Could an SMR Power Your City? Here's What the Numbers Say
The math on small modular reactors is more interesting than the hype, and a lot more complicated.
Picture a city of 200,000 people. Coffee shops, hospitals, traffic lights, office towers, data centers humming away in the background. This city needs electricity around the clock, and it doesn’t care whether the wind is blowing or the sun is out. How much power does that actually take? And could a small modular reactor supply it? The answers are less obvious than most headlines suggest, and more encouraging than most critics will admit.
Let’s do the math together.
What it actually takes to power a city
Numbers first, then the context. According to analysis from the U.S. Energy Information Administration, the average American household consumes roughly 10,791 kilowatt-hours per year, or about 899 kWh a month. That’s just homes. Stack on commercial buildings, street lights, factories, hospitals, and the picture gets considerably heavier. 🏙️
A useful engineering rule of thumb: expect about 1 to 2.5 kilowatts of continuous average load per person in a typical developed-world city. Peak demand, which is what grid planners actually design for, runs 1.5 to 2 times higher than that average, typically hitting on a hot summer afternoon when every air conditioner in town switches on simultaneously. The actual figures stack up like this:
A small city of 50,000 to 100,000 people: roughly 50 to 100 MW of peak capacity needed
A mid-size city of around 500,000 people: between 500 and 1,500 MW, depending heavily on climate and industry
A large metropolitan area of 5 million or more: upward of 8,000 to 12,000 MW at peak ⚡
Austin, Texas, a city of around 1 million people, consumes roughly 20 billion kilowatt-hours per year. That’s not a number you can drop casually into conversation. It works out to continuous average demand somewhere north of 2,000 MW. And Austin is not an industrial powerhouse. It’s a tech and government hub with a hot climate. The electricity demand of a real city is humbling.
What makes this harder is that power grids don’t run on averages. They’re engineered for worst-case moments. A city planner who builds just enough capacity for average demand will face blackouts whenever demand spikes. 🔋 This is one reason the dispatchability argument for nuclear keeps coming back in policy debates: it’s not about average output, it’s about reliable output whenever you need it.
How many SMRs does a city actually need?
The International Atomic Energy Agency defines an SMR as a reactor producing up to 300 megawatts electric (MWe) per unit. That’s about one-third the output of a traditional large nuclear plant. The European Commission notes that SMRs can range all the way down to 20 MWe, which is not powering anyone’s metropolis on its own.
The realistic commercial designs in front of utilities today cluster in a more useful range. The GE Vernova Hitachi BWRX-300 produces 300 MWe. The Rolls-Royce SMR targets 470 MWe, a deliberate design choice to cross the threshold where you’re powering something meaningful on a single unit. NuScale’s VOYGR-6 stacks six 77 MWe modules to reach around 462 MWe total. These are not the same product, and the differences in output matter enormously when doing city-scale math. 🔬
Run the numbers for that 200,000-person city with mid-range demand:
Estimated continuous average load: roughly 200 to 400 MW
Peak demand at 1.5x average: 300 to 600 MW
One BWRX-300 (300 MWe): covers the city’s average load comfortably, falls short of peak
Two BWRX-300s: covers peak demand with some margin left for industrial growth
Three BWRX-300s: provides comfortable headroom, redundancy, and potential export capacity
This is not a plug-and-play scenario, which is partly the honest limitation and partly the appeal. SMRs are designed to be added incrementally as demand grows, unlike a traditional gigawatt plant that you either build in full or don’t. The IAEA notes that SMRs should represent no more than 10 percent of a grid’s total installed capacity to avoid stability problems, which means isolated deployment only makes sense if the city is also connected to a regional grid. 🌐
Think about this for your own city: could you imagine the local utility adding one SMR module now, then another as demand grows, rather than committing to an all-or-nothing megaproject? That’s the pitch.
The economics: what the numbers actually say
Here is where things get genuinely complicated, and where I think a lot of coverage fails readers by picking a side instead of reporting what the data shows.
The current levelized cost of electricity (LCOE) for SMRs runs roughly $89 to $102 per MWh based on first-of-a-kind project estimates. That is unambiguously higher than utility-scale solar ($28 to $117/MWh) and onshore wind ($23 to $139/MWh). Honest journalism requires acknowledging that gap upfront. 💡
But the ranges for renewables mask a critical issue. Those lower bounds assume ideal conditions: premium solar sites in the American Southwest or wind corridors in the Great Plains. They don’t include the cost of backup generation for when the sun sets or the wind calms, which is the key economic argument SMR developers make. A nuclear plant runs at a capacity factor of roughly 92 percent. Solar averages about 27 percent. Onshore wind, about 37 percent. What the LCOE comparison leaves out is the system cost of intermittency: the cost of backup capacity, grid upgrades, and storage that renewables require to be firm power.
