Intel "Crescent Island" GPU Gets D-Series Name, Supports Up to 480 GB of LPDDR5X

Live updates

1. 2h ago

   Intel Names Crescent Island AI Accelerator as GPU D-Series

   Intel officially named its Crescent Island AI accelerator the GPU D-Series at OCP on October 12, 2026. The first product in the line, the D270, is an air-cooled PCIe card designed for inference. It features 32 Xe3P cores, 256 XMX engines, and 32 MB of unified L2 cache. The hardware delivers 1.3 peak PFLOPS of FP8 compute and 1.5 TB/s of memory bandwidth, supporting up to 480 GB of LPDDR5X memory within a 350W power envelope. It supports data formats ranging from FP4 to FP64.

   Why it matters

   Intel is positioning the D-Series to compete with Nvidia by boosting AI turnaround. This server-form factor accelerator shifts the Xe3P architecture toward AI inference rather than traditional graphics. The high memory capacity allows larger models more space to operate during complex tasks.

   What is confirmed

   • The Crescent Island AI accelerator is officially named the GPU D-Series.
   • The D270 is the first product in the D-Series line.
   • The GPU D-Series provides 1.3 peak PFLOPS of FP8 compute.
   • The hardware supports up to 480 GB of LPDDR5X memory.
   • The D-Series delivers 1.5 TB/s of memory bandwidth.
   • The GPU D-Series is an air-cooled PCIe card with a 350W power rating.
   • The chip contains 32 Xe3P cores, 256 XMX engines, and 32 MB of unified L2 cache.
   • The D-Series supports data formats from FP4 to FP64.

   What to watch next

   • Launch timeline for the D270 and subsequent D-Series models
   • Performance benchmarks against Nvidia AI accelerators
   • Integration details for Xeon 6+ rack designs
   Sources used for this update (15)

   1. Yahoo Finance — Intel aims to boost AI turnaround with new GPU chips to take on Nvidia
   2. Intel — Intel Accelerates Open Infrastructure for the Agentic AI Era
   3. ServeTheHome — Intel’s Crescent Island AI Server GPU Gets a Name and Specs: Meet the Intel GPU D-Series
   4. Tom's Hardware — Intel's Crescent Island AI accelerator officially powers new GPU D-Series products — chip delivers 1.3 peak PFLOPS of FP8 compute and 1.5 TB/s of memory bandwidth
   5. StorageReview.com — Intel GPU D-Series Packs 480GB LPDDR5X and 1.3 PFLOPS FP8 into a 350W PCIe Card
   6. TechPowerUp — Intel "Crescent Island" GPU Gets D-Series Name, Supports Up to 480 GB of LPDDR5X
   7. www.tech-critter.com — Intel shows off its Crescent Island inference GPU and Xeon agentic AI racks at OCP
   8. the-gadgeteer.com — Intel’s new D270 AI chip gives bigger models more room to work
   9. www.techpowerup.com — Intel "Crescent Island" GPU Gets D-Series Name, Supports Up ...
   10. prismix.dev — Intel "Crescent Island" GPU Gets D-Series Name, Supports Up ...
   11. www.thefpsreview.com — Intel’s GPU D-Series Arrives with Up to 480 GB of LPDDR5X ...
   12. shattered.io — Intel Crescent Island GPU D-Series: 1.3 PFLOPS, 480GB
   confidence 100%