The hardware behind the future


NVIDIA Makes Omniverse Free for Production Use, and Sells the Support Separately

The simulation platform behind digital twins and robot training no longer needs an AI Enterprise subscription to ship in a commercial product, and partners can redistribute the runtime under the same terms.

NVIDIA Makes Omniverse Free for Production Use, and Sells the Support Separately

Somewhere inside NVIDIA a licensing page just got a great deal shorter. NVIDIA has removed the subscription requirement from Omniverse, its 3D simulation platform, so the software can now be run in production, and shipped inside somebody else's commercial product, without an NVIDIA AI Enterprise licence attached to every GPU it touches.

The news arrived as a post in the company's own Developer Forums, under Announcements, from an Omniverse community manager whose forum badge reads Employee. The post says: "Omniverse is now freely available for both development and production use with no NVIDIA AI Enterprise subscription required." NVIDIA's licensing documentation carries the same terms and dates them to May 2026.

Redistribution is the part worth slowing down for. The announcement adds: "Software built with Omniverse may also be freely redistributed under the same terms." Anyone who builds a simulation tool on top of Omniverse and sells it can now hand the runtime along with it, which removes a licence conversation that used to sit between an independent software vendor and every customer it wanted to sign.

Quick orientation for readers who have only met Omniverse in a demo reel. It is not really a renderer, although it does render beautifully. Think of it as a warehouse that obeys physics, where a CAD model, a robot controller and a camera simulation can all stand in the same coordinate system and agree about where the floor is. That agreement is the actual product. Underneath sits OpenUSD, a scene description format that lets a building imported from one tool and a prop authored in another describe themselves to each other without a translator losing the plot.

Which is why this matters beyond the graphics crowd. Digital twins of factories, and the simulated miles that autonomous driving stacks train on, run on exactly that kind of shared, physically consistent world. Rare and dangerous edge cases are cheaper and safer to generate in software than to wait for on a public road. Making the simulation layer free lowers the entry price for everyone building in that direction, including people who will never render a single pretty frame.

Nothing is free forever, and here the meter simply moved. Support is where NVIDIA still charges. "support is limited to community support via the NVIDIA Developer Forum and Discord" without a subscription, the announcement says, and organisations that want enterprise support buy NVIDIA AI Enterprise from a reseller or a cloud marketplace. Partners who embed Omniverse can take an arrangement in which they handle front line support and NVIDIA handles the back line.

That is the open source infrastructure playbook, minus the open source. Give the software away, sell the answer to the three in the morning phone call. It is a familiar trade, and for a platform whose whole value depends on other people adopting it, the arithmetic is not hard to follow.

The quietness of it is the odd part. A licensing reset of this size would normally get a blog post and a keynote slide. Instead it went up as a post in a developer forum, with the licensing documentation updated to match.

Leave a Reply