List of NVIDIA-Certified Hypervisors
The following partner platforms are certified as part of the NVIDIA-Certified Hypervisors program for the indicated certification track.
Certification Track: Arm Platforms
Arm platform certification is currently scoped to NVIDIA GB200 NVL systems. This track validates the performance of a single 4-GPU, 2-CPU passthrough virtual machine (VM) within a single GB200 NVL compute tray.
Unless otherwise noted, certifications are validated using the NVIDIA software components supported for the certified platform at the time of testing. Refer to the NVIDIA AI Enterprise Infrastructure Support Matrix for currently supported component versions.
| Partner | Certified Partner Platform and Version | Certification Scope | NVIDIA Platform |
|---|---|---|---|
| Mirantis | Mirantis k0rdent AI Osprey | Mirantis k0rdent AI Osprey | Grace Blackwell |
| OpenNebula Systems | OpenNebula 7.2.1 | OpenNebula version 7.2.1 and all subsequent maintenance and minor releases within the 7.x series | Grace Blackwell |
| Rafay | Rafay-powered VMaaS 3.1 | Rafay-powered VMaaS Release 3.1-XX | Grace Blackwell |
The Certification Scope column is partner-provided. It describes which versions of the certified virtualization platform the partner considers covered by the listing. Versions represented in this column are considered within the scope of this listing based on the partner’s representation, but they are not specifically certified unless listed separately.
Refer to the partner’s compatibility and support guidance for release-specific details. Major releases or material platform changes that affect the certified configuration can require additional review or recertification.
Certification Track: x86 Platforms
This certification covers systems built on the NVIDIA HGX™ platform, as well as NVIDIA DGX™ systems. The x86 platform certification is scoped to NVSwitch-based 8-GPU NVIDIA Hopper™ SXM systems. The x86 track requires the following two phases:
- Single-Node: Validates the performance of a single 8-GPU VM on one NVIDIA Hopper SXM node, using full GPU and network interface card (NIC) passthrough.
- Multi-Node: Validates the performance of two 8-GPU VMs deployed across two identical NVIDIA Hopper NVL8 nodes, with each VM spanning one node.
Unless otherwise noted, certifications are validated using the NVIDIA software components supported for the certified platform at the time of testing. Refer to the NVIDIA AI Enterprise Infrastructure Support Matrix for currently supported component versions.
The Certification Scope note above also applies to this table.
| Partner | Certified Partner Platform and Version | Certification Scope | NVIDIA Platform |
|---|---|---|---|
| Rafay | Rafay-powered VMaaS 3.1 | Rafay-powered VMaaS Release 3.1-XX | Hopper NVL8 |
| Red Hat | Red Hat OpenShift Container Platform 4.22 with OpenShift Virtualization 4.22 | Red Hat OpenShift Container Platform 4.22 with OpenShift Virtualization 4.22 | Hopper NVL8 |
| VMware | VMware vSphere 9.1 | VMware vSphere 9.1.0 and all future vSphere 9 releases | Hopper NVL8 |
| Nutanix | Nutanix Cloud Infrastructure 7.6.9 | Nutanix Cloud Infrastructure 7.6.9 with Nutanix AHV 11.3 and all future releases of Nutanix Cloud Infrastructure | Hopper NVL8 |