasers as possible. It relies on two main lasers. Large laser sites use many laser paths and need a lot of equipment and setup for each test.
The key number is shots per day. The company wants 50 to 80 tests, or shots, per day. Faster testing means faster fixes to fuel targets, laser pulses, squeeze evenness and measurements. It is like running dozens of small kitchen tests in a day instead of one big one.
The idea came from the lab of MIT chemist Keith Nelson. MIT's Deshpande Center lists Cambridge Heavy Industries as a spinout and describes its approach as new engineering to make laser fusion smaller and more efficient. Deshpande Center
According to that listing, the machine is designed to control density, heat and hold time separately. Together they form the Lawson triple product, the combined measure that sets fusion energy gain. Separate dials, if they work, would let researchers tune each factor on its own. Dr. Yun Kai is also listed as the November 5, 2026 speaker in the University of South Carolina Department of Physics and Astronomy colloquia series.
The seed round also included Overlay, Neta Foundation, P3A, MassCEC and the Rothrock family. The company plans to build a power plant from the idea in about a decade. It said that future plant should be about the size of two basketball courts.
In the nearer term, the company plans to use its smaller machines to make tritium, a rare fuel needed by fusion plants. Making that fuel is a separate business and technical problem from making fusion work well.
The broader context here is a move from one big fusion test at a time toward small machines that can test again and again. Big sites aim for maximum power and detailed measurements. A tabletop system aims for fast turnaround, cheap targets and simple operation. Those choices bring different practical problems in lens life, aiming targets, handling debris and managing data.
Looking at what this means for business, the tritium plan matters alongside the ten-year power plant plan. A small machine that runs dozens of tests per day does not need to put out as much energy as it uses to be useful for testing materials, improving tools or making needed isotopes. In my view, that second path is practical. It lets the company test manufacturing and daily operations while the science improves, and a small neutron or fusion source that can make fuel could find customers before large fusion power exists.
Worth flagging for specialists is the claim of separate control of density, heat and hold time. Laser squeezing usually links them tightly through fluid motion and laser interaction. How the company separates them, through target design, laser timing or chamber shape, will decide whether fast testing brings faster learning or just faster repetition of the same test.
In my view, the bet is clear. Make the test small enough to run often, cheap enough to change, and useful enough to sell something along the way. Fusion still faces hard problems in energy output, wear on walls, fuel supply and final energy cost. A system that shortens the wait between idea and test gives engineers more chances to solve them.


