ASRock Rack Shows Liquid-Cooled NVIDIA Systems
The company displayed rack-scale and server hardware at the OCP Global Summit 2026 in San Jose.
Updated on Nov. 2, 2026 in Data Centers
ASRock Rack showcased liquid-cooled NVIDIA systems at the OCP Global Summit 2026 in San Jose. The booth featured a full NVIDIA Vera Rubin NVL72 rack, using direct-to-chip liquid cooling.
Why it matters
The displayed hardware reflects the continued move toward liquid cooling for dense AI computing infrastructure. Showcasing full-rack systems at a major open-hardware conference signals how data center builders are preparing for next-generation GPU deployments.
The full NVL72 rack is rated at 3,600 PFLOPS NVFP4 inference (sparse) and carries 20.7TB of HBM4 memory, with each Rubin GPU providing 288GB. Commercial availability and shipment dates for these systems were not disclosed.
The players
ASRock Rack
ASRock Rack is a server and motherboard subsidiary of ASRock that builds rack-scale and GPU-accelerated systems for data centers.
NVIDIA
NVIDIA is a semiconductor company whose GPUs and rack-scale AI platforms are widely used in data centers worldwide.
The details
The booth included a full NVIDIA Vera Rubin NVL72 rack, HGX Rubin NVL8 servers in 2U and 5U form factors, two NVIDIA MGX RTX PRO servers, and a dual NVIDIA Vera CPU server. The 2U16X-GNR2/DLC RUBIN is fully liquid-cooled with no fans and is powered over a 54V DC busbar.
Timeline
Coverage of the booth was published on November 2, 2026, three days after the core event on October 30, 2026.
Roadmap
The displayed systems build on NVIDIA's MGX modular rack architecture, extending it into fully liquid-cooled configurations for dense AI workloads. The showcase illustrates how server makers are adapting to rising power and thermal demands in AI data centers.
For businesses planning AI infrastructure, the displayed liquid-cooled systems show the hardware forms now available for high-density deployments. Operators weighing cooling and power upgrades can compare these configurations when evaluating future data center builds.
The takeaway
Liquid cooling is becoming a standard design consideration for AI racks, so infrastructure planners should evaluate cooling compatibility early in purchasing decisions. Reviewing rack power and busbar specifications alongside GPU counts helps clarify total facility requirements.
Further reading
Explore more Data Centers coverage for context on the infrastructure behind AI computing.
Source note: This article includes information reported by ServeTheHome.
