Intel Names Data Center GPU Line 'D-Series'
Intel announced at the 2026 OCP Global Summit that its "Crescent Island" data center inference GPU is officially named Intel GPU D-series. The first product will be the Intel GPU D270.
Intel has officially named its "Crescent Island" codenamed data center inference GPU as the Intel GPU D-series. The announcement was made at the 2026 OCP Global Summit. The first product in this new series is the Intel GPU D270.
The D270 features 32 Xe cores and 256 XMX engines, delivering 1.5 TB/s of FP8 compute performance. It supports up to 480GB of LPDDR5X memory with a bandwidth of 1.5 TB/s. The card has a power consumption of 350W and utilizes a standard PCIe AIC form factor with PCIe Gen5 support, designed for air-cooled configurations.
The Intel GPU D-series is designed for inference workloads such as large language models (LLM) and long-context tasks. Intel stated that a single card can handle models with 300 billion parameters, and an 8-card system can support models up to 1 trillion parameters. Target applications include multi-tenant AI services, robotics, and other latency-sensitive workloads.
Intel also showcased new rack designs developed with partners at the event. These solutions, leveraging Xeon processors, can scale to over 46,000 CPU cores and deliver up to 133.45 kW of power.