“We continue to set the price-performance ratio benchmark for the AI accelerator market,” said Fabrizio del Maffeo, CEO of Axelera AI. “Our goal is to make it possible for our customers to deploy transformative edge AI applications at scale. Metis M.2 Max and our original Metis M.2 card enable customers to choose the solution that best fits their requirements – from computer vision to large language models to the latest of transformer models. We are also adding enhanced security to better address the growing threat landscape at the edge.”
Metis M.2 Max builds on Axelera AI’s proven Metis AIPU and offers high performance and efficiency for large, demanding models such as vision and GenAI, inclusive of LLMs and vision-language models (VLMs). M.2 Max delivers a 33% performance uplift in convolutional neural networks (CNNs) and double the token/second for LLMs and VLMs, all while staying within a typical average power range of 6.5W
With security being paramount at the edge, the Metis M.2 Max incorporates enhanced security features including firmware integrity protection through a securely provisioned Root of Trust, enabling secure boot and secure upgrade capabilities. These security measures operate transparently, requiring no user intervention while helping ensure only authenticated Axelera AI-issued firmware runs on customer systems.
Metis M.2 Max, a 2280 M.2 M-key module, will be available as a standalone board or with an optional low profile heatsink. This design reduces the height of the card by 27% over the current M.2 card, allowing a wider range of configurations. Also, for users that require AI capabilities in power- or thermal-constrained deployments, an onboard power probe can be used to automatically adjust performance to specific settings.
Axelera AI makes it easy for developers to incorporate Metis M.2 Max into their AI projects with the Voyager SDK. Voyager unlocks the full potential of Metis AIPUs and simplifies deployment of proprietary and industry standard models.
Metis M.2 Max starts shipping in Q4 2025 from the Axelera AI Webstore or via Axelera AI Channel Partners.
