The Open Compute Project Foundation (OCP) has launched a dedicated AI portal on its Marketplace platform to serve as a central hub for AI cluster builders and infrastructure developers. Announced at the OCP EMEA Summit in Dublin, the new portal consolidates specifications, open hardware contributions, best practice documents, and deployment blueprints to streamline the design and deployment of AI-capable data centers. With AI clusters now consuming up to 1 megawatt per rack, OCP’s open standards approach addresses industry bottlenecks in power delivery, cooling, interconnects, and rack architecture.
OCP’s strategic Open Systems for AI initiative, launched in early 2024, has already produced a blueprint for scalable AI infrastructure and hosted technical workshops on AI physical infrastructure. Recent vendor contributions to OCP include Meta’s Catalina AI Compute Shelf—an ORv3-based design supporting up to 140 kW for NVIDIA GB200 NVL72—and NVIDIA’s MGX-based GB200 NVL72 reference platform. With over 6,000 engineers and 400 corporate members collaborating, OCP is now emphasizing three pillars to guide its AI efforts: standardizing hardware interfaces, supporting open systems development, and offering educational programming through the OCP Marketplace and Academy.
- OCP has launched a new AI portal on its Marketplace as a one-stop destination for AI infrastructure reference designs, white papers, and vendor solutions
- Meta has contributed the Catalina AI Compute Shelf, an ORv3-based 140 kW high-density design for NVIDIA GB200 NVL72 deployments
- NVIDIA previously contributed its MGX-based GB200-NVL72 platform, including 1RU liquid-cooled compute and switch trays
- OCP’s Open Systems for AI initiative addresses rack standardization (250 kW–1 MW), liquid cooling, high-efficiency power delivery, and scale-out fabrics
- The initiative brings together workstreams across power, cooling, silicon, and system design with a focus on open, sustainable architectures
- OCP plans multiple global events to showcase AI cluster advancements, including the OCP Global Summit and regional Tech Days
- IDC’s Ashish Nadkarni emphasizes the need for community-driven standards after siloed first-gen AI cluster deployments led to fragmentation and higher costs
“The AI capable data center build-out is now in its third year… It is the right time for an organization like the OCP to be facilitating a community to determine commonalities leading to standardizations that can help accelerate the market for future generations of AI cluster deployments,” said Ashish Nadkarni, Group Vice President and GM, Infrastructure at IDC.