Intel Unveils Xeon Rack Designs and D-Series GPU Details for AI
Intel’s Xeon rack designs and D-series GPUs target agentic AI. Learn what IT teams should check on memory, power, compatibility and deployment readiness. The post Intel Unveils Xeon Rack Designs and D-Series GPU Details for AI appeared first on TechRepublic .
Intel unveils Xeon rack designs and D-series GPU specifications aimed at advancing agentic AI. The announcements, made at the 2026 Open Compute Project (OCP) Global Summit, focus on memory-intensive AI inference workloads for enterprises. Intel's Xeon-based rack configurations can house up to 46,000 CPU cores per rack, consuming 133.45 kW of power.
The D-series GPU supports up to 480 GB of LPDDR5X memory, 1.5 TB/s memory bandwidth, and 1.3 petaflops of FP8 compute. Designed for memory-intensive inference workloads, the GPU D-series is available in configurations with up to 160 GB of memory per server, while partner-built cards can reach 480 GB. Intel's Xeon rack designs utilize OCP's Open Rack v3 (ORv3) standard, ensuring compatibility with other open rack infrastructure.
However, buyers must verify rack, power, and cooling requirements before deployment. Intel stresses the need for open, heterogeneous infrastructure to support agentic AI, encouraging IT teams to evaluate these options alongside existing investments. Customer sampling of the GPU D-series is currently underway, with no general availability dates disclosed.
Before investing, IT teams should assess performance under real-world workloads, including retrieval, context length, concurrent users, and response times. Comparing total costs, including hardware, memory, energy, software integration, and support, is crucial before finalizing any configuration.
Written by urgent.news from TechRepublic's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
