Hypervisors for AI, NVIDIA-Certified
Performance-optimized virtualization platforms for AI enterprise data centers and AI factories.
What are NVIDIA-Certified Hypervisors?
Run AI workloads on virtualized infrastructure with greater flexibility, efficiency, and operational control. The NVIDIA-Certified Hypervisors program validates virtualization solutions for modern agentic and inference workloads, so teams get near-bare-metal performance. Certified for the latest x86-based NVIDIA HGX™ and Arm®-based systems, including NVIDIA Grace Blackwell, these solutions let enterprises and cloud providers confidently deploy GPU-accelerated AI and accelerated computing at scale.
What are the Benefits of Choosing an NVIDIA-Certified Hypervisor?
Workload-Based Validation
Certification testing is based on representative performance-critical behaviors across compute, memory, data-path efficiency, and LLM inference.
Optimized Performance
By accurately capturing hardware topology and implementing key performance optimizations, NVIDIA-Certified Hypervisors enable virtualization platforms to deliver near-bare-metal performance for representative AI and accelerated computing applications.
Enterprise-Grade
The program provides a clear, reliable way for enterprise customers and cloud partners to identify hypervisor platforms and virtualized stacks that satisfy the rigorous demands of production deployments.
Ecosystem Readiness
The certification targets the latest NVIDIA GPU architectures—including x86-based systems and Arm-based NVIDIA platforms—to support broad deployment flexibility across enterprise AI factories and NVIDIA Cloud Partners.
Learn More About Hypervisor Certification
Discover what the certification covers, how solutions are evaluated, and certification scope and requirements.
NVIDIA-Certified Hypervisor Partners
GPU Virtualization Use Cases
NVIDIA-Certified Hypervisors validate virtualization platforms and infrastructure software stacks for GPU-accelerated AI infrastructure, supporting enterprise AI factories and NVIDIA Cloud Partners.
Frequently Asked Questions: NVIDIA-Certified Hypervisors Program
The validation suite assesses hypervisor performance across representative single-node and multi-node GPU-accelerated workloads, including AI inference, accelerated compute, inter-GPU communication, and high-speed GPU networking.
NVIDIA-Certified Hypervisors certification is focused on AI and accelerated compute workloads on GPU-based infrastructure, including model-serving workloads used by generative and emerging agentic AI applications. It is not intended for graphics or VDI use cases.
The program validates GPU virtualization platforms that expose NVIDIA GPUs to guest VMs. Certification applies only to platform configurations that NVIDIA has validated and listed.
The certification currently offers two validation tracks: an x86 track scoped to NVIDIA Hopper™ SXM-based systems, and an Arm track scoped to NVIDIA Grace Blackwell systems. The program is designed to expand with new NVIDIA AI infrastructure platforms as they become generally available.
The certification validates that a hypervisor platform or infrastructure software stack has met the requirements of the NVIDIA-Certified Hypervisor validation suite. Any product support inquiries, issues, or bugs related to the hypervisor, virtualization platform, management tools, orchestration software, or core virtualization capabilities must be directed to the respective hypervisor or infrastructure software vendor.
The NVIDIA AI Cloud Ready validation initiative qualifies and validates AI infrastructure software for deployment on NVIDIA Cloud Partners (NCPs) and AI cloud providers. Infrastructure software solutions are assessed against the NCP Software Reference Guide, which defines functional requirements across networking, compute, orchestration, and AI platform layers without prescribing specific technologies. NVIDIA-Certified Hypervisors complement this program by certifying the performance of virtualization-based solutions that may be a part of an AI Cloud Ready solution.
Questions about NVIDIA-Certified Hypervisors?