AI News Feed
Market watch
Inscryption to Launch on Mobile Five Years After ReleaseStanford Study Finds Bleached Corals Can Recover Fully Within Weeks by Touching Healthy NeighborsIntel unveils 480GB GPU D-series AI inference acceleratorUCLA's PhoneBot humanoid uses cheap Android phones for vision and controlTinyFish Makes Search and Fetch Free for AI Agents, Pointing to Token Volume as the Real CostWebsites Close Doors to AI Agents, Blocking Shopping, Travel and Banking
Inscryption to Launch on Mobile Five Years After ReleaseStanford Study Finds Bleached Corals Can Recover Fully Within Weeks by Touching Healthy NeighborsIntel unveils 480GB GPU D-series AI inference acceleratorUCLA's PhoneBot humanoid uses cheap Android phones for vision and controlTinyFish Makes Search and Fetch Free for AI Agents, Pointing to Token Volume as the Real CostWebsites Close Doors to AI Agents, Blocking Shopping, Travel and Banking
AI Chips & Compute

Intel unveils 480GB GPU D-series AI inference accelerator

Intel introduced the GPU D-series at the Open Compute Project Global Summit, an AI inference accelerator with up to 480GB of LPDDR5X memory and support for 300-billion-parameter models on a single card.

AI News Feed · 2026-10-19 06:11

The card, previously known by the codename Crescent Island, includes specialized computing hardware and has more memory than Nvidia's Blackwell Ultra B300, which carries 288GB of HBM3e memory, the report said. Intel selected LPDDR5X, a memory technology used in laptops and other power-conscious devices, rather than GDDR7. The choice indicates that Intel is prioritizing memory capacity and power efficiency for AI inference rather than designing a conventional gaming graphics card.

Intel says a single D-series GPU can support large language models with up to 300 billion parameters, while an eight-GPU system can accommodate models exceeding one trillion parameters. The company is aiming the accelerator at lengthy prompts, simultaneous inference requests, enterprise AI services and robotics applications that require responsive processing.

“Inference is becoming the defining AI workload, and enterprises need infrastructure that delivers the right performance at the right cost,” said Anil Nanduri, vice president, AI Products & GTM, Intel Data Center Group. “With Intel GPU D-series, we're focused on making high-capacity inference practical to deploy and economical to scale.”

Intel rates the D-series at 1.3 petaflops for FP8 computation and 1.5TB/s of memory bandwidth. TechRadar noted that memory capacity alone does not determine how quickly a GPU will serve requests, because performance also depends on bandwidth, computing resources, software and workload characteristics. The 480GB capacity is relevant when AI systems must keep substantial model weights and associated data accessible during inference.

The accelerator has a 350W power envelope, a standard PCIe form factor, air cooling and software compatibility across Intel's GPU platforms. Intel says these characteristics are intended to simplify deployment in existing data centers without specialized cooling or entirely new rack infrastructure. Intel also plans to connect the accelerator with its Xeon processors, allowing CPUs to coordinate tasks, manage data movement and handle workloads that do not require GPU acceleration.

“Open models are expanding the options enterprises have for deploying AI,” said Nicolas Dubé, corporate vice president, Data Center Systems and Solutions Group, Intel. “To realize their full potential, organizations need infrastructure that can match different workloads with the right intelligence, right compute resources and right economics. That's what we mean by 'put every prompt in its place.'”

Intel had not revealed a public retail price or a specific general-availability date for the GPU D-series as of the time of writing, according to TechRadar.