Cambridge Heavy Industries Bets on Small, Fast Laser Fusion

Cambridge Heavy Industries has emerged from stealth in Cambridge, Massachusetts, with a benchtop laser fusion device and a $6.2 million seed round 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 operating quietly. TechCrunch

The reactor is deliberately small and modular. It compresses a small fuel target with lasers while using as few lasers as possible, relying on two main lasers to drive ignition. That differs from large, facility-scale laser systems that depend on many beamlines, or separate laser paths, and carry high facility cost and setup for each shot.

The company describes shot rate as the central specification. It intends the benchtop system to support 50 to 80 experiments, or shots, per day. For laser fusion, iteration speed controls how quickly researchers can test and revise target designs, pulse shapes, or how laser energy is delivered over time, drive symmetry, or how evenly the fuel is squeezed, and diagnostic feedback, or measurement data from each test.

The concept emerged from the lab of MIT chemist Keith Nelson. MIT's Deshpande Center lists Cambridge Heavy Industries as a spinout and describes its approach as novel engineering to make laser fusion more compact and efficient. Deshpande Center

According to that listing, the machine is designed to independently improve density, temperature and confinement time, the three parts of the Lawson triple product, a combined measure used to judge fusion conditions. Separate control of those variables, if achieved in practice, would give experimenters individual controls for the core factors that together determine fusion gain, or energy output relative to input. 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 included participation from Overlay, Neta Foundation, P3A, MassCEC and the Rothrock family. The company plans to develop a power plant from the concept in about a decade. It said its future scaled-up power plant design should be about the size of two basketball courts.

For the nearer term, Cambridge Heavy Industries plans to use its smaller-scale devices to make tritium for use by fusion power plants. Tritium, a rare form of hydrogen used as fusion fuel, is a separate technical and commercial problem from plasma performance.

The broader context here is a shift in fusion development from single large shots toward high-repetition experimental platforms. Large laser facilities optimize for peak energy and diagnostic fidelity, or measurement detail. A benchtop system optimizes for cycle time, target cost and operational simplicity. Those are different engineering tradeoffs, with different failure modes in optics lifetime, target alignment, debris management and data handling.

Looking at what this means for commercialization, the tritium plan deserves attention alongside the power plant timeline. A small device that can run dozens of shots per day does not need to reach engineering breakeven, the point where a system puts out as much energy as it consumes, to be useful for materials exposure, diagnostic development or isotope production. In my view, that parallel product path is pragmatic. It creates a way to test reliability, manufacturing and operations while the physics scales, and a compact neutron or fusion source that can breed fuel could find users before grid-scale fusion power exists.

Worth flagging for specialists is the claim of independent control over density, temperature and confinement time. Laser-driven compression usually couples those variables tightly through hydrodynamics, or how hot fluid and plasma move, and laser-plasma interaction. How Cambridge Heavy Industries isolates them, through target design, pulse formatting or chamber geometry, will determine whether high shot rate translates into faster learning or simply faster repetition of the same regime.

In my view, the bet is legible. Make the experiment small enough to run often, cheap enough to modify, and useful enough to sell intermediate output along the way. Fusion still faces hard problems in gain, wall loading, or wear on chamber walls, tritium breeding ratio, and levelized cost, or average lifetime cost of energy. A platform that shortens the loop between idea and shot gives engineers more chances to work through them.