CoreWeave, Inc. (Nasdaq: CRWV) announced a new product called CoreWeave Physical AI Launchables at NVIDIA GTC Berlin, offering pre-built and pre-tested configuration templates for robotics, autonomous vehicle, and simulation engineering teams.

The product is designed to reduce environment setup time before engineering runs. According to the company, teams can launch a workload with a single command rather than configuring storage, scheduling, and software environments from scratch each time.

At launch, the catalog includes templates for five tools: Nvidia Isaac Sim, CARLA, MuJoCo, Nvidia AlpaSim, and Nvidia Cosmos 3. The templates cover data generation, evaluation, and policy training. Nvidia's Cosmos Data Factory agentic skills for data augmentation, auto-labeling, and data curation are also included.

Each template can be directed at a team's own data and scenes and run against reserved capacity or as a burst job. Blank templates are included for each tool, allowing teams to configure workloads beyond those pre-packaged by CoreWeave.

The product also ships agent skills tuned for CoreWeave's infrastructure, compatible with Claude Code, Codex, and other coding agents.

"Robotics and autonomous vehicle teams are doing some of the most demanding engineering in this industry, and they should be spending their time on it," said Richard Ahlfeld, senior vice president of physical AI at CoreWeave.

Nvidia's director of product management, Nakul Arora, said CoreWeave built its own version of Nvidia's reference agent skills optimized for its cloud infrastructure.

CoreWeave also announced partnerships with Cadence, Convergent Science, Encord, nTop, PhysicsX, and Voxel51 to run those companies' software on CoreWeave infrastructure as part of the physical AI offering.

CoreWeave completed its public listing on Nasdaq in March 2025. The announcement was based on a press release issued by the company.