Flex and Cerebras scale AI supercomputer production in California
EMS provider Flex and AI systems company Cerebras have expanded their manufacturing partnership to increase production of the Cerebras CS-3 AI accelerator system at Flex's Milpitas, California facility. Production capacity is expected to grow approximately seven times through 2026.
The expansion includes new assembly and integration lines, expanded floor space, advanced test infrastructure and additional manufacturing staff in California, according to a press release from the companies.
The CS-3 is built on Cerebras' wafer-scale engine architecture – featuring a processor larger than any conventional AI chip – and integrates liquid cooling, high-density power delivery, precision mechanical assembly and networking infrastructure into a single platform for large-scale AI training and inference.
Manufacturing the CS-3 required Flex engineers to develop dedicated assembly flows, custom tooling, automated test stations and new methodologies tailored specifically to wafer-scale systems – processes that differ significantly from conventional server production.
"The CS-3 does not resemble a conventional server or rack-scale compute platform. Every stage of the manufacturing process required deep collaboration between our engineering teams," said Rob Campbell, President of Communication, Enterprise and Cloud at Flex.
The Milpitas facility now runs parallel integration lines with enhanced burn-in and validation areas, automated test infrastructure and expanded logistics capacity. The expansion is contributing to growth in high-skilled manufacturing roles across the region.