Canada’s Darlington project, now under construction, puts the real cost of first-of-a-kind SMR deployment in sharp focus:
First BWRX-300 unit: CAD $7.7 billion (approximately USD $5.6 billion) for 300 MWe
Four units total: CAD $20.9 billion for 1,200 MWe of installed capacity 📈
Expected operation: 65 years
Construction timeline: approximately three and a half years from breaking ground to grid
The cost per unit drops significantly for units two, three, and four, as the supply chain, regulatory learning, and factory production ramp up. GE Hitachi originally designed the BWRX-300 with a target cost of USD $700 million per reactor. That target has not been met at Darlington, and the International Energy Agency says SMRs need to reach $4,500 per kilowatt by 2040 to achieve mainstream adoption. That’s the goal line. The gap between where costs are now and where they need to be is real, and nobody in the industry pretends otherwise.
The Wood Mackenzie analysis puts it plainly: if SMR costs fall to $120/MWh by 2030, they become competitive with conventional nuclear, natural gas, and coal. That’s a meaningful target, and whether first movers like OPG can pull costs down to that level will determine whether this technology becomes mainstream or stays a niche. 💰
What do you think: is first-of-a-kind cost overrun acceptable if it brings down costs for the next fifty projects? That’s the bet that governments in Canada, the UK, and the US are making right now.
Who is already betting on this
Real money is moving. That matters more than projections. 🚀
Russia has been running its Akademik Lomonosov, a 70 MWe floating nuclear plant, in the Far Eastern city of Pevek since 2020. It’s not glamorous, but it has generated over a billion kilowatt-hours of electricity. China’s ACP-100, a 125 MWe pressurized water reactor at Changjiang in Hainan Province, became the world’s first operational commercial land-based SMR in late 2025. Canada is building its BWRX-300 at Darlington. These aren’t announcements. They’re done deals or active construction sites.
The tech sector’s involvement is the other major signal. Consider what’s been committed:
Google has contracted 500 MWe of SMR power from Kairos Power, targeting 2030 operations, to power its AI data centers in the southeastern US
Amazon committed over $500 million toward SMR development with the Energy Northwest consortium in Washington State, targeting four units totaling 960 MW
The US Department of Energy allocated $400 million each to Tennessee Valley Authority and Holtec in December 2025 to support early deployment 🏗️
The UK is running its own parallel track. The government has backed Rolls-Royce SMR’s 470 MWe design with GBP 280 million in funding, with four sites identified and grid connection targeted for the mid-2030s. The UK, Canada, and US signed the Atlantic Partnership for Advanced Nuclear Energy to synchronize regulatory approvals and share safety assessments.
What’s striking about the tech sector’s involvement is that it reframes the question entirely. Google and Amazon aren’t betting on SMRs because they’re cheap. They’re betting on them because they need guaranteed, dispatchable, 24/7 carbon-free power that solar and wind alone can’t reliably provide for always-on computing infrastructure. The economics look different when reliability is the primary requirement, not lowest cost per MWh.
What this means for your city
Could an SMR power your city? The straight answer: yes, probably, depending on the city’s size and where you are on a regional grid. 🌍
A city of 100,000 people could realistically be served by a single 300 MWe SMR, with surplus power going to the grid. A city of 500,000 would need four to six units running in parallel. A major metropolitan area like London or Tokyo needs dozens of gigawatts total, so SMRs work as part of a diverse mix rather than the whole answer. The World Nuclear Association tracker currently shows 127 modular reactor designs globally, with seven already operating or under construction. That pipeline is growing fast.
The realistic timeline for a first Western commercial SMR serving a city’s grid is roughly 2029 to 2032, based on current project schedules. That’s not this decade’s solution for climate targets already due. It’s the next decade’s workhorse if construction goes to plan and costs come down at scale. The gap between where costs sit today and where they need to be is the central challenge for this technology, not the physics, not the safety record.
The physics already works. The safety record of nuclear energy is strong. The question is whether the economics can follow, and whether cities and utilities are willing to be early adopters who absorb first-of-a-kind costs to bring those prices down for everyone who comes after. 🔌
Here’s the question worth sitting with: if your city’s utility offered residents a vote on building an SMR 50 miles away, fully carbon-free, with power for 65 years, at a cost 20 to 30 percent higher than current rates for the first decade but stable thereafter, would it pass?