There is no single number, because “a city” can mean anything from a town of 50,000 to a sprawling metro of 20 million. As a rough guide, a small city of around 100,000 people might draw somewhere in the range of 100 to 300 megawatts on an average day, while a large metropolitan area can require several thousand megawatts at its peak. Even for one specific city, the answer swings wildly depending on the time of day, the season, and whether the local economy runs on office buildings or steel mills. The interesting part is understanding what drives those differences and why the question is harder to pin down than it sounds.
A Real-World Example From Phoenix
One of the clearest snapshots of city-scale power demand comes from a study of the greater Phoenix, Arizona area during a heat wave. Air conditioning alone pushed demand to roughly 8,000 megawatts at its peak, settled around 5,000 MW on average across the day, and dropped to about 2,000 MW at its lowest point overnight.1Applied Energy. Reducing a semiarid city’s peak electrical demand using distributed cold thermal energy storage That is just air conditioning in a metro area of roughly 4.5 million people during extreme heat. Add in lighting, commercial buildings, data centers, industry, and everything else plugged in, and the total demand is even higher.
The fourfold swing between the overnight minimum and the afternoon peak is itself remarkable. At 2 AM, the grid is coasting. By 4 PM on a 115°F day, it is straining. This is why grid planners do not think in terms of a single megawatt figure for a city. They think about the peak, the trough, and how fast demand ramps between them.
Why Bigger Does Not Simply Mean More
You might assume that a city twice the size of another uses exactly twice the electricity. The reality is more interesting, and researchers have found that the relationship between population size and electricity use depends heavily on where the city is and how its economy is structured.
In the United States, water and electricity consumption scale sublinearly with population, meaning that as cities get bigger, they tend to use resources more efficiently per person.2Environmental Research: Infrastructure and Sustainability. Urban scaling of water and electricity demand across the United States Shared infrastructure, denser housing, and more efficient commercial systems all contribute. A resident of a large American city effectively uses a bit less electricity than someone in a smaller town, on average, when you account for the city’s total load divided by its population.
In China, the pattern flips for residential electricity. Larger Chinese cities consume disproportionately more residential electricity per person, with the scaling exponent hovering between 1.11 and 1.18 over time.3Journal of Urban Management. Residential electricity use scales superlinearly with city size in China Residents in bigger Chinese cities tend to have higher incomes, more appliances, and more air conditioning, which collectively push per-capita consumption above what you would predict from population alone. Total and industrial electricity use in China also showed a super-linear pattern, though that trend has been moving toward a more linear relationship over time.
The takeaway is that “how many megawatts per million people” is not a fixed ratio. It shifts depending on a country’s development stage, its building stock, climate, and whether the economy is driven by industry or services. A rough American ballpark often cited by grid operators is somewhere around 1 to 1.5 kilowatts of average demand per person, but that collapses a lot of variation into one number.
Climate Is One of the Biggest Wild Cards
Air conditioning and heating together represent a massive chunk of urban electricity demand, and climate dictates how large that chunk is. Phoenix’s extreme summer demand illustrates the point in a desert city, but the relationship holds worldwide and is getting more dramatic as temperatures rise.
A study modeling climate change impacts on electricity demand across world cities found that the change in annual per-capita demand ranged from about a 2.7% decrease to a 5.7% increase relative to current conditions by mid-century.4Scientific Reports. Heterogeneous climate change impacts on electricity demand in world cities circa mid-century Those numbers sound modest, but the impact on peak demand is where it gets serious. The change in peak hourly demand ranged from a 3.4% decrease to a 9.5% increase. In most cities studied, the extra electricity needed for cooling on hotter days outweighed the savings from reduced heating in milder winters. Grids are sized to handle the peak, not the average, so even a modest percentage increase in peak demand can force billions of dollars in new generation and transmission infrastructure.
