IoT and AI in Action: How Smart Cities Are Cutting Urban Carbon Footprints

Cities do not become lower-carbon simply by installing more sensors or adding artificial intelligence to municipal software. The climate benefit appears only when digital systems change how energy, transportation, buildings, waste, and public infrastructure are actually operated—and when the resulting emissions are measured against a credible baseline.

That distinction matters because the urban climate challenge is large. A 2024 UN-Habitat report on the urban content of national climate plans states that cities account for approximately 67% of global primary energy consumption and 70% of global greenhouse gas emissions. The opportunity for better urban operations is therefore substantial, but “smart city” technology is a means, not an emissions-reduction result by itself.

To make the technology concrete, this article uses a fictional example. Harborview is a hypothetical city created only for illustration. It is not a real deployment, and none of the operating results described here are testimonials or measured field outcomes. Imagine that Harborview has adopted a citywide emissions target and wants to use Internet of Things systems and AI where they can produce verifiable operational improvements.

What IoT and AI actually do in a low-carbon city

The Internet of Things (IoT) is a network of physical devices—such as meters, traffic detectors, thermostats, air-quality sensors, lighting controllers, EV chargers, and equipment monitors—that collect data or receive control commands. The ITU-T Y.4223 recommendation describes common IoT and information-and-communications capabilities for smart cities and communities, including integrated sensing, management platforms, and urban services.

Artificial intelligence (AI) adds pattern recognition, forecasting, anomaly detection, optimization, and decision support. A building system can forecast cooling demand. A traffic platform can predict queue formation. A charging system can schedule EV loads. A waste department can identify collection patterns that deserve attention. Not every useful smart-city control needs AI—many effective systems use conventional optimization or rules—but AI becomes valuable when conditions are too dynamic or complex for static schedules.

The useful chain is therefore:

Sense → understand → decide → act → measure.

If that chain stops at “sense,” the city has a dashboard. If it reaches “act” but never “measure,” the city has an automation project without a verified carbon result.

Harborview starts with a carbon inventory, not a technology shopping list

Before Harborview buys sensors, it establishes where its greenhouse gas emissions come from. That means defining a base year, collecting activity data, applying consistent emissions factors, and separating sectors such as stationary energy, transportation, and waste.

The GHG Protocol for Cities, formally the Global Protocol for Community-Scale Greenhouse Gas Emission Inventories (GPC), provides a widely used framework for cities to identify, calculate, report, and track citywide emissions. This step prevents a common smart-city mistake: optimizing what is easy to instrument rather than what materially contributes to the carbon footprint.

For the fictional Harborview program, suppose the baseline shows that buildings, road transport, electricity demand, and organic waste are the priority operating areas. IoT and AI can then be deployed against specific mechanisms rather than vague “digital transformation” goals.

1. Smart buildings: turning sensors into lower energy demand

Buildings are one of the clearest places where IoT data can lead directly to energy actions. Occupancy sensors, smart meters, equipment telemetry, indoor temperature sensors, and weather data can feed building automation systems. AI or model-based controls can then forecast demand and adjust heating, ventilation, air conditioning, lighting, storage, or other flexible loads.

The U.S. Department of Energy defines a grid-interactive efficient building as an efficient building that uses smart technologies and distributed energy resources to provide demand flexibility while co-optimizing cost, grid services, and occupant needs. DOE's work also emphasizes integration among building systems, sensors, smart meters, and two-way grid communications.

In Harborview, consider an office district during a hot afternoon. Instead of cooling every floor according to a fixed schedule, buildings use verified occupancy and temperature data. The control system can avoid conditioning empty zones, pre-cool selected spaces when electricity is cleaner or less constrained, and reduce noncritical loads during grid peaks while preserving comfort and safety.

The carbon mechanism is straightforward: lower total energy use can reduce emissions, while shifting flexible loads can make it easier to use lower-carbon electricity. But the city should measure kilowatt-hours, peak demand, operating conditions, and the emissions intensity of the electricity actually consumed. A comfortable building that simply moves energy use to another time without reducing or improving the carbon profile may save money without necessarily delivering the expected climate benefit.

2. Traffic systems: use real-time data to reduce unnecessary delay

Roadside detectors, connected signals, transit vehicle locations, incident feeds, and travel-time data can give a city a current view of traffic conditions. Adaptive traffic signal systems then adjust signal timing to changing demand instead of relying only on fixed schedules.

The U.S. Department of Transportation's Intelligent Transportation Systems program describes adaptive signal control as using sensors, data analytics, communications, and increasingly AI or machine learning to reduce stops, improve travel time, and decrease fuel use and emissions.

