Atomic Machines Comes Out of Stealth To Help the Next Generation of Data Centers
Atomic Machines says its first product, the PrimeSwitch PS-150, carries 150 amps continuously, holds its state with zero power, and opens in 50 microseconds. The company puts that at about a thousand times faster than a conventional contactor. It unveiled the relay on October 7 after six years in stealth.
Founder and CEO Jeff Holden describes the gap it targets. Semiconductor switches open in microseconds but give off heat the whole time they conduct. Electromechanical relays conduct efficiently but take milliseconds to open. He says the move to 800-volt DC power in AI data centers has made the gap more pressing.
The manufacturing system
PS-150 is built on what Atomic Machines calls the Matter Compiler. Holden says it uses no molds, masks, or fixtures, so a new device design costs the effort of writing it down. Its operations are composable primitives that take design intent as input, and each operation is measured so errors are corrected before they compound. The company’s stated goal is prompt-to-product manufacturing, which Holden compares to working with an AI coding agent where some of the output tokens are physical. He describes that as a destination.
According to Holden, the relay combines micron-scale features, the alloys used in full-size machines, and three-dimensional assembly of moving parts. He says semiconductor fabs cannot process those materials or assemble those parts, and that no conventional tool reaches the feature sizes.
Where Holden sees the opening
In a second post, Holden defines a machine as something that converts energy or matter, shapes a flow through engineered geometry, and directs that flow. He sorts machines by seven physical domains and by whether they deliver energy and matter or encode information. Chip fabrication, in his account, is tuned for one cell of that grid: electrical domain, information purpose.
MEMS, which has adapted chip fabs to build small machines for about forty years, covers roughly half the cells, he writes. Examples include phone accelerometers, micro-mirrors, inkjet nozzles, and the blood-chemistry cartridges used in emergency rooms. Holden says they share limits: no magnetics, no real force, no sustained pressure, and no true third dimension.
His origin post takes the argument further. Humanity’s manufacturing catalog, he writes, grows by adding machinery one product at a time, since each design needs its own mold, mask, or die. The ribosome builds working molecular machines from a string of digital symbols. A 150-amino-acid protein drawn from 20 amino acids has 20^150 possible structures, all made by one machine. Holden says no human system has had a repertoire whose elements are working devices.
He gives three reasons for starting at microscale. Surface forces dominate inertial ones, and thermal time constants fall to microseconds. Batch production can put thousands of devices in one run. And outside computation, the territory holds MEMS and little else. A sensing and acting loop could also close inside a package a few microns long, next to the chips that already work at that scale.
What the posts leave out
Both pieces are company announcements. For PS-150 they give three specifications. They do not give a voltage rating, which matters for the 800-volt DC use Holden cites. They also omit cycle life, test conditions, independent measurements, price, and production volume.
The larger claims are Holden’s own and come as argument without data. His one sourced passage is a footnote on information storage, citing Seth Lloyd’s 2000 estimate of about 10^25 bits per milligram at the physical limit, against humanity’s stored information at 10^23 to 10^24 bits.
Anyone evaluating protection for a high-voltage DC bus would need the voltage rating and endurance data first. The posts also do not say whether a second device from the Matter Compiler is in development.
