Intel is reshaping the landscape of AI infrastructure with a series of advancements that prioritize performance density, power efficiency, and scalability. At the heart of this strategy are the Xeon 6+ processors, built on Intel's 18A process node, which promise to deliver sustained performance under real-world power constraints. This marks a significant shift in how AI workloads are orchestrated, with a focus on concurrency and data movement that aligns with the growing demands of agentic AI systems.

The Xeon 6+ processors are engineered to address the challenges posed by increasingly complex AI workloads. They feature up to 288 efficient-cores, offering up to 2.5 times more performance compared to the previous generation and a 45% improvement in performance per thread per watt over competitors. This efficiency is crucial for environments where watts per rack and throughput per core are critical metrics.

  • Efficient-cores: Up to 288 efficient-cores, delivering up to 2.5 times more performance compared to the previous generation.
  • Power Efficiency: Up to 45% better performance per thread per watt versus competition.
  • Memory Support: 12-channel DDR5 memory with scalable bandwidth for high-density systems.
  • PCIe and CXL: 96 lanes of PCIe Gen 5 and CXL support to accelerate data movement across heterogeneous infrastructure.
  • Energy Telemetry: Intel Application Energy Telemetry (AET) enables real-time workload-level CPU energy and activity telemetry.
  • Server Consolidation: Up to 9:1 server consolidation, reducing footprint and total cost of ownership.
  • Security: Built-in security features including Intel SGX and Intel TDX for confidential and multi-tenant deployments.

The Xeon 6+ processors are already being integrated into telecom network infrastructures and data center systems, with support from major OEMs such as ASUS, Dell Technologies, Ericsson, GIGABYTE, HPE, Lenovo, and Supermicro. This broad ecosystem reflects Intel's commitment to delivering deployable infrastructure for running, scaling, and orchestrating agentic AI workloads on x86 platforms.

Complementing the Xeon 6+ processors is the Intel Ethernet E835 portfolio, designed to deliver high-efficiency networking for next-generation data center environments. The E835 controllers and network adapters offer flexible connectivity with multiple configurations supporting data rates from 10GbE to 200GbE. They are engineered for high performance-per-watt, delivering up to 1.9 times higher performance per watt than comparable solutions from NVIDIA and Broadcom.

Intel's Xeon 6+ and Crescent Island: A New Era for AI Infrastructure
  • Connectivity: Delivers 200 GbE throughput with multiple controller and adapter configurations supporting data rates from 10GbE to 200GbE.
  • Power Efficiency: Up to 1.9 times higher performance per watt than NVIDIA ConnectX-6 DX (CX614106A) and 1.4 times higher than Broadcom BCM957508-P2100G.
  • Network Optimization: Implements RDMA (RoCEv2/iWARP) to reduce CPU utilization and maximize efficiency.
  • Security & Management: Integrates Hardware Root of Trust and signed SPDM with DMTF-based manageability for secure, deterministic operations.
  • Compatibility: Supports multiple operating systems including Linux, ESXi, and Windows.
  • Lifecycle: 10+ year lifecycle built for long-term reliability and support.

The E835 portfolio is supported by industry leaders such as Cisco, Dell, HPE, Lenovo, and Supermicro, ensuring broad compatibility and reliability for modern distributed environments. This networking solution is designed to handle the increasing demands of AI training and enterprise cloud services, providing scalability and high-efficiency features required for next-generation infrastructure.

In addition to the Xeon 6+ processors and Ethernet E835 portfolio, Intel has also announced progress on its AI accelerator roadmap with the introduction of Crescent Island. This next-generation data center GPU is built on the Xe 3P architecture and extends the proven Xe architecture, delivering enhanced efficiency and performance-per-watt while maintaining broad software compatibility for modern AI workloads.

  • Memory Capacity: Up to 480 GB LPDDR5x memory capacity to efficiently handle large, token-intensive workloads.
  • Power Efficiency: 350 W air-cooled PCIe design for highly efficient scaling and strong performance-per-watt.
  • Compatibility: Supports a wide range of datatypes and microscaling formats, from native FP4/MXFP4 to FP64, including expanded support for advanced AI operations.
  • Software Stack: Intel's open programmable AI software stack supports a heterogeneous compute platform designed to reduce friction and enable AI deployment at scale.

Crescent Island is designed to leverage the multi-generational Xe install base, providing an ideal development platform for developers to build, validate, and optimize workloads on familiar hardware. This forward and backward compatibility ensures seamless deployment across different generations of Intel's AI accelerators.

For IT teams looking to optimize their data center infrastructure for agentic AI workloads, the Xeon 6+ processors and Crescent Island GPU offer a robust set of features that address performance density, power efficiency, and scalability. The Ethernet E835 portfolio further enhances networking efficiency, providing a comprehensive solution for modern AI environments.

As Intel continues to advance its AI infrastructure capabilities, the focus remains on delivering platforms that are not only powerful but also energy-efficient and scalable. This approach is expected to set new benchmarks for performance-per-watt in data center environments, making it a critical development for IT teams navigating the complexities of modern AI workloads.