Specifications
Model
Axelera Server 150p
Vendor
Axelera AI
AIPU
4x Metis AI Processing Unit
Peak Performance
856 TOPS (INT8)
System Memory
16 GB (64 GB configuration available on request)
Host Interface
PCIe Gen3 x16
Form Factor
PCIe single slot, HHHL (half height, half length)
Thermal Solution
Active air cooling
Typical Power Consumption
30 W - 58 W
Operating Temperature
0 C to 60 C
Security Features
Secure Boot, Root of Trust
Software
Axelera Voyager SDK
Workload Capability
Dozens of camera streams per board; multiple parallel neural networks
Content Protection
DRM support
Model Support
Wide range of end-to-end AI pipelines and models available out of the box
Bracket
Full-height bracket fitted; spare low-profile (half-height) bracket included
Target
High-performance edge and server AI inference
Overview
The Axelera Server 150p is the highest-performance card in Axelera's PCIe line: four Metis AI Processing Units on a single half-height half-length card, delivering 856 TOPS of INT8 inference from a PCIe Gen3 x16 host link. Like the rest of the family it is a single-slot card that keeps host requirements low while scaling AI throughput per slot rather than per server.
The 150p adds server-grade features to the Axelera architecture: active air cooling for sustained duty cycles, Secure Boot and Root of Trust for protected deployment, DRM support and a 0 C to 60 C operating window. Typical power runs 30 W to 58 W, so the card's TOPS-per-watt positioning carries into multi-card deployments. Software integration, quantization tooling and model support are shared with the rest of the Metis family through the Voyager SDK.
Key Benefits
Maximum density per slot. 856 TOPS INT8 from four Metis AIPUs in one half-height half-length card. Server-grade security. Secure Boot, Root of Trust and DRM support for protected inference workloads. Efficient at scale. 30 W - 58 W typical power keeps rack thermal budgets manageable across many cards. Shared toolchain. The Voyager SDK and quantization tooling carry across the whole Axelera Metis family.
Applications
Server AI inference, high-density camera analytics, enterprise and telco edge inference, secure inference deployments, media and content workloads requiring DRM, and scale-out inference clusters.
Request a Quote — AXELERA SERVER 150P — 856 TOPS QUAD-METIS PCIE INFERENCE ACCELERATOR
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →