Atomic Machines exits stealth with $250M to build machines from code
What's the deal? Atomic MachinesDealroom has a profile for this one. Try Dealroom → has raised $250 million and emerged from six years in stealth to launch the Matter Compiler, an AI-native manufacturing system that builds working micro-machines directly from code. Its first product, PrimeSwitch, is an electromechanical power relay that opens in 50 microseconds — about a thousand times faster than a conventional contactor — aimed at the 800-volt power systems in AI data centers.
Who's backing it? The $250 million came from investors including OneIMDealroom has a profile for this one. Try Dealroom →, GigafundDealroom has a profile for this one. Try Dealroom →, XTX VenturesDealroom has a profile for this one. Try Dealroom →, KAS Venture PartnersDealroom has a profile for this one. Try Dealroom →, Valor Equity Partners, Tru Arrow PartnersDealroom has a profile for this one. Try Dealroom →, Construct CapitalDealroom has a profile for this one. Try Dealroom →, Sozo VenturesDealroom has a profile for this one. Try Dealroom →, and the Regents of the University of California.
What's the endgame? The Matter Compiler targets micro-machines — motors and gears the size of a grain of sand, robots small enough to work inside the body, and cooling built into AI chips. A single system in which code alone determines the device, it builds three-dimensional machines with moving parts and features down to single-digit microns, with no hard tooling.
Why now? Conventional manufacturing hits its limits in the low tens of microns, and semiconductor fabrication was built to pattern flat circuits, not machines. Under the industry maxim "one product, one process," each new micro-device class has taken years of custom development — a bottleneck Atomic Machines says is about manufacturing, not physics.
The company was founded by Jeff Holden, Amazon's tenth engineer and builder of Amazon Prime, who later served as Uber's first chief product officer. PrimeSwitch will be shown publicly for the first time at the Open Compute Project Global Summit in San Jose, October 12–15.
What they're saying: "3D printing made shapes programmable: any geometry, straight from a file. We're making machines programmable," Holden said. Gigafund managing partner Luke Nosek framed the bet in blunter terms: "If vibe coding is how software gets made now, this is the start of vibe manufacturing."
The signal: Investors are extending the logic of AI-driven software — where the marginal cost of a new design approaches the cost of the file — into physical hardware. If it works, it could unlock a long-blocked frontier of micro-machines tied directly to the AI infrastructure build-out.
Read more: business.theantlersamerican.com
Image credit: Sergei Golyshev (AFK during workdays)
