Intel Unveils 480GB AI Chip, the GPU D-series, Targeting AI Inference Workloads

Intel has introduced its new GPU D-series AI chip, built on the Xe3P architecture, featuring up to 480GB of LPDDR5X memory on a single air-cooled PCIe card. This launch is significant as Intel aims to optimize for AI inference workloads by prioritizing memory capacity per card, rather than raw FLOPS, to reduce data center serving costs and compete with Nvidia and AMD in the rapidly growing AI inference GPU market.