NVIDIA expands AI infrastructure and agentic services for global cities

From data to action

Cities currently generate massive amounts of data from sensors, cameras, and operational systems. However, turning this raw information into useful actions for maintenance or public planning remains a significant hurdle for local governments. NVIDIA is addressing this by deploying physical and agentic AI to bridge the gap between data collection and practical use.

Agentic AI refers to systems that can understand needs, provide answers, and take specific actions autonomously. These tools allow city teams to deliver faster and more responsive public services without requiring a massive initial investment in hardware. Depending on the specific workload, these services can run on relatively small GPU systems or standard workstations.

In Spain, 1MillionBot is using these models to handle municipal procedures and appointment requests for several regions. These deployments help manage urban-planning incidents and service inquiries across more than 150 locations. This approach allows for localized, conversational AI that can be tailored to specific languages and community needs.

Vision and traffic management

Vision AI agents are being used to turn existing city camera feeds into operational insights. These systems help teams identify mobility issues and coordinate emergency responses across different city services. In Poland, K2K is using these tools to transform video streams into actionable intelligence for safety and event response.

In Malaysia, ITMAX is scaling vision agents to support over 17,000 cameras by 2027 for traffic management. These agents use a traffic simulation platform to test signal timing and reduce travel times during peak hours. The goal is to provide long-term managed services that help cities manage congestion more effectively.

These technologies rely on specific blueprints for video search and summarization. By using shared infrastructure, different city functions can access the same computing power. This prevents each department from having to acquire and manage its own separate resources for video analytics.

Simulating urban growth

Digital twins are being used to expedite infrastructure planning and minimize community disruption. A digital twin is a virtual model of a physical asset or process that can be used for real-time simulation. These models help city planners make faster decisions by testing different scenarios before they are built in the real world.

In Taiwan, architecture firms are using these tools to assess the environmental impact of urban planning. They can simulate the 15-year outcomes of an urban forest canopy in just 10 minutes. This increases the speed of design iterations from two weeks per cycle to less than one hour.

In Nigeria, a digital twin of a refinery is being planned to support training and simulation. The platform helps operators identify bottlenecks and anticipate equipment issues. By using these virtual models, industrial sites can optimize performance and improve safety for workers.

Sovereign AI and ecosystems

Many cities are moving toward building sovereign AI factories to maintain control over sensitive data. In Morocco, Cires Technologies is expanding a factory for applications like worker safety and cargo routing. This model ensures that local governments keep control over how their data is processed and used by AI systems.

These factories also create ecosystems where local application providers can develop specific models. For example, the Barcelona Supercomputing Center provides startups with access to sovereign computing and technical support. This helps small companies build tools for agriculture, medical imaging, and speech services.

By building on shared resources, municipalities can extend AI capabilities across different departments. This collaborative approach allows for a broader local AI ecosystem where multiple organizations benefit from a common foundation. It reduces the cost of entry for smaller municipalities that lack the budget for independent AI infrastructure.