A look at Emeryville, CA-based Atomic Machines, which is training AI on materials and designs to revamp how microelectromechanical systems (MEMS) are built

New AI-driven chip manufacturing tools become available for early access customers.

Atomic Machines, an Emeryville, CA-based startup, has launched its first product, the PrimeSwitch PS 150, a 150-amp relay. This device is the inaugural product manufactured using the company's proprietary Matter Compiler technology. The Matter Compiler is an AI-driven system designed to revolutionize the fabrication of microelectromechanical systems (MEMS) by training on materials and designs. The PS 150 is reportedly being shipped to early access customers for evaluation as of October 7th, 2026. The company, founded by CEO Jeff Holden, has not yet disclosed pricing or specific production volume details for the new relay.

Atomic Machines' Matter Compiler signifies a potential paradigm shift in MEMS manufacturing, leveraging AI to accelerate design and production cycles. This approach could drastically reduce the time and cost associated with developing and bringing new MEMS devices to market, impacting sectors from automotive to consumer electronics. The company's focus on training AI with materials and designs suggests a path toward more efficient, customized, and potentially novel MEMS functionalities.

The introduction of the PS 150 provides a tangible example of the Matter Compiler's capabilities, though its broader market impact will depend on scalability and cost-competitiveness. If successful, this technology could challenge traditional MEMS foundries by offering a faster, more agile alternative for chip production. Companies relying on specialized MEMS components may soon have new avenues for sourcing and innovation, prompting established players to reconsider their own R&D and manufacturing strategies.