Axelera AI introduces Metis M.2 Max for edge inference

Axelera announced the Metis M.2 Max as a new member of its Metis AI processor unit family. The company positions the module to deliver the performance of a PCIe card in the M.2 form factor, targeting the most compute-intensive Artificial Intelligence inference workloads at the edge. The product is described as improving performance for large language models (LLMs) and vision transformer networks, while fitting into compact edge devices.

Compared with the existing Metis M.2, the Metis M.2 Max doubles memory bandwidth and presents a slimmer profile. Axelera highlighted advanced thermal management features and additional security capabilities as part of the update. The module will be offered with up to 16 GB of memory, aiming to accommodate larger models and datasets commonly used in modern inference tasks.

Axelera will supply Metis M.2 Max in versions for both a standard operating temperature range and an extended range for harsh environments. The standard version covers minus 20 degrees Celsius to plus 70 degrees Celsius, while the extended version spans minus 40 degrees Celsius to plus 85 degrees Celsius. The company names industrial manufacturing, retail, security, healthcare and public safety as target application areas where the M.2 form factor and environmental options are expected to be advantageous.