Intel launched its Xeon 6+ processors, marking a significant shift in data center architecture as AI becomes more agentic. The new chips, built on Intel 18A process technology, deliver up to 288 Efficient-cores and 2.5 times better performance than previous generations. These processors address orchestration, concurrency, and data movement demands of emerging AI workloads while emphasizing scale-out performance without requiring disruptive data center redesigns.

The company also unveiled the Ethernet E835 controllers and adapters, delivering 200 GbE throughput with configurations from 10GbE to 200GbE. Designed for dense, virtualized deployments, the E835 provides industry-leading power efficiency—up to 1.9 times better performance per watt than comparable NVIDIA solutions. These networking components reduce CPU utilization through RDMA support and streamline packet processing with Dynamic Device Personalization, creating an efficient fabric for AI training and enterprise cloud services.

Intel further detailed its Crescent Island GPU, purpose-built for agentic AI workloads with enhanced efficiency and performance-per-watt. The GPU features LPDDR5x memory delivering up to 480GB capacity and a 350W air-cooled design. With these announcements, Intel reinforces its systems-first approach, positioning the CPU as the control plane for modern AI infrastructure while providing a heterogeneous compute platform designed to reduce friction in AI deployment at scale.