Nvidia has unveiled the Blackwell Server Edition GPU designed for enterprise servers. The reveal was made at the ACM SIGGRAPH 2025 computer graphics conference, which started Sunday and runs through Thursday in Vancouver. The company also introduced a host of resources for robotics developers that include a new AI family called the Cosmos World Foundation Models, or Cosmos WFMs, which generate “physics-aware” videos. Notable among them is Cosmos Reason, an open and customizable 7-billion-parameter reasoning vision language model (VLM) for physical AI and robotics.

At SIGGRAPH, the Nvidia Research theme was graphics and simulation innovations enabling physical and spatial AI. Nvidia AI Research VP Sanja Fidler explores in a blog post the “authentic and powerful coupling between the two fields.”

“Joining the existing batch of Cosmos world models are Cosmos Transfer-2, which can accelerate synthetic data generation from 3D simulation scenes or spatial control inputs, and a distilled version of Cosmos Transfers that is more optimized for speed,” TechCrunch reports.

“Physical AI needs a virtual environment that feels real, a parallel universe where the robots can safely learn through trial and error,” Nvidia Research VP Ming-Yu Liu explained, citing the example of “an agricultural robot using the exact amount of pressure to pick peaches off trees without bruising them.”

To drive the deployment of AI agents across physical AI systems Nvidia is also expanding the Nemotron model family, which has reasoning capabilities. Leading enterprise clients including CrowdStrike, Uber and Zoom are already using Cosmos and Nemotron to drive productivity via teams of AI agents and humanoid robots, Nvidia VP of Generative AI Software Kari Briski said in a reasoning model blog post.

New Omniverse libraries and SDKs are now available for building and deploying industrial AI and robotics simulation applications, Nvidia announced at SIGGRAPH.

Although the new WFMs are designed for robotics and real world applications, capabilities such as generating photorealistic 3D scenes, reasoning about physical environments, and integrating with Omniverse’s 3D world-building tools — used for gaming, VR and VFX — could theoretically allow for crossover into those sectors. Cosmos Transfer-2’s ability to create photorealistic video from 3D simulations is a potential cinematic resource.

Helping to power those efforts, the new Nvidia RTX Pro 6000 Blackwell Server Edition GPU “will allow organizations to run large language models at high speed” using 2U form-factor rack-mountable severs for high-performance AI inference workloads, reports SiliconANGLE, explaining the new chips bring “GPU acceleration to traditional CPU-based workloads — including data analytics, simulation, video processing and graphics rendering.”

“What started in the cloud is now transforming the architecture of on-premises data centers,” SiliconANGLE quotes Nvidia CEO Jensen Huang saying.

Related:
Making Safer Spaces: Nvidia and Partners Bring Physical AI to Cities and Industrial Infrastructure, Nvidia, 8/11/25
Nvidia Blackwell Architecture Powers AI Acceleration in Compact Workstations, Nvidia, 8/11/25Nvidia, 8/11/25
Nvidia RTX PRO Servers With Blackwell Coming to World’s Most Popular Enterprise Systems, Nvidia, 8/11/25
Nvidia’s New GenAI Model Helps Robots Think Like Humans, Computerworld, 8/11/25
How a Once-Tiny Research Lab Helped Nvidia Become a $4 Trillion-Dollar Company, TechCrunch, 8/12/25
Dell and HPE Extend AI Infrastructure Lines with New Nvidia-Powered Systems, SiliconANGLE, 8/11/25