Using NVIDIA AI Enterprise license for vGPU compute (formerly vCS) on VMware vSphere – confirmation needed

Virtual GPU Forums More vGPU Forums
1

Hello,

I would like to get an official clarification regarding the usage of vGPU compute (formerly NVIDIA Virtual Compute Server / C-Series) when NVIDIA AI Enterprise (NVAIE) entitlement is available.

Background

In earlier deployments, I successfully used NVIDIA Virtual Compute Server (vCS) on VMware vSphere with the following flow:

  • Install vGPU Manager on ESXi host
  • Install vGPU guest driver in VM
  • vGPU profiles were visible, including C-series (compute) profiles
  • Obtain a .tok license token
  • Configure licensing with legacy token-based method
  • Product name shown as NVIDIA Virtual Compute Server
  • Setup worked as expected

Current situation / confusion

Recent documentation states:

“vCS is not supported on VMware vSphere.
C-series vGPU types are not available.
Instead, vCS is supported with NVIDIA AI Enterprise.”

At the same time, the NVIDIA AI Enterprise documentation (vGPU appendix) describes vGPU C-Series (compute) usage on vSphere, especially from vSphere 8.0 onwards, including MIG-backed vGPU.

Clarification requested

I would like to confirm the correct and supported workflow today when NVIDIA AI Enterprise entitlement is available:

  1. Is it correct that vCS as a standalone product is deprecated, but the same compute vGPU capability is supported under NVIDIA AI Enterprise?
  2. When using NVAIE:
  • Is the installation logic still the same (host vGPU manager + guest driver)?
  • Will compute-capable vGPU profiles still appear (even if not explicitly labeled as “C-Series”)?
  1. For licensing:
  • Is Platform Type = 4 (NVIDIA License System) the correct and supported method?
  • Should NVIDIA AI Enterprise entitlement via NLS be used instead of legacy .tok-based licensing?
  1. Is the legacy vCS / token-based workflow still functional but considered unsupported, while NVAIE + NLS is the supported approach?

Summary question

If I have a valid NVIDIA AI Enterprise license, can I use compute vGPU (formerly vCS) on VMware vSphere, and is the only supported method today:

  • vGPU Manager + guest driver
  • Assign compute-capable vGPU profile
  • License via NVIDIA License System (Platform Type 4) using NVAIE entitlement?

I would appreciate an official confirmation to align our deployments with NVIDIA-supported best practices.

Thank you.