In Harborview, a morning crash diverts vehicles onto a parallel arterial. The signal network detects the change and adjusts timing rather than waiting for a manually scheduled plan. The same data can help the transit agency prioritize buses or identify where reliability is deteriorating.

However, smoother car traffic is not the same as a complete urban decarbonization strategy. If faster roads induce additional driving, some of the benefit can be offset. Harborview therefore treats traffic optimization as one layer alongside reliable public transit, safe walking and cycling networks, parking policy, and transport electrification.

3. Managed EV charging: connect transportation electrification to the grid

Electric vehicles can reduce tailpipe emissions, but large fleets and charging hubs create new electricity loads. Networked chargers, vehicle schedules, building loads, electricity prices, and grid signals can be coordinated instead of allowing every vehicle to charge immediately when plugged in.

The U.S. Department of Energy describes managed EV charging as strategic control of when and how vehicles charge while meeting operational needs. DOE notes that networked charging can shift or modulate charging according to equipment capacity, building demand, fleet schedules, and dynamic energy conditions. Its smart charge management guidance also identifies the opportunity to prioritize cleaner energy sources and integrate local renewable generation.

For Harborview's fictional electric bus depot, AI does not need to “drive” the buses to be useful. It can forecast which vehicles need how much energy before their next route, then schedule charging so that the fleet is ready without creating an unnecessary facility peak. If local solar generation is abundant in the afternoon, some charging can be aligned with that production when operationally feasible.

4. Connected lighting: make public infrastructure responsive and measurable

Streetlights are another example of a simple asset becoming a networked energy system. Controllers can report energy use, detect faults, and support dimming or scheduling strategies where appropriate. The U.S. Department of Energy's connected streetlighting research highlights the value of networked controllers that can measure power and energy while supporting better accountability, performance, and maintenance.

In Harborview, lighting controllers identify a corridor with unusually high overnight energy use and several fixtures that are not responding to dimming commands. Maintenance teams can target the problem instead of relying solely on resident reports or periodic inspection.

There is an important design constraint: lighting exists first to provide appropriate visibility and public safety. The goal is not to dim aggressively for the sake of a dashboard metric. Controls should operate within engineering requirements and be commissioned so that expected energy savings correspond to actual light output and power consumption.

5. Waste systems: optimize collection, but focus on methane where it matters

IoT-equipped waste containers can report fill level, fleet telematics can reveal inefficient routes, and analytics can help planners understand where collection patterns are changing. Route optimization may reduce collection mileage and fuel use.

But for climate impact, Harborview looks beyond truck routing. Organic waste sent to landfills can generate methane. The U.S. Environmental Protection Agency's research on landfilled food waste reports that food waste makes up about 24% of U.S. municipal solid waste disposed of in landfills and estimates that it is responsible for 58% of fugitive methane emissions from municipal solid-waste landfills.

That means a smart waste strategy should connect data to diversion, not just collection efficiency. Harborview could use route and bin data to improve organics collection design, identify contamination hotspots, or target education and service changes. The climate outcome would then be measured through actual tonnage diverted and an accepted emissions methodology—not by counting the number of “smart bins” installed.

6. Grid-aware city operations: coordinate buildings, storage, and flexible loads

At city scale, the next step is coordination. Buildings, municipal batteries, solar systems, EV chargers, water infrastructure, and other flexible loads increasingly interact with the electric grid. IoT provides visibility into device state and power flows; optimization systems decide when resources should consume, store, or potentially supply energy.

International standards work is moving in this direction. ITU-T Recommendation Y.4512 (05/26), which is in force as of 2026, defines a functional architecture for IoT-based distributed energy storage management systems in smart cities.

In Harborview, the municipal energy platform forecasts a late-afternoon peak. It does not indiscriminately shut equipment off. Instead, controllable loads respond according to preapproved operating constraints: some EV charging is delayed, selected buildings reduce flexible demand, and storage can be dispatched where its operating rules allow. Critical services remain protected.

This is where smart-city decarbonization becomes a systems problem rather than a collection of isolated devices.

Where edge computing fits

Not every sensor reading should travel to a central cloud before the city can act. Edge computing processes data close to the device or local site. A traffic controller may need to react locally to changing demand. A building controller should maintain safe operation if an internet connection fails. A charger cluster may enforce a local power limit without waiting for a remote service.

Cloud platforms are still useful for fleet-wide analytics, model training, long-term optimization, cross-department data integration, and reporting. A practical smart-city architecture often uses both: local control for resilience and low latency, with centralized systems for broader coordination and learning.

A useful reference architecture: from sensor to carbon result

Air-quality sensors are useful, but they are not a carbon inventory

Smart-city discussions often mix air quality and greenhouse gas measurement. They are related environmental issues, but they are not interchangeable. Low-cost air sensors can provide dense local observations of pollutants or environmental conditions, yet they have limitations in accuracy, calibration, placement, and interpretation.

