Intel Unveils GPU D-Series Accelerators for AI Inference

Intel has introduced its new GPU D-series inference accelerators and high-density Xeon rack architectures at the Open Compute Project Global Summit, targeting memory-intensive artificial intelligence workloads with an air-cooled PCIe form factor. Shares of Intel Corp (NASDAQ: INTC) closed down 5.79% at $89.85 on October 16, 2026, falling from a previous close of $95.37 amid broader semiconductor market movements.

"Open models are expanding the options enterprises have for deploying AI," said Nicolas Dubé, corporate vice president, Data Center Systems and Solutions Group at Intel, in a statement at the summit. "To realize their full potential, organizations need infrastructure that can match different workloads with the right intelligence, right compute resources and right economics."

Technical Specifications of Crescent Island

Previously known by the code-name Crescent Island, the GPU D-series is engineered on Intel's Xe3P architecture. According to details released in a report from Investing.com, each accelerator delivers up to 1.3 petaflops of FP8 compute performance within a 350-watt power envelope.

The hardware is equipped with up to 480 GB of LPDDR5X memory, providing 1.5 TB/s of memory bandwidth to handle large-scale model inference tasks. Customer sampling for the new GPU D-series is currently underway.

Open Compute Rack Configurations

Alongside the accelerators, Intel showcased new server infrastructure built on the Open Compute Project's Open Rack v3 standard. Select Xeon 6+ rack configurations scale beyond 46,000 cores per rack, operating at a total rack power density of 133.45 kW.

These high-density layouts are designed to support data center deployments requiring dense parallel processing for real-time model serving and enterprise retrieval-augmented generation systems.