At the SIGGRAPH conference on August 11, 2025, Nvidia announced significant advancements in artificial intelligence and robotics, introducing new models and infrastructure designed to accelerate the development of intelligent robotic systems. The company unveiled Cosmos Reason, a 7-billion-parameter vision language model specifically engineered for physical AI applications and robotics. This model empowers robots to understand and predict physical interactions, a crucial step for tasks ranging from data curation to complex planning and video analytics. Nvidia also introduced Cosmos Transfer-2, an enhanced model aimed at speeding up the generation of synthetic data from 3D simulations and spatial inputs, along with a distilled version optimized for training efficiency.
Complementing these models, Nvidia presented new neural reconstruction libraries, including a rendering technique that simulates the real world in 3D using sensor data. These libraries are being integrated into the CARLA simulator, a popular platform for autonomous vehicle development, benefiting a large community of over 150,000 developers. To support these advancements, Nvidia unveiled RTX Pro Blackwell Servers, a unified architecture designed to streamline robotic development workloads, offering a single platform for both AI model training and complex simulations. These servers, built on the Blackwell architecture, promise up to 45 times better performance and 18 times higher energy efficiency compared to traditional CPU-only systems. The RTX Pro Servers are being offered through partnerships with major system providers including Cisco, Dell, HPE, Lenovo, and Supermicro. These developments underscore Nvidia's commitment to advancing the field of robotics by providing developers with sophisticated tools and robust infrastructure, aiming to foster the creation of more intelligent and capable robotic systems. The global robotics market is projected to experience substantial growth, with analysts estimating a compound annual growth rate of 18.2% through 2030, driven by such AI and automation advancements. Nvidia's strategic focus on physical AI and its integration with simulation technologies positions it as a key enabler for the next generation of intelligent machines, transforming industries through enhanced capabilities in AI reasoning and real-world interaction.