This creates a self-reinforcing pattern in hot-climate cities. Hotter summers drive more air conditioning adoption, which pushes up peak demand, which strains the grid, which requires more power plants that often produce waste heat themselves. Cities in the tropics and subtropics face the steepest demand growth curves under warming scenarios, and these are also the cities growing fastest in population.
The Peak-to-Trough Problem
When people ask how many megawatts it takes to power a city, they usually imagine a steady, consistent number. But a city’s electricity demand is more like a heartbeat than a flatline. The Phoenix data makes this vivid: the same metro area that needed 8,000 MW at its afternoon peak dropped to 2,000 MW overnight, a fourfold difference within a single day.1Applied Energy. Reducing a semiarid city’s peak electrical demand using distributed cold thermal energy storage
This daily cycle follows a fairly predictable pattern in most cities. Demand is lowest in the early morning hours, begins climbing as people wake up and businesses open, and hits its highest point in the late afternoon on hot days or early evening in colder climates when heating and lighting overlap. The exact shape of this curve varies by season and by what drives a city’s economy. A city dominated by heavy industry will have a flatter demand curve because factories run around the clock. A residential and commercial city will have sharper peaks and deeper valleys.
The peak is what grid operators have to build for. Every power plant, every transmission line, every transformer must be sized to handle the single worst hour of the year, even though most of that infrastructure sits partly idle the rest of the time. This is one reason utilities care so much about demand management: shaving even a small percentage off the peak can defer or eliminate the need for an entirely new power plant.
How Electric Vehicles and New Loads Are Changing the Math
Urban electricity demand is not static, and several new categories of consumption are growing fast enough to reshape city-level power needs within the next decade. Electric vehicles are among the most discussed.
The impact depends enormously on charging speed. A study examining the effect of EV charging on retail building sites found that fast-charging stations, with power levels as high as 350 kilowatts per charger, have the potential to dwarf a large retail building’s own power consumption if they share the same electrical connection, increasing monthly peak power demand at the site by over 250%.5Advances in Applied Energy. Impact of electric vehicle charging on the power demand of retail buildings That is a single commercial site, but multiply it across a city with thousands of fast chargers and the aggregate effect on the grid becomes substantial. The timing matters as much as the total: if everyone plugs in their car when they get home from work at 6 PM, the existing evening demand peak gets even sharper.
Data centers represent another fast-growing source of urban electricity demand. A single large data center campus can draw 100 MW or more, equivalent to a small city on its own. As artificial intelligence workloads scale up, some projections suggest that data center electricity consumption could double or triple within a decade in certain regions. For a mid-sized city that hosts one or two major data center clusters, this can become the dominant driver of new demand growth, outpacing all residential and commercial growth combined.
Solar Panels and How Local Generation Reshapes the Curve
Rooftop solar panels are quietly changing how much electricity cities actually pull from the centralized grid, even if the total amount of electricity consumed in the city does not drop. Research in Spain found a significant reduction in grid electricity demand during daylight hours as the number of self-consumption solar installations increased.6PubMed Central. The impact of photovoltaic self-consumption on the daily electricity demand in Spain: Definition of a model to estimate it The total electricity consumed in those buildings did not necessarily change much; instead, a portion of it was generated on-site rather than drawn from the grid.
This has a useful side effect. In sunny climates, solar output peaks in the early afternoon, which overlaps with the hours when air conditioning is ramping up. That overlap means rooftop solar can chip away at the very time period when the grid is under the most stress. The effect is even more pronounced when combined with smart management of electric vehicles. A study of the Yangtze River Delta region in China found that combining orderly EV charging and discharging schedules with rooftop solar could reduce peak grid load by roughly 21.9 gigawatts and cut load fluctuation intensity by about 50% compared to a baseline scenario.7Transport Policy. Assessing the potential of combined vehicle-to-grid and rooftop photovoltaics strategies for reshaping city-level power load profiles
The catch is that solar disappears after sunset, right when residential demand often peaks in cooler climates. Without battery storage or other flexible resources to fill the gap, solar alone can create a new problem: a steep ramp in grid demand as the sun goes down and millions of homes switch back to grid power simultaneously. Grid planners call this the “duck curve” because of the shape it produces on a demand chart, and managing it is one of the central challenges of integrating high levels of solar into urban grids.
