NVIDIA contributed Vera Rubin rack and compute tray designs to the Open Compute Project through its open MGX architecture, giving system builders a common foundation for AI factory infrastructure. The updated MGX 1.2 specification adds rack-scale blind-mate liquid cooling, interoperable interconnect infrastructure and a liquid-cooled 5,000A busbar for higher-power AI systems. NVIDIA also launched open source software components for its DSX AI Factory platform, including tools for automating switch configurations and stress-testing GPU clusters. The Vera CPU rack design extends the MGX architecture beyond accelerated computing to support tool calling and orchestration for agentic AI workloads.

NVIDIA

Wiwynn showed both NVIDIA Vera Rubin NVL72 and AMD Helios racks at its summit booth, with cooling and co-packaged optics demonstrations.

Unite.AI

Meta is unveiling its first liquid-cooled network rack and a 102.4T switch platform at the 2026 OCP Global Summit, alongside a new scale-out fabric design and an open-source AI transport protocol. The IcePack rack and IceCube switch blades support gigawatt-scale AI clusters with 2x bandwidth over previous 51.2T platforms while fitting tighter power and cooling envelopes. Meta's Flattened Planar Fabric (FPF) replaces traditional single-plane uplinks with eight independent 100G links per GPU, reducing latency and infrastructure cost. Meta is also launching the Open Switch Platform Architecture (OSPA) initiative within OCP, co-led with Microsoft and Alibaba, and open-sourcing its MetaRoCE RDMA transport protocol.

Meta

Microsoft announced its next-generation AI accelerator Maia 300 and agent-native CPU Cobalt 200, designed to improve inference efficiency and handle emerging agentic workloads. Maia 300 is expected to deliver 40% more tokens per watt than leading GPU systems, while Cobalt 200 offers over 50% performance improvement over its predecessor and reduces agent call latency by 33%. The company also detailed networking innovations including Multipath Reliable Connection for scaling beyond 100,000 GPUs and the Optical Compute Interconnect specification for multi-rack scale-up systems. Microsoft contributed Mt. Diablo power architecture specifications to the Open Compute Project and is advancing 800-volt DC infrastructure through industry collaboration.

Microsoft