Intel® Xeon® 6 processors.
The host CPU for AI-accelerated systems.
Maximise utilisation across your AI platforms while reducing total cost of ownership. Intel® Xeon® 6 processors power modern AI infrastructures, keep accelerators reliably fed with data, and maintain consistently high utilisation.
Intel® Xeon® processors are the host CPUs of choice for high-performance AI accelerator platforms and provide the foundation for efficient, scalable AI workloads.
Why choose an AI-accelerated system?
AI workloads are becoming increasingly demanding, as predictive AI, generative AI (GenAI) and high performance computing (HPC) continue to push requirements for performance, efficiency and cost control ever higher, with purely CPU or GPU-based architectures quickly reaching their limits.
Instead, rely on a system that combines both:
- A host CPU that orchestrates, prepares and optimises workloads.
- Discrete AI accelerators designed to handle compute-intensive training processes.
The host CPU ultimately determines how efficiently your system runs by keeping training pipelines moving, ensuring consistent accelerator utilisation and directing workloads precisely where they are needed.
Intel® Xeon® 6 processors with Performance-cores (P-cores) are purpose-built for exactly this role.
Intel® Xeon® 6 as a host CPU – Five key benefits.
- 1. High I/O performance for fast data movement.
With up to 192 PCIe 5.0 lanes per processor and up to 20% more I/O bandwidth than the previous generation, data can be offloaded faster while keeping GPUs and AI accelerators consistently well utilised.
- 2. High core counts with strong single-thread performance.
Up to 128 Performance-cores per CPU, combined with high turbo frequencies, ensure accelerators are fed with data efficiently, helping to reduce model training times.
- 3. Exceptional memory bandwidth and capacity.
Support for MRDIMMs (up to 2.3× more bandwidth), 2 DIMMs per channel, up to 504 MB of L3 cache and CXL for coherent memory creates an architecture well suited to memory-intensive AI and HPC workloads.
- 4. Advanced RAS features.
Enhanced telemetry, monitoring, shared resource control and real-time firmware updates help minimise unplanned downtime and improve availability across large-scale AI and HPC environments.
- 5. Flexibility for mixed workloads, including AI on the CPU.
Intel® Xeon® 6 supports a wide range of workloads, from traditional applications to AI, while Intel® AMX and FP16 support enable the host CPU to handle AI pre-processing and selected AI tasks directly.
P-cores and E-cores – Choose the architecture that fits your workload.
Intel® Xeon® 6 processors give you two clear options, depending on whether your priority is maximum performance or maximum efficiency.
Performance-cores (P-cores) – Maximum performance for demanding workloads.
- High core counts combined with strong single-thread performance
- Up to 2x the memory bandwidth compared to earlier generations
- Built-in AI acceleration in every core
When moving from 2nd Gen Intel® Xeon® (4208) to Xeon® 6 (6952P), you can significantly consolidate your infrastructure while creating additional capacity for new AI workloads
- Up to 17:1 server consolidation
- Up to 94% fewer servers
- Up to 77% lower CO₂ emissions and energy consumption
- Up to 87% lower TCO
- Return on investment in approximately 4 months
Efficient-cores (E-cores) – Maximum efficiency for scaling workloads.
- High core density for parallel workloads.
- Optimised for performance per watt
- Ideal for environments with limited space, power or cooling
When transitioning from 2nd Gen Intel® Xeon® (6252) to Xeon® 6 (6740E), you can run highly parallelised workloads more efficiently while reducing costs where it matters most.
- Up to 4:1 server consolidation
- Up to 70 % fewer servers
- Up to 53 % lower CO₂ emissions and energy consumption
- Up to 53 % lower TCO
- Return on investment in approximately 13 months
Why choose Intel® Xeon® 6 processors?
When running AI infrastructure, the host CPU plays a decisive role in determining overall system efficiency.
Intel® Xeon® 6 processors manage data flows, distribute workloads and keep AI accelerators consistently utilised, while reliably handling large volumes of data and supporting diverse workloads on a single platform.
This allows you to combine high performance with efficient resource use and reduce both energy consumption and infrastructure costs.
We use YouTube to embed video content on our website. This service may collect data on your activity. For more information, please go to the settings page.
Frequently asked questions about Intel® Xeon® 6 processors.
Intel® Xeon® 6 processors are purpose-built for use as host CPUs in AI-accelerated systems, managing data flows, preparing workloads and ensuring consistent utilisation of GPUs and AI accelerators. High I/O performance, large numbers of Performance-cores and extensive memory bandwidth keep training and inference pipelines running smoothly, helping organisations maximise the efficiency of their AI infrastructure while reducing total cost of ownership (TCO).
Pure CPU or GPU-based systems quickly reach performance and efficiency limits with modern AI workloads such as generative AI, predictive AI and HPC, whereas AI-accelerated systems combine a high-performance host CPU with discrete AI accelerators. The host CPU handles orchestration, data preparation and workload distribution while accelerators take care of compute-intensive training processes, improving hardware utilisation, shortening training times and enabling tighter cost control.
I/O performance is critical to the efficiency of AI and HPC systems, with Intel® Xeon® 6 processors providing up to 192 PCIe 5.0 lanes per CPU and up to 20 % more I/O bandwidth than the previous generation. Faster data movement between the CPU, GPUs, memory and AI accelerators reduces latency, improves data offload and ensures consistently high utilisation across the entire system.
Intel® Xeon® 6 processors are available with Performance-cores (P-cores) or Efficient-cores (E-cores), with P-cores delivering maximum performance and strong single-thread capability for demanding AI and HPC workloads, while E-cores are designed for high core density and efficiency per watt, making them ideal for highly parallel, scaling applications. This allows organisations to choose the architecture that best matches their performance, energy and space requirements.
Intel® Xeon® 6 processors are designed to support a broad range of workloads, from traditional applications to HPC and AI, and extend their role beyond hosting accelerators by enabling AI processing directly on the CPU through technologies such as Intel® AMX and FP16. Pre-processing, selected inference tasks and smaller AI workloads can therefore run without additional hardware, simplifying infrastructure and improving utilisation of existing resources.