AI with Intel® Xeon®
Intel® Xeon® processors handle every AI workload, from traditional ML inference to agentic and frontier AI, without GPU lock-in.
Intel® Xeon® wins across the entire AI workload spectrum
From traditional ML inference to frontier GPU systems, Xeon has a performance story at every stage.
Traditional AI
Intel® Xeon® 6 is the Best CPU for AI
Intel® Xeon® 6787P (2S) delivers 4x throughput on Speech Recognition (OpenAI Whisper) and 2.2x throughput on Computer Vision (RetinaNet) compared to NVIDIA® L4 in MLPerf v5.1 inference. 1
- Machine Learning (ML)
- Deep Learning (DL)
- Natural Language Processing
- Image Classification
- Recommender Systems
- Object Detection
CPU Cores
Sequential nature of ML/DL workloads have been optimized for CPU cores.
Acceleration
Built-in Intel® AMX accelerates higher precision ML/DL workloads.
Integration
Better data integration with structured enterprise databases, records, and logs.
History
Long development history of ML/DL applications on CPU architecture.
Scale
Smaller scale workloads more typical for ML/DL compared to generative AI.
Generative AI
AI Inference on the Infrastructure You Know
Run AI inference at the scale you need today, on standard enterprise architecture, with no specialized hardware expertise required.
- Design and Programming
- Customer Support
- Content Creation
- Scientific Discovery
Simplicity
Standard enterprise architecture with no need for specialized HW expertise.
Flexibility
Maintain flexibility for non-AI workloads and avoid idle GPU compute cycles.
Security
Enterprise-grade security with established protocols and governance.
Right-sizing
Scale in smaller increments than GPU systems to meet service level targets.
Data Access
Fast access to local enterprise data for Retrieval and Augmentation (RAG).
Ready-to-Deploy Solutions
Skip the Setup. Start Building.
These solutions pair Intel® Xeon® 6 with optimized software stacks and partner integrations so you can get production RAG pipelines running quickly.
- TTFT
- Time to First Token — how quickly the first word of a response arrives
- TPOT
- Time Per Output Token — generation throughput, how fast subsequent tokens are produced
Agentic AI
Optimal Balance of CPU and GPU Resources
Agentic AI needs the right mix of compute. Intel® Xeon® handles AI workloads with built-in accelerators while keeping your infrastructure consistent.
Accelerated
Offload AI workloads to the best CPU for AI with built-in accelerators.
Consistency
Maintain infrastructure consistency with the most-deployed host CPU, trusted across enterprise deployments worldwide.
SW Enabling
Out-of-box support for Intel® AMX with virtualization and orchestration software.
Faster Data
Storage with more I/O bandwidth and vector search performance advantage.
Security
Enterprise-grade security combined with Confidential AI via Intel® TDX and Intel® TDX Connect.
Frontier AI
Benefits of Intel® Xeon® as Host CPU for Frontier AI
What Intel® Xeon® 6 brings to the table as host CPU for large-scale GPU systems.
Leadership
World's most deployed host CPU and vendor of choice for most OEMs.1
Partnership
Co-validated reference designs for DGX, HGX, MGX fuel OEM systems.
Performance
High per-core performance with max turbo (PCT) and Intel® AMX accelerator.
Bandwidth
30% higher memory bandwidth with MRDIMMs, expandability with CXL 2.0.
Security, RAS
Confidential AI with Intel® TDX Connect and AI-Ready RAS for host CPU.
The Most Deployed Host CPU for AI systems
Available across a wide range of AI systems from leading OEMs for NVIDIA GPUs.
Intel® and NVIDIA® collaboration
- Co-validation of Intel CPUs with NVIDIA GPUs on DGX, HGX, MGX reference designs resulting in broad availability with leading OEMs
- Intel will build custom x86 CPUs (with NVIDIA NVLink) that NVIDIA will integrate into its AI infrastructure platforms and offer to the market
- Co-engineering on Confidential AI³ across CPUs and GPUs for host CPU AI security with encrypted bounce buffer and Intel® TDX Connect
“This historic collaboration tightly couples NVIDIA’s AI and accelerated computing stack with Intel’s CPUs and the vast x86 ecosystem — a fusion of two world-class platforms. Together, we will expand our ecosystems and lay the foundation for the next era of computing.”
Confidential AI
Privacy First: Protect Your AI Data & Models
The most complete confidential computing portfolio for AI security.
Accelerated Encryption
- Hardware acceleration with Intel® QAT & Intel® Crypto Acceleration
- Quantum-Resistant AES-256 data & memory encryption
Hardware Isolation & Control
- App Isolation with Intel® SGX
- VM Isolation with Intel® TDX
- CPU to GPU encrypted connection with Intel® TDX Connect
Verification
- Attestation Services with Intel® Trust Authority