As urban populations continue to grow—with two out of every three people expected to live in cities by 2050—cities worldwide are turning to AI-driven solutions to manage congestion, improve safety, and enhance sustainability. At the Smart City Expo World Congress in Barcelona, NVIDIA and its ecosystem of partners showcased cutting-edge physical AI technologies designed to transform urban environments in cities including Dublin, Ireland; Ho Chi Minh City, Vietnam; and Raleigh, North Carolina. The NVIDIA Blueprint for smart city AI, introduced in June at NVIDIA GTC Paris, integrates digital twins, synthetic data generation, vision language models (VLMs), and video analytics to create a unified workflow for real-time urban intelligence. New updates include advanced NVIDIA Cosmos world foundation models and VLMs that enable photorealistic simulations and physical reasoning, alongside the latest version of the Video Search and Summarization (VSS) blueprint within the NVIDIA Metropolis platform. Five key partners are deploying these technologies to deliver scalable, secure AI solutions across cities: Esri is collaborating with NVIDIA to develop an AI agent for Raleigh, North Carolina, that processes live camera feeds and generates real-time alerts on an interactive geospatial map. This system helps city operators respond quickly to incidents, optimize traffic, and reduce carbon emissions by adjusting streetlight timing. Milestone Systems has introduced generative AI to its XProtect video management platform, using NVIDIA Cosmos Reason VLMs trained on 75,000 hours of compliant traffic video from Europe and the U.S. The new features allow users to generate reports, filter false alarms, and reduce operator fatigue by up to 30%. Cities like Dubuque, Iowa, and Genoa, Italy, are preparing to adopt these capabilities. Linker Vision is expanding its physical AI platform to Ho Chi Minh City and Danang, Vietnam, building on its success in Kaohsiung, Taiwan, where incident response times dropped by 80%. Using simulation-ready 3D digital twins from AVES Reality and NVIDIA Omniverse, the company models traffic and construction activity to improve urban planning and efficiency. In Dublin, Bentley Systems and VivaCity Ireland’s Smart Dublin initiative are using Cesium’s 3D geospatial platform and NVIDIA Omniverse to visualize micromobility data. VivaCity’s AI-powered sensors, built on NVIDIA Jetson and Metropolis, deliver real-time insights on cyclists, pedestrians, and vehicles, helping identify high-risk areas and improve road safety. A digital twin now integrates VivaCity data, with future plans to add natural language search and weather impact analysis. Deloitte is using NVIDIA Cosmos to automate street inspections across thousands of crosswalks. Cosmos Predict turns static images into realistic video simulations, while Cosmos Transfer generates scenarios like rain, fog, and snow. Cosmos Reason evaluates these environments to identify safety risks, enabling proactive maintenance and protection for vulnerable road users. These innovations rely on a global network of partners including AAEON, Advantech, Dell Technologies, HPE, and others, who are demonstrating physical AI applications on NVIDIA RTX PRO Servers, DGX Spark, and Jetson Thor modules. Additional real-world deployments include Akila’s AI systems in Monaco-Monte-Carlo train stations and a digital twin for Morocco’s University Mohammed VI Polytechnic campus, as well as K2K’s real-time analytics for roadway safety and waste management. NVIDIA and its partners are making city-scale AI a reality—enabling smarter, safer, and more sustainable urban futures. Attendees at the Smart City Expo World Congress can see these technologies in action and learn more about the upcoming public release of the NVIDIA Blueprint for smart city AI. Follow NVIDIA AI on LinkedIn, Instagram, X, and Facebook for updates.