Everspin Technologies and MaxLinear signed a memorandum of understanding (MOU) to evaluate next-generation memory architectures intended to improve efficiency and performance in AI servers, with an emphasis on using persistent MRAM alongside storage acceleration capabilities.
The companies said the technical work will center on pairing Everspin’s persistent, low-latency MRAM with MaxLinear’s platform for storage compression, encryption, and hardware acceleration. The target use cases include metadata management, log data, write buffering, and cache functions, plus other data-intensive server workloads.
AI inference is increasingly constrained by memory behavior and data movement, especially as models grow and key-value (KV) caches expand. Bringing persistence closer to active data structures can change failure behavior and restart characteristics, while acceleration and compression are often used to reduce the amount of data that has to traverse memory and storage paths. But an MOU is an evaluation framework, not a product commitment—so the near-term output to watch for is validation results, not deployment claims.
“AI is forcing system architects to rethink where persistent memory fits in the server memory hierarchy,” said Sanjeev Aggarwal, president and CEO of Everspin Technologies. Vikas Choudhary, senior vice president, connectivity and storage at MaxLinear, said the collaboration is aimed at evaluating architectures that “improve resource utilization, reduce data movement and deliver more predictable performance” for AI workloads.
Under the MOU terms, the companies plan to evaluate technical validation of Everspin MRAM with MaxLinear’s acceleration, compression, and encryption platform for AI and data-intensive server workloads, pursue market development initiatives targeting hyperscale cloud, AI infrastructure, and enterprise server customers, and explore long-term manufacturing, supply, and commercialization opportunities.
Everspin Technologies and MaxLinear did not provide timelines, product SKUs, or deployment details as part of the MOU.
Source: Everspin Technologies
