Cities Worldwide Build and Scale AI Services With NVIDIA Physical and Agentic AI

NVIDIA says cities and infrastructure operators around the world are adopting its physical and agentic AI tools for planning, transportation and public services, in a November 2 blog post published ahead of the Smart City Expo World Congress, held Nov. 3-5 in Barcelona. The company argues that shared computing lets departments and neighboring municipalities build services without each buying its own resources, and that some projects can start on RTX workstations or a small GPU system.

Among the examples, Spain's 1MillionBot pairs ALIA models from the Barcelona Supercomputing Center with NVIDIA Nemotron in resident-service agents across more than 150 deployments, including for Madrid's regional government and Catalonia's Administració Oberta. In Malaysia, ITMAX is scaling its SamurAI V2 vision agents to support over 17,000 cameras by 2027, and NVIDIA says signal-timing tests with the TeraSim simulator cut peak-hour corridor travel times by up to 26%. K2K is building shared camera-analysis infrastructure in Wrocław, Poland, while LiveX is deploying conversational AI hosts in the San Francisco Bay Area.

Digital twins feature heavily as well. Honeywell's Forge Plant Cognition uses Nemotron-powered agents at Dangote's Lekki refinery in Nigeria, with an Omniverse-based twin planned. In Taiwan, Linker Vision and architecture firm MVRDV simulate 15 years of urban forest canopy growth in 10 minutes, reportedly shrinking design iterations from two weeks to under an hour.

The post also highlights Cires Technologies' sovereign AI factory at Morocco's Tanger Med port. The performance figures are NVIDIA's own and come from a promotional post, but they show the company pushing local, government-controlled AI as a growing market.