Edge AI for Smart Cities

Edge AI is transforming how cities operate: not by adding another dashboard to a control room, but by embedding intelligence directly into the infrastructure that keeps communities safe, secure, moving, and connected. Intel® Core™ Ultra Series 3 processors, paired with the Metro AI Suite and an open software ecosystem, give state, local, and federal or central governments the performance, durability, and developer tools to deploy agentic AI systems that sense, reason, and act autonomously across transportation networks, public safety operations, and critical infrastructure.

This is not a future promise. It’s production-ready technology, built on open standards, proven at scale, and designed to deliver measurable gains in safety, efficiency, and total cost of ownership.

The Smart City Promise Falls Short Without Real-World AI at the Edge

Investment in safety, security, smart campuses, transportation, and related national infrastructure has become a competitive imperative for cities across the globe. But two decades into the “smart city” movement, the results have been uneven. Many deployments remain reactive: cameras record footage for after-the-fact review, sensors generate data that sit in silos, and command and control centers rely on human operators to spot patterns across hundreds of hours of video feeds. The bottleneck isn’t a lack of data. It’s a lack of timely insights at the point where data is collected and intelligent decisions need to be made.

The next evolutionary stage of a smart city is what I think of as the cognitive city: urban infrastructure that doesn’t just collect information, but actively senses the environment, analyzes and reasons over multimodal data, and responds in real time with minimal human .

And agentic AI makes this possible.

Unlike traditional AI that requires step-by-step instructions, agentic AI sets its own goals, plans a sequence of actions, and adapts its approach based on changing conditions. It breaks down information silos and turns disconnected sensor networks into coordinated systems of intelligence.

Achieving this shift requires a specific kind of computing: low-latency inference on complex vision and language models, industrial-grade reliability for outdoor and mobile environments, and open software that prevents vendor lock-in across diverse city operations. That’s the challenge Intel is built to solve.

Intel® Core™ Ultra Series 3 Delivers Integrated AI Acceleration Purpose-Built for Urban Environments

Intel® Core™ Ultra processors (Series 3) are the first Intel processors to combine AI acceleration and real-time control in a single SoC. For smart city deployments, this matters enormously.

Think about what a smart network video recorder, a Video Security as a Service (VSaaS) gateway, or a roadside unit actually needs to do: decode multiple high-resolution video streams, run object detection and classification models in real time, and increasingly, execute generative AI workloads like natural language video search and summarization. All this needs to happen on-device, at the edge, where latency to the cloud is unacceptable and data privacy is non-negotiable.

The Intel® Core™ Ultra Series 3 delivers on every dimension of that workload. The enhanced NPU and built-in Intel® Arc™ GPU with up to 12 Xe cores provide 50 percent more Xe cores versus the previous generation, translating directly to powerful graphics and inference performance for large, complex models, including GenAI. In head-to-head comparisons, the Intel® Core™ Ultra X9 388H delivers up to 1.9x faster LLM performance, up to 1.7x faster image classification, and up to 2.3x better performance per watt per dollar on video analytics workloads versus NVIDIA® Jetson Orin AGX 64GB.

That’s the kind of advantage that determines whether an agentic AI deployment pencils out at citywide scale or stalls at a pilot.

But performance alone doesn’t define a smart city processor. Durability does.

Roadside units bake in summer heat and freeze through northern winters. Fleet computers absorb constant vibration. Infrastructure monitoring devices sit on bridges and overpasses for years without maintenance. Core Ultra Series 3 supports extended temperatures and outdoor use conditions with industrial-grade reliability, and 10-year product availability means cities can plan long deployment cycles without worrying about forced hardware refreshes. That combination of AI performance and physical endurance is what makes this processor uniquely suited for urban edge environments.

