Intel Begins Customer Sampling of D-Series GPUs for AI Inference at OCP Summit

Intel has started customer sampling of its D-Series GPUs, known as Crescent Island, as the chipmaker escalates efforts to compete with Nvidia and AMD in the artificial intelligence data center market. The announcement, made at the OCP Global Summit, marks a milestone in Intel's broader turnaround strategy, which is centered on capturing a share of the fast-growing AI inference segment rather than the training market where Nvidia has long dominated.

The D-Series chips, built on Intel's Xe architecture, are designed exclusively for AI inference workloads. Intel CTO Sachin Katti has described the strategic rationale as a shift in the industry: "AI is shifting from static training to real-time, everywhere inference, driven by agentic AI." Alongside the GPU sampling announcement, Intel said it is expanding its open infrastructure to support the emerging agentic AI era, positioning the platform for the orchestration demands of multi-agent systems.

The move is a critical test for Intel, which previously failed to gain meaningful traction in the AI accelerator market with its Gaudi line. Customer sampling is the precursor to commercial availability, and real revenue from the D-Series is unlikely before 2027. Intel's ability to attract data center customers at this stage will signal whether its AI strategy can translate into competitive ground against Nvidia and AMD.