Atomic Machines

Atomic Machines is coming out of six years in stealth with the Matter Compiler, an AI-native manufacturing system that produces working micro-machines directly from code. The system builds complete, functional, three-dimensional machines with moving parts, many materials and assembled subsystems, with features down to single-digit microns, and with no hard tooling of any kind. The pitch to hardware teams and semiconductor-adjacent operators is straightforward. Conventional manufacturing hits its limits in the low tens of microns, and the MEMS industry operates under a "one product, one process" maxim that forces years of custom development for each new device class. Atomic Machines is trying to collapse that cycle so a new design costs roughly what the design file costs to produce. The company is based in Emeryville and was founded by Jeff Holden, who joined Amazon as its tenth engineer, built Amazon Prime and rose to senior vice president, then became Uber's first chief product officer and founded its self-driving, AI labs and air-taxi divisions . The company has raised $250 million to date from investors including OneIM, Gigafund, XTX Ventures, KAS Venture Partners, Valor Equity Partners, Tru Arrow Partners, Construct Capital, Sozo Ventures and the Regents of the University of California. A key design choice is that the physical manufacturing system sits inside the AI's loop, with every operation measured by high-precision metrology and corrected in real time, and the AI running its own experiments on its own hardware to test what simulation proposes. The launch matters now because the first device off the Matter Compiler is already shipping for evaluation. PrimeSwitch is an electromechanical power relay that opens in 50 microseconds, about a thousand times faster than a conventional contactor, with no sacrifice in conduction efficiency , aimed at high-density electrical infrastructure including the 800-volt DC architectures showing up in AI data centers. PrimeSwitch will be shown publicly for the first time at the Open Compute Project (OCP) Global Summit in San Jose, October 12-15 , which gives operators a concrete way to judge whether the broader claim about code-to-device manufacturing holds up in a real product.