Axelera Metis AIPU
Axelera AIAn edge AI inference accelerator from Axelera AI (Netherlands), built on a quad-core in-memory-compute architecture and sold as M.2 modules, PCIe cards, and an integrated compute board. A single quad-core Metis delivers on the order of 214 TOPS, with PCIe configurations scaling to multiple units. It is paired with the Voyager SDK (100+ models) and targets computer-vision workloads such as multi-channel video analytics and quality inspection.
Verified live 2026-06-22 via primary sources. Proprietary in-memory-compute silicon with a publicly available Voyager SDK on GitHub and retail purchasing, but board design files are not open.
Openness
3 medium confidence- schematics
- none
- toolchain
- open (Voyager SDK on public GitHub)
- datasheets
- brief (product pages/briefs public)
- blobs
- required (proprietary AIPU silicon + firmware)
- retail
- open_market (webstore + partners)
Proprietary in-memory-compute silicon with a publicly available Voyager SDK on GitHub and retail purchasing, but board design files are not open.
- https://www.axelera.ai/ recorded 2026-06-22
Metis AIPU form factors, ~214 TOPS, Voyager SDK, webstore availability
- https://github.com/axelera-ai-hub/voyager-sdk recorded 2026-06-22
publicly available Voyager SDK repo under Axelera's official org
Adoption
3 medium confidenceCommercially available via webstore and partners with a public, versioned SDK, but a younger ecosystem than Hailo/Coral.
- https://github.com/axelera-ai-hub/voyager-sdk recorded 2026-06-22
actively released SDK branches indicating developer traction
Capability
4 medium confidenceHigh raw inference throughput via in-memory compute, currently oriented toward CNN/computer-vision pipelines.
- https://www.axelera.ai/ recorded 2026-06-22
~214 TOPS Metis, PCIe scaling, Voyager SDK model coverage
Unchanged since 2026-08-01 (last edited, not re-checked)