Cities collect enormous amounts of data from cameras, sensors and operational systems, but turning that information into practical action for infrastructure planning, maintenance and public services remains difficult. NVIDIA and its partners are using physical AI, agentic AI and digital twins to help cities analyze existing data, simulate urban and industrial environments, and deliver services while giving local governments greater control over how sensitive information is processed.
Initiatives announced at GTC Berlin highlight a shared-computing approach in which city departments, neighboring municipalities and local developers can build on common infrastructure. This model is intended to help organizations expand AI services without each one having to acquire and manage its own dedicated resources.
Agentic AI for city services
Cities are deploying NVIDIA agentic AI for citizen services, video analysis and other applications designed to make public services faster and more responsive. Depending on the workload, city teams may be able to begin with an NVIDIA RTX workstation or a small GPU system rather than making a large infrastructure investment.
Resident services in Spain
In Spain, resident-service initiatives combine the open-source ALIA model, trained with the Barcelona Supercomputing Center, NVIDIA Nemotron and other models. The source describes the deployment as involving 1 million bots and identifies Madrid Digital of the Community of Madrid and the Oberta de Catalunya Administration among the participating organizations.
Across more than 150 locations in Spain, these systems handle municipal procedures, appointment requests, urban-planning incidents and service inquiries.
Interactive AI hosts in the San Francisco Bay Area
In the San Francisco Bay Area, LiveX is developing an “AI City” concept using the NVIDIA Cosmos world model, the Nemotron open model, Omniverse and NVIDIA libraries for intelligent video analytics running on NVIDIA RTX PRO GPUs.
Interactive conversational AI hosts can understand people’s needs, answer questions and take action. They can also adapt to local languages and service requirements in airports, hospitals, local events and other public spaces.
Vision agents for mobility and safety
Vision AI agents can turn existing city-camera feeds into operational insights. These systems are intended to help teams identify mobility and safety issues and coordinate emergency responses across city services.
In Wroclaw, Poland, K2K is building shared AI infrastructure with NVIDIA accelerated computing. The company is using NVIDIA Cosmos world models and the NVIDIA Metropolis Blueprint for Video Search and Summarization (VSS) to convert existing camera streams into actionable information for mobility, safety and event response. The same infrastructure can support multiple city functions.
In Malaysia, ITMAX combines infrastructure and agentic AI applications with long-term management services for cities, including traffic management. The systems are built with reference to Cosmos, Nemotron and Metropolis VSS Blueprints.
ITMAX says its SamurAI V2 vision agent will scale on NVIDIA RTX PRO 6000 GPUs to support more than 17,000 cameras by 2027. The company is also integrating Laplace Intelligence’s open-source traffic simulation platform, TeraSim, to test signal timing and reduce corridor travel times by up to 26% during peak hours.
Using digital twins to plan and operate critical infrastructure
Beyond public-facing services, digital twins can help cities accelerate infrastructure planning and reduce disruption to local communities. A city-scale digital twin can also function as a real-time operational simulation, supporting faster decision-making.
Industrial operations at the Dangote refinery
At the Lekki refinery in Dangote, Nigeria, Honeywell Technologies, Inc. is using Forge Plant Cognition with Nemotron-powered AI agents to interpret operational data. Digital twins of plants and their processes, built using the Omniverse library, are intended to support simulation and training.
Dell PowerEdge servers with NVIDIA RTX PRO accelerators help operators identify bottlenecks, anticipate equipment and process issues, and optimize plant performance.
Urban planning simulations in Taiwan
In Taiwan, Linker Vision and Dutch architecture company MVRDV are using NVIDIA Omniverse and Nemotron to build an environmental, social and governance agent.
The system can assess the environmental effects of different urban-planning scenarios and simulate 15 years of urban-canopy outcomes in 10 minutes. According to the source, the technology reduced design iteration times from two weeks per iteration to less than one hour.
Building a broader local AI ecosystem
Cities can extend AI to more services by building on existing infrastructure and adding computing capacity as needed. Shared resources and managed services can help distribute these capabilities across departments and municipalities.
Sovereign AI for the port of Tangier
At the Mediterranean port of Tangier in Morocco, Siles Technologies is expanding its Sovereign AI Factory for applications including worker safety, congestion analysis, cargo routing and port operations.
Cires Technologies is also developing an ecosystem of application providers that create local models and applications for ports and cities.
Support for startups and research
The Barcelona Supercomputing Center’s AI Factory Accelerator Program provides startups with access to sovereign computing, technical support and training. BSC AI Factory also supports startup research and innovation in areas including agriculture, medical imaging, knowledge access and voice services.
NVIDIA and its ecosystem are scheduled to present physical AI, agentic AI and local AI factories at Smart City Expo World Congress in Barcelona from November 3 to 5. NVIDIA’s vice president of EMEA, Paolo Guglielmini, is scheduled to speak at the event in a keynote speech. A subsequent panel will be moderated by Charbel Aoun, NVIDIA’s director of Smart Cities and Spaces for EMEA.
Source: blogs.nvidia.com