A new presentation on Intel's GPU roadmap has revealed the architecture succeeding the Xe3P architecture. According to the shared slide, a new phase named "Xe Next" will follow Xe3P. This new architecture signifies a new era in the company's AI and inference-focused GPU strategy.
Xe Next is Coming
The presentation indicates that after the Xe3P-based GPU codenamed Crescent Island, the next generation of inference-focused GPU family will arrive. The "Xe Next" phase specifically covers new-generation accelerators focused on agentic AI and inference workloads in data centers. Another notable detail is the codename Jaguar Shores. Intel positions Jaguar Shores as the company's first general-purpose programmable GPU AI accelerator.
It was previously announced that Falcon Shores would remain an internal test chip. This suggests that Jaguar Shores could be the first major step to be commercially offered. Intel's data center GPU roadmap has been revised several times in recent years. While intermediate steps like Rialto Bridge and Lancaster Sound were canceled, the company shifted its focus to more long-term and scalable architectures.
The fact that Xe3P will be used in both data center and consumer segments indicates that architectural integrity will be maintained. Moreover, the phrase "Xe Next" had been used by Intel as a transition label before. However, its direct placement after Xe3P in the current roadmap confirms a clearer timeline for the next-generation AI accelerators than previously planned.
0 Comments: