We plan to deploy Enterprise Nucleus Server on our own Cloud infrastructure as an internal backend. All assets, USDs, extensions and applications stored on it are built and managed by us. Omniverse Kit SDK will be deployed in individual instances that connect to Nucleus Server for asset sync, and it exposes a WebRTC/HTTP port. End customers will only connect to our main application on HTTPS, which presents the stream from the Kit instances. End customers will never access Nucleus Server or SDK Kit directly.
As of May 2026 licensing change, does this architecture require an NVIDIA AI Enterprise subscription for either the Kit SDK or the Enterprise Nucleus Server (If so, where can we buy a commercial license that can permit the SaaS use case in commercial project?), or is it covered under the free production/enterprise use terms?
Here are some information that we reviewed -
Omniverse SDK Kit
Nucleus Server
Hi there and thanks for posting. I am not aware of any May 2026 licensing change, but let me ask about this specifically.
This is what I have found in my research…
You almost certainly do need NVIDIA Enterprise (which now bundles Omniverse Enterprise into “NVIDIA Enterprise”) for a commercial SaaS deployment of Kit-based apps backed by Enterprise Nucleus, even if end customers never see Nucleus or Kit directly. The “free”/developer terms are aimed at development, testing, and very small teams or non‑production use; once you deploy in production as a paid service with more than a couple of internal users/projects, you fall under the Enterprise licensing model.
Below is how this maps to your specific architecture and where/how you’d actually buy the license—but you should treat this as guidance, not legal advice, and have your counsel or procurement validate directly with NVIDIA or an authorized reseller.
Key May‑2026 changes and terminology
NVIDIA has consolidated “NVIDIA AI Enterprise” and “Omniverse Enterprise” into a single entitlement called NVIDIA Enterprise, which includes Omniverse Enterprise rights. As of these changes:
-
The NVIDIA Software License Agreement (NSLA) is the master contract for all Enterprise software, including Omniverse (Nucleus, Kit, etc.).
-
Product‑specific terms for AI Products explicitly include NVIDIA Omniverse (including Kit) for production distributions and deployments.
-
“Enterprise Products” under the NSLA are licensed via subscription (per GPU, per CPU socket, per node, per CCU, etc.), and once you move from Community/Developer to Enterprise, all use of that Enterprise Product must be under a paid license.
NVIDIA also continues to describe a free tier for Omniverse that allows up to 2 users collaborating on the same project content; beyond that, an Enterprise license is required.
Hello @Richard3D, thank you very much for the detailed response, it is very helpful.
I wanted to follow up on the information published in the official NVIDIA Omniverse documentation on May 2026 (Omniverse Overview — Omniverse) as detailed below.
"As of May 2026, Omniverse is now free for development, production, and redistribution use with community support available through the NVIDIA Developer Forums and Discord.
NVIDIA AI Enterprise licenses are required to receive Enterprise Support."
Since this does not make any distinction between small/large teams, production/commercial/paid SaaS use, instead it says “free for development, production, and redistribution use with community support”. It means we only need the NVIDIA AI Enterprise licenses to receive Enterprise Support.
We are also hoping that an NVIDIA staff member can weigh in with an authoritative answer, particularly on whether the May 2026 “free for production” announcement extends to the Enterprise Nucleus Server and/or SDK Kit in a cloud production deployment, or whether Nucleus and/or SDK kit still falls under a separate paid enterprise license.
Any clarification, especially from NVIDIA, would be very helpful.
Thanks.
If that is what it says, then you are free to proceed under those conditions.