The EPA's current air-sensor guidance explains that sensors vary substantially in design, pollutants measured, data management, and use conditions. These devices can support local insight, but they do not replace regulatory monitoring networks or a citywide greenhouse gas inventory.

Harborview therefore uses air sensors to understand local pollution patterns and evaluate operational conditions, while greenhouse gas reporting remains tied to energy, fuel, waste, and other emissions-accounting data.

The biggest risks are governance problems as much as technical problems

Cybersecurity

A connected traffic controller, building automation network, or charger is also a potential attack surface. Security must be designed into procurement, authentication, network segmentation, patching, monitoring, and incident response. NIST's principles for designed-in security and privacy for smart cities emphasize cybersecurity management processes, expertise, and partnerships rather than treating security as a later add-on.

Privacy and surveillance

A city may need traffic counts without identifying individual drivers, occupancy patterns without tracking named employees, or pedestrian volumes without building a face-recognition database. Harborview applies data minimization: collect only what is necessary for the operational goal, limit retention, control access, and separate climate optimization from unrelated surveillance.

Interoperability and vendor lock-in

If each department buys a closed platform with incompatible data formats, the city can end up with many “smart” silos. The NIST IoT-Enabled Smart City Framework was created specifically to help smart-city stakeholders think about convergence and harmonization across deployments.

AI's own energy use and rebound effects

AI is not environmentally free. The International Energy Agency's Energy and AI report, published in April 2025, estimates that data centers used about 1.5% of global electricity in 2024. The IEA also estimates that broad adoption of existing AI-led applications could reduce emissions equivalent to around 5% of energy-related emissions in 2035, while warning that rebound effects and deployment barriers can erode those gains.

For Harborview, this means using AI where it improves a meaningful operational decision—not deploying large models simply because they are fashionable. A small forecasting model running near a building controller may be more appropriate than sending every sensor stream to an energy-intensive centralized AI service.

How Harborview would know whether the program is working

The fictional city should evaluate outcomes at three levels:

• Operational metrics: kilowatt-hours, peak demand, vehicle delay, charging load, route miles, equipment uptime, waste tonnage, and service reliability.
• Carbon metrics: emissions calculated with documented boundaries, activity data, and emissions factors using a recognized framework such as the GPC.
• Public-value metrics: affordability, comfort, safety, accessibility, privacy, reliability, and whether benefits are distributed fairly across neighborhoods.

The city should also use a reasonable counterfactual. If a mild winter reduces building energy use, the smart-building platform should not receive all the credit. If road construction changes traffic patterns, the city should account for that when evaluating signal optimization. Measurement and verification matter because weather, population, economic activity, electricity mix, and service levels can all change independently of the digital intervention.

A day in the fictional Harborview system

At 6:00 a.m., connected buildings report their overnight energy state and expected occupancy. The municipal bus depot knows which electric buses must leave first. Traffic detectors begin to observe the morning peak. Streetlights report faults before the maintenance shift starts.

By late morning, building controls adjust ventilation and cooling based on actual conditions instead of a rigid schedule. The waste department reroutes one collection crew around a blocked street. An adaptive signal corridor responds to an unexpected traffic surge.

In the afternoon, the energy platform forecasts a grid peak. Flexible chargers slow temporarily without compromising the next bus departures. Selected buildings reduce noncritical loads within approved comfort limits. Local controls continue operating even if a cloud connection becomes unreliable.

At night, the city does not declare success because a dashboard turned green. Energy, transportation, and waste data feed the emissions-accounting process. Engineers review anomalies. Climate staff compare results with the baseline. Cybersecurity teams review alerts. Operations managers check whether efficiency gains affected service quality.

Again, this is a hypothetical example—not a reported deployment or performance result. Its purpose is to show what “IoT and AI in action” should mean when the goal is genuinely lower urban emissions.

The smartest carbon strategy is the one that closes the loop

Smart cities reduce carbon footprints when digital tools are connected to real physical decisions: less wasted building energy, better-managed electricity demand, more efficient transport operations, cleaner charging schedules, responsive lighting, smarter maintenance, and less organic waste reaching landfills.

IoT supplies visibility. AI can turn that visibility into predictions and optimized actions. Edge computing can keep local systems responsive and resilient. But carbon accounting, cybersecurity, privacy, interoperability, and public policy determine whether those technologies create durable value.

The practical lesson is simple: start with the emissions problem, instrument the systems that matter, automate only where the control action is safe and useful, and verify the result. A city is not lower-carbon because it is full of connected devices. It becomes lower-carbon when those devices help people operate urban systems measurably better.