News

Nvidia Vera Rubin platform drives triple demand for AI server cabinets with backlogs through 2027

Wednesday, March 18, 2026 at 12:42 AM

Nvidia Vera Rubin AI server demand is tripling cabinet orders and extending backlogs into 2027. Due to a new DC power design requiring up to 600kW, the architecture necessitates 2-3 additional sidecar cabinets for power and cooling. Shipments are expected to begin in the second half of 2026.

Context

As of March 23, 2026, the semiconductor supply chain is facing a critical bottleneck due to the upcoming launch of Nvidia’s Vera Rubin platform. Demand for AI server cabinets has tripled, leaving chassis manufacturers with a massive order backlog extending into 2027. This surge is driven by the platform's extreme power and cooling requirements; the high-end Rubin Ultra architecture is projected to consume up to 600kW per rack, necessitating two to three additional "sidecar" cabinets to house dedicated power supplies and liquid cooling units. This infrastructure scramble impacts major hyperscalers including Microsoft, Meta, Amazon, Google, and Oracle, who are racing to secure slots for the H2 2026 rollout. The Vera Rubin platform marks a shift toward pod-scale supercomputing, integrating seven new chips—including the Vera CPU and Groq 3 LPU—to enable next-generation agentic AI. Investors are closely monitoring the capacity of the broader power and cooling ecosystem to support this unprecedented 10x increase in rack-level density compared to current standards.