Computer Vision Remains the Workhorse of Smart City AI

The generative AI conversation has captured enormous attention, and rightly so. But in the context of urban infrastructure, computer vision remains the foundational workload. It powers everything from intersection management and pedestrian safety monitoring to license plate recognition, infrastructure condition assessment, and crowd flow analysis. What’s changing is the complexity of what vision systems are expected to do. First-generation smart city cameras counted vehicles. Today’s systems need to classify objects across dozens of categories, track movement patterns across multiple cameras, fuse video data with radar and lidar inputs, and reason about what’s happening in a scene to make autonomous decisions.

Intel has one of the most mature video processing stacks in the industry, with optimized decode pipelines that take standard video streams and prepare them for AI inference with low latency and high throughput. The integrated image processing unit in the Intel® Core™ Ultra Series 3 can directly ingest feeds from high-resolution cameras and process that video on-device. For applications like real-time traffic management or public safety video analytics, this architecture delivers faster response times with fewer dropped frames and lower power consumption than solutions that rely on discrete accelerators.

What makes this particularly powerful for cities is the layering of GenAI capabilities on top of established vision pipelines. An operator can now search hours of recorded video using natural language queries, generate automated incident summaries, or deploy a digital avatar for citizen interaction. These aren’t science fiction use cases. They’re sample applications available today through the Metro AI Suite, running on Intel silicon.

The Metro AI Suite Accelerates Time to Value for City Deployments

Hardware performance is necessary but not sufficient. What has historically held back edge AI adoption in municipalities is the software complexity of moving from a working demo to a production deployment. Cities don’t have the engineering resources of a hyperscaler. They need solutions that compress the development cycle and reduce integration risk.

The Metro AI Suite is Intel’s answer to that challenge. It’s an ecosystem partner-driven software suite that provides everything a developer, system integrator, or hardware builder needs to get video AI and generative AI applications into production on Intel platforms. The suite includes optimized AI models, modular open-source libraries and microservices, dimensioned benchmarks for real-world scenarios, and ready-to-adapt sample applications covering smart intersection management, advanced video search, video summarization, natural language search, smart parking, smart tolling, and live video stream analysis.

What makes the Metro AI Suite different from a collection of tools is the emphasis on end-to-end solution enablement. It includes a hardware catalog of pre-validated, AI-ready systems from Intel’s ecosystem partners and a platform sizing tool that helps buyers right-size their hardware for specific performance, power, and form factor requirements. Developers can bring their own models, create custom models using Geti™, or leverage popular open-source frameworks. All of it is free to download, open-source, and designed to maximize AI and video performance across a broad range of Intel platforms.

The latest Open Edge Platform release enhances the Metro AI Suite with new capabilities that directly address agentic AI requirements for cities: a Smart Network Video Recorder, Video Search and Summarization applications, and a Smart Traffic Intersection application that delivers real-time, multi-camera scene analysis with spatial intelligence for traffic management scenarios. These aren’t isolated demos. They’re composable building blocks designed to be deployed, customized, and scaled on production Intel® AI Edge Systems.

Open Systems and a Proven Ecosystem Reduce Risk and Total Cost of Ownership

City infrastructure projects are long-cycle, high-accountability investments. Municipal procurement teams rightly demand transparency, interoperability, and a clear path to long-term support. Proprietary AI platforms create vendor dependency that conflicts with public sector governance requirements. That’s why Intel’s commitment to open software, open systems, and a proven ecosystem is central to how we approach the smart city market.

OpenVINO™ optimizes and scales AI across CPU, GPU, and NPU, maximizing performance and portability while protecting R&D investment across hardware generations. Over 200 million x86 processors are already deployed at the edge, supported by a global ecosystem of more than 4,000 integrators and ISVs. Intel® AI Edge Systems deliver pre-validated, benchmarked configurations that collapse the time from procurement to production. This isn’t just about making deployment faster. It’s about making it lower risk.

