NVIDIA Expands Cosmos With Five Data Factory Blueprints and Simulation Checkpoints

NVIDIA has expanded its Cosmos platform with new training tools, simulation checkpoints and Cosmos Data Factory blueprints for developers building world models, which help physical AI systems understand and simulate the real world. The NVIDIA Blog said the platform ranks top five on six different open models leaderboards.

What the Cosmos Data Factory Blueprints Do

Cosmos Data Factory brings together tools for labeling, curating, augmenting and evaluating data at scale for physical AI training. Cosmos Curator, a GPU-accelerated solution for curating video training datasets, has added support for multimodal multisensory inputs and processing blocks. The Data Factory now offers five cloud-native blueprints and is available through GitHub, Microsoft Azure, Nebius and CoreWeave.

The blueprints cover autonomous vehicle dataset generation, image attribute augmentation, event video generation, defect image generation and video data augmentation. Porsche is leading development of a shared AV data flywheel across Volkswagen Group brands, using Cosmos Data Factory through Voxel51 on Microsoft Azure, which is expected to reduce repeated physical test drives by up to 90% and shorten development cycle times by up to 85%.

ANYbotics generated over 1,500 synthetic video clips with Cosmos Data Factory blueprints, curating real-world data and validating outputs with Encord's platform, to train its industrial-inspection robots to recognize liquid spills and other changes between facility visits. Ultralytics retrained its YOLO model with real images and targeted synthetic defects generated using Cosmos Data Factory, improving defect localization and recall. Ultralytics CEO Glenn Jocher discussed YOLO, an open-source computer vision model, and the company's platform for deployment, with YOLO26 targeting edge and CPU environments, according to Pulse 2.0.

NVIDIA's Funding and Acquisition Record

NVIDIA's recent company records include a June 2026 funding round in which it issued $25 billion in high-grade bonds to expand liquidity to roughly $80 billion and fund a 2,400% dividend increase alongside a commitment to return more than 50% of free cash flow to shareholders, according to akta.pro company records. The same records show a January 2026 acquisition of AI inference chip startup Groq for $20 billion and a December 2025 acquisition of SchedMD, the developer of Slurm workload management software used on about 60% of the world's supercomputers.

Earlier NVIDIA Open Model Deployments

NVIDIA's open models have been deployed across several European projects. In October 2026, European retailers including Shopify, Zooplus and L'Oréal began deploying NVIDIA open models and physical intelligence across retail operations, with Shopify's post-trained model cutting inference costs by 90% and speeding responses 7x, according to the NVIDIA Blog. That month, NVIDIA Earth-2 open models were deployed across Europe for weather and climate prediction, with StormScope EU trained on EUMETSAT satellite data, and NVIDIA partnered with European governments on public sector projects including AI@EC, healthcare imaging and agentic factories.

Cosmos Training and Deployment Tools

At GTC Berlin, NVIDIA extended Cosmos Framework training, optimization and deployment tools and recipes as agent skills to more open models, along with Cosmos 3. Developers and agents can adapt open models including GR00T, Alpamayo, QWEN, Pi0.5 and other compatible Hugging Face models through fine-tuning and reinforcement learning, optimize with distillation and quantization, and deploy with high-performing inference.

New Cosmos 3 Nano model variants for simulation enable developers to test autonomous policies in repeatable, hard-to-capture, multiview multisensor scenarios. Updated NVIDIA Cosmos 3 NIM microservices deliver up to 2x faster inference with no quality loss in validated configurations, while continuous video analysis enables sub-second alerts. The new NVIDIA Metropolis Blueprint for video search and summarization 3.3, built on Cosmos NIM microservices, helps teams monitor 40% more video streams with the same GPU.