Solar power broke every global record in 2025. Nuclear power had one of its biggest years too, for reasons that rarely make the headlines.
Last updated June 2026 | 13-minute read
Image Credit: Leonardo AI
News summary
- Solar power reached nearly 2,900 GW of global installed capacity by the end of 2025, the fastest energy scale-up in recorded history.
- California's grid operator curtailed a record 3.4 million megawatt-hours of solar and wind output in 2024 because midday supply kept outpacing demand.
- More than 40 countries now have active plans to expand nuclear power, with over 70 GW currently under construction worldwide.
- HALEU, the specialized fuel several next-generation reactors need, remