Hedgehog Contributes OCP Accepted AI Fabric Reference Architectures

Hedgehog said it contributed AI training fabric and AI inference fabric designs to the Open Compute Project (OCP) as reference architectures, where they received OCP Accepted™ recognition and are available through the OCP Marketplace. The move gives operators and system builders access to published designs for deploying open, Ethernet-based AI networks with disaggregated hardware and Hedgehog AI network software.

Hedgehog made the contribution in conjunction with the 2026 OCP EMEA Summit in Barcelona, Spain, April 29–30. The reference architectures were described as validated and intended to support production deployment for Ethernet-based AI fabrics at scale.

The training fabric designs focus on large-scale GPU clusters and include congestion-aware routing, lossless Ethernet, and automated network lifecycle management for predictable performance. The inference fabric designs were described as optimized for efficiency and low latency, with multi-tenant security, hybrid multi-cloud routing, simplified operations, and consistent performance at scale.

OCP and ecosystem partners outlined their role in the effort. FarmGPU said Hedgehog made its backend fabric setup straightforward and that getting these designs into OCP would let more operators run real AI infrastructure. Celestica said it worked with Hedgehog to validate the architectures across open systems using Celestica’s OCP-inspired switches and open source networking software, and Open Compute Project Foundation VP James Kelly said the reference architectures were published through the OCP Contributions Database and OCP Marketplace.