NVIDIA Cosmos Data Factory generates physical AI training data with two prompts
NVIDIAAI · x · 2026-11-19
NVIDIA launched Cosmos Data Factory (CDF), offering 5 cloud-native blueprints and 6 composable skills for curating, augmenting, generating, and labeling physical AI training data, orchestrated by a coding agent.
- Five use cases: AV driving scenario generation, image attribute augmentation for search/re-identification, video augmentation across weather/lighting, rare event video generation, and defect image generation for visual inspection.
- Core skills: Auto Labeling, AnomalyGen, Augmentation, Curation and Retrieval, Simulation, and Orchestration — usable independently or combined.
- Measured results: on a server with 8 RTX PRO 6000 Blackwell GPUs, up to 60x faster dataset assembly and 45x cheaper AV scenario generation vs. traditional methods.
One prompt sets up the environment; a second runs the chosen workflow. A technical walkthrough is on the NVIDIA blog.
More from coding & agent
- Google Labs launches Stitch Loop, an agent that keeps improving shipped code — tokumin · 2026-11-19
- Speed up CI for AI-written code: caching, faster runners, test sharding — rseroter · 2026-11-19
- Netlify rebrands around agent-driven app building with live preview workflow — thisiskp_ · 2026-11-19
- Agentcard raises $8M to let agents buy from 5,000+ stores via Stripe Link — jeff_weinstein · 2026-11-19
- hstack open-sources one-command self-hosting for the Hermes agent — TheMoonMidas · 2026-11-19
- 16 AI models ran produce buying for 90 days — only 3 beat a simple reorder formula — ycombinator · 2026-11-19