Can smart city technologies reduce urban carbon emissions?

How much can smart cities actually cut carbon emissions?

The short answer is that smart city technologies can reduce urban carbon emissions by roughly 4% to 25%, depending on which technology you use and where you deploy it. A 2023 study of China's smart city pilot policy found that pilot cities cut their carbon emissions by about 4.36% compared to non-pilot cities, with the effect growing stronger over time [10]. Another 2024 analysis using a similar method reported an even larger average reduction of 11.4% in carbon dioxide emissions [5]. These are not tiny numbers: for a mid-sized city emitting 10 million tons of CO2 per year, an 11% cut equals 1.1 million tons saved annually.

The biggest single-technology gains come from AI-powered digital twins. A 2025 study found that combining artificial intelligence with digital replicas of city systems cut energy use in smart buildings by 25-30%, reduced transport carbon emissions by 20%, and improved renewable energy integration by 20% [1]. Similarly, cities that deployed cyber-physical systems and Internet of Things (IoT) networks reported energy savings of up to 25% and waste management improvements of 30%, directly lowering their carbon footprint [2]. These figures show that the technology works, but the total city-level reduction depends on how many systems are upgraded and how well they are integrated.

What are the main ways smart cities reduce emissions?

Smart cities cut carbon through three main pathways: making energy use more efficient, spurring green technology innovation, and improving how resources are allocated across the city. The efficiency pathway is the most direct. AI-enhanced digital twins continuously monitor and optimize energy use in buildings, traffic systems, and power grids, cutting waste without sacrificing performance [1]. IoT sensors and real-time data analytics let cities adjust street lighting, heating, and traffic flows to match actual demand, reducing unnecessary energy consumption [11].

The innovation pathway is equally important. Smart city construction pushes companies and governments to develop and adopt cleaner technologies. A 2024 study found that smart city pilots in China boosted green patent output and total factor productivity, meaning cities produced more economic output per unit of carbon emitted [5]. Another study showed that smart city policies improved carbon emission efficiency by 1.4% through green technology innovation, industrial structure upgrading, and energy structure optimization [6]. The third pathway—resource allocation—works by using data to match energy supply with demand more precisely. Smart grids cut transmission losses by 15% and increased renewable energy use by 20% [1]. Digital infrastructure also helps cities shift away from coal and toward cleaner energy sources [7][9].

When do smart city carbon reductions work best—and what are the limits?

Smart city carbon reductions are not automatic; they depend on city characteristics, timing, and policy design. The benefits are strongest in cities with high human capital, strong financial resources, and advanced digital infrastructure [5][9]. For example, smart city policies cut carbon more effectively in eastern and southern Chinese cities, non-resource-based cities, and large or medium-sized cities [6][8]. Resource-based cities (those dependent on mining or heavy industry) see smaller gains unless they are at a specific development stage [3].

There is also a timing lag. One study found that the carbon reduction effect of smart city construction does not appear until the third year after implementation and then grows stronger over time [10]. Another study noted that in the early years, building digital infrastructure can actually increase energy use because data centers and networks are energy-intensive [8]. This means cities need patience and sustained investment. Additionally, the carbon reduction effect varies by sector: smart city policies have a bigger impact on industrial production and waste treatment than on household consumption [5]. Finally, privacy-preserving approaches like federated learning can predict carbon footprints accurately while protecting data, but they require coordination across multiple cities to work at scale [4].

Sources used in this answer

Related research questions