Demand Response and Shaving the Peak
Since the peak hour is what determines how much generation capacity a city needs, even small reductions during those critical hours can have an outsized impact. Demand response programs aim to cut electricity use during peak periods by adjusting building systems, shifting industrial processes, or incentivizing consumers to reduce consumption temporarily.
A study of buildings in Changsha, China, tested several demand response strategies during a three-hour peak event. Adjusting thermostat setpoints alone cut consumption by about 8%, while controlling plug loads achieved a similar reduction. Combining multiple strategies together, including thermostat resets, lighting adjustments, and plug load controls, achieved a 22% reduction in electricity consumption during the event.8Energy and Buildings. Assessment of electricity consumption reduction potential for city-scale buildings under different demand response strategies Applied across an entire city during a heat wave, a 22% reduction in building electricity demand for even a few hours can translate to hundreds or thousands of megawatts of avoided peak demand, potentially keeping the grid from tipping into blackout territory.
The limitation is that demand response works best for short-duration peaks. You can ask people to tolerate a slightly warmer building for three hours; you cannot ask them to do it for three days. And the savings come from deferring consumption, not eliminating it. Once the demand response event ends, there is often a rebound as thermostats recover and deferred loads kick back in. Still, as a tool for managing the difference between what a city needs at 3 PM versus what it needs at 3 AM, demand response is one of the cheapest options available.
Microgrids and What Happens When the Main Grid Fails
The question of how many megawatts it takes to power a city implicitly assumes the city is connected to a large regional grid. But portions of cities increasingly operate on microgrids, smaller self-contained power systems that can disconnect from the main grid and run independently during outages. Hospitals, military bases, university campuses, and critical municipal facilities are common candidates.
A microgrid does not need to power an entire city. Its value lies in keeping essential services running when the larger grid goes down. Research on microgrid deployment at critical facilities found that a well-designed system can withstand over 700 hours of grid outage over its lifetime while saving the facility tens of thousands of dollars compared to relying solely on the main grid and backup diesel generators.6PubMed Central. The impact of photovoltaic self-consumption on the daily electricity demand in Spain: Definition of a model to estimate it That is not powering the whole city, but it reframes the question: maybe a city does not need every megawatt to come from one centralized system. Distributing generation and storage across many smaller systems can make the overall network more resilient, even if the total megawatts consumed stay the same.
Putting Numbers in Context
If you want a rough mental model, here is how to think about it. A single large power plant, whether gas-fired, nuclear, or a major wind farm, typically has a capacity somewhere between 500 and 1,500 MW. A small city of 100,000 people might need one or two such plants’ worth of capacity to handle its peak demand. A major metro area of 5 to 10 million people might need the equivalent of a dozen or more large plants, depending on climate and economic activity. And those plants do not all run flat out all the time; the fleet ramps up and down to follow the city’s demand curve throughout the day.
The number also depends on what you count. If you draw the boundary around just the city limits, you get one figure. If you include the surrounding suburbs and exurbs that share the same grid infrastructure and commute into the city for work, you get a much larger one. Grid operators typically think in terms of service territories and balancing areas rather than municipal boundaries, which is why published demand figures for “a city” can vary wildly depending on who drew the map.
What makes the question genuinely hard to answer with a single number is that electricity demand is not a fixed property of a city the way its area or altitude is. It is a living, breathing thing that changes hour by hour, season by season, and decade by decade. The megawatts a city needed ten years ago are not the megawatts it needs today, and the megawatts it needs today are not what it will need in 2040 after millions of electric vehicles, more air conditioning, and more data centers have been added to the mix. The best answer to “how many megawatts does it take” is always “it depends, and it is changing.”