Intel reveals Crescent Island AI GPU details at Computex

Intel ↗10.64% unveiled new specs for its Crescent Island data center GPU at Computex 2026, built on the Xe3P architecture with a 350W air-cooled design.datacenterdynamics
The chip bypasses costly HBM memory for LPDDR5X, offering up to 480 GB capacity but significantly lower bandwidth than Nvidia ↗2.95% and AMD ↗4.86% competitors.tomshardware
Customer sampling remains on track for the second half of 2026, with Intel targeting cost-sensitive agentic AI inference workloads.intel

Intel Unveils Crescent Island AI Chip Details at Computex, Targeting Nvidia and AMD

Intel used Computex 2026 in Taipei to reveal new technical details about its upcoming data center GPU, code-named Crescent Island, positioning the chip as a cost-effective alternative to offerings from Nvidia and AMD for AI inference workloads.

A Different Approach to AI Hardware

The Crescent Island GPU, built on Intel’s Xe3P architecture, will be a PCI Express add-in card with a 350-watt thermal design power that supports air cooling — a departure from the liquid-cooled designs increasingly common among high-end AI accelerators. Rather than using high-bandwidth memory (HBM) like its competitors, the chip will use LPDDR5X memory, with a reference design of 160 GB and partner configurations scaling up to 480 GB.datacenterdynamics

With an estimated 640-bit memory bus and 10.7 Gbps LPDDR5X, Crescent Island would deliver roughly 684 GB/s of memory bandwidth — well below the nearly 5 TB/s offered by Nvidia’s older H200 GPUs. Intel is betting that the tradeoff in raw bandwidth will be offset by lower cost, greater memory capacity, and reduced power consumption, making the chip attractive for inference rather than training workloads.tomshardware

Targeting “Performance Per Dollar”

Intel’s pitch centers on what it calls optimized “token economics” for customers running agentic AI systems. The chip supports data types ranging from FP4 for AI inference to FP64 for scientific computing. Intel CTO Sachin Katti first announced the chip at the 2025 OCP Global Summit, describing it as part of a new annual GPU release cadence.intel

Reuters reported in October 2025 that the chip represents Intel’s second attempt to break into the AI accelerator market after its Gaudi chips “struggled to compete with industry leaders Nvidia and AMD”.reuters

Timeline and Competition

Intel confirmed that customer sampling of Crescent Island remains on track for the second half of 2026. The company has not yet released specific performance benchmarks. The chip will compete against products like AMD’s MI350P with 144 GB of HBM3E and Nvidia’s H200 NVL in the inference-focused segment of the data center market.tomshardware