Rapid urbanization is increasing pressure on city infrastructure and services, and the article highlights how NVIDIA and its ecosystem are promoting physical artificial intelligence to meet those demands. The companies are exhibiting city-scale solutions at the Smart City Expo World Congress in Barcelona, focusing on traffic management, public safety and infrastructure planning. The piece cites United Nations projections that two out of every three people will live in urban centers by 2050 and frames the technology on display as a response to rising congestion, aging infrastructure and the need for climate resilience.
At the center of the demonstrations is the NVIDIA Blueprint for smart city artificial intelligence, which combines digital twins, NVIDIA Omniverse libraries, synthetic data generation, vision language models and video analytics artificial intelligence agents using the video search and summarization blueprint. Updates include NVIDIA Cosmos world foundation models and VLMs to produce photorealistic synthetic data and physical reasoning, along with the latest video search and summarization components within the NVIDIA Metropolis platform. The article also references practical cookbooks for NVIDIA Cosmos Predict, Cosmos Transfer and Cosmos Reason to support intelligent traffic system workflows.
Several partner deployments illustrate the blueprint in action. Esri is building an agent to ingest and visualize camera data in an interactive geospatial map for the City of Raleigh to speed responses and optimize traffic flows. Milestone Systems is adding generative artificial intelligence to its XProtect platform using Cosmos Reason vision language models that were post-trained with 75,000 hours of compliant traffic video per model, offering GDPR-compliant features that could cut operator alarm fatigue by up to 30 percent. Linker Vision is expanding deployments to Ho Chi Minh City and Danang after a Kaohsiung City implementation that cut incident response times by as much as 80 percent.
Bentley Systems, with VivaCity, is supporting Smart Dublin through Cesium and Omniverse for micromobility insights while VivaCity’s sensors run on NVIDIA Jetson and Metropolis. Deloitte is applying Cosmos Predict, Transfer and Reason to automate street inspections and simulate varied conditions such as fog and low light. The article notes hardware and platform support from a wide range of partners and lists additional demonstrations at the show, including Akila and K2K, underscoring a broad ecosystem approach to bringing physical artificial intelligence to urban environments.