The TCO story is equally compelling. By consolidating AI inference, video processing, and system management onto a single SoC with integrated acceleration, the Intel Core Ultra Series 3 eliminates the need for expensive discrete GPUs, reduces system power draw, and simplifies thermal design. In real-world deployments, customers have achieved 39 to 67 percent TCO savings compared to higher-cost, higher-power alternatives. For a city deploying thousands of edge devices across intersections, transit hubs, and public spaces, those savings compound into transformative budget impact.

Add 10-year product availability, Intel vPro® for resilient remote management, and hardware-level security, and the result is an edge computing platform that city IT teams can deploy with confidence and maintain for the full lifecycle of a municipal infrastructure project.

From Reactive to Proactive: The Cognitive City Is Already Taking Shape

The shift from smart cities to cognitive cities is not a rebrand. It’s a fundamental change in how urban infrastructure operates. A smart city monitors. A cognitive city anticipates. Smart sensors at intersections don’t just analyze what happened after the fact; they provide real-time insights into what is happening at levels beyond human capability and identify emerging patterns before incidents occur. This shift from reactive to proactive is the defining characteristic of cognitive urban infrastructure, and it requires exactly the kind of integrated edge AI that Intel is bringing to market.

Cities around the world are already on this path. Barcelona has built an open-source urban operating system that unifies sensor data across the city. Singapore’s Virtual Singapore initiative uses AI-driven traffic optimization and digital twins. Helsinki monitors air quality with AI systems that predict rather than just report. These are not pilot projects. They are operational deployments that demonstrate what becomes possible when municipalities have access to the right combination of hardware performance, open software tools, and ecosystem support.

The goals and requirements of cities are as varied as the cities themselves. That’s precisely why Intel collaborates with a wide array of partners to provide tailored solutions that address the unique technical and regulatory challenges of each municipality. By integrating edge AI into existing infrastructures, any city can evolve into a cognitive environment capable of comprehensively understanding its internal dynamics and addressing the needs of its residents.

Intel's Edge AI Ecosystem Is Already Delivering Results for Cities and Transportation Networks

Let’s look at how our customers are using Intel’s Edge Portfolio and the results they’re getting

• GRIDSMART, an intelligent traffic system provider, uses real-time computer vision running on Intel® processors and optimized with OpenVINO™ to track vehicles, bicyclists, and pedestrians as they approach and move through intersections. Deployed across nearly 10,000 intersections in more than 1,500 communities globally, the system uses omnidirectional cameras and AI-powered detection to improve safety for vulnerable road users while optimizing signal timing and traffic flow.
• SONDA USA, Intel, and Intelligent Security Systems (ISS) have partnered to deliver an AI-powered video analytics platform that processes live feeds from traffic and security cameras to identify violations, public safety incidents, and emerging threats in real time. The solution combines SONDA's smart city integration expertise, Intel's edge hardware, and ISS's computer vision and deep learning models to give municipalities a unified system for traffic management and public safety response.
• ISS, operating in more than 50 countries across six continents, is leveraging Intel® Core™ Ultra Series 3 processors to improve how quickly operators can react to events across intersections and public spaces while reducing the amount of hardware required in the field. Their SecurOS® platform delivers AI-powered video intelligence for tolling, traffic management, and pedestrian safety, built on more than 20 years of ITS experience and a portfolio of over 30 patents.
• Saimos, a video analytics solution provider, is enabling richer and more efficient video analytics on existing Intel-based hardware, allowing customers to support more camera streams, reduce energy consumption, and improve real-time situational awareness across city operations.

Real-World Challenges Deserve Real-World Performance

Cities don’t need another technology vendor promising a smarter future. They need production-ready, open, and affordable AI systems that work where the physical world demands most.

At Embedded World 2026, we’re demonstrating what edge AI looks like when it’s engineered for the realities of urban deployment: outdoor temperatures, 24/7 uptime, limited maintenance windows, strict data privacy requirements, and the relentless pressure to do more with constrained budgets

That’s what Intel delivers. That’s the power of Intel Inside®.

For notices, disclaimers, and details about certain performance claims, visit www.intel.com/PerformanceIndex