Tuesday, August 25, 2026

Intel unveils three AI-focused architectures at Hot Chips 2026

Input
2026-08-25 16:09:16
Updated
2026-08-25 16:09:16
Intel's Diamond Rapids architecture diagram. Provided by Intel Korea.

[Financial News] Intel unveiled three architectures optimized for agentic artificial intelligence (AI) and enterprise workloads at Hot Chips 2026, a major semiconductor conference held at Stanford University in the United States.
The company also introduced a broad lineup of next-generation products, including the Diamond Rapids processor for high-performance orchestration, the Crescent Island graphics processing unit (GPU) for efficient inference, and the Wildcat Lake system-on-chip (SoC) for intelligent client and edge computing.
At the conference on the 25th, Pushkar Ranade, Intel's Chief Technology Officer (CTO), said, "Agentic AI is fundamentally changing how computing is designed and delivered, from transistors and packaging to the entire system architecture." He added, "The future lies in tightly integrating general-purpose computing, purpose-built acceleration, advanced packaging and open chiplet technologies to build systems that can adapt and scale to workloads even under real-world constraints such as power, cost and deployment."
The processors are part of Intel's 18A process-based lineup and are built on Intel Foundry technologies, including Foveros Direct 3D packaging and early adoption of the industry standard Universal Chiplet Interconnect Express (UCIe) for open chiplet interconnects.
Diamond Rapids is a multipurpose processor designed for enterprise-scale agentic AI.
It is built entirely on Intel technologies, including Intel 18A-P, which improves power and performance, and introduces a new architecture. It combines adaptive compute elements, an integrated memory fabric and flexible input/output (I/O).
Crescent Island is an architecture designed to increase token throughput and create greater business value. It is a low-power GPU that is easy to deploy. With strong computing performance and large memory capacity, it enables larger models, longer context windows and more AI agents to run simultaneously even in conventional air-cooled data center environments. By improving the economics of large-scale inference, it can help companies accelerate AI adoption, increase infrastructure utilization and improve returns on AI investment. 
Intel Core Ultra Series 3 processors, codenamed Wildcat Lake, deliver AI capabilities suited for cost-conscious laptops and intelligent edge platforms. 
[email protected] Jang Min-kwon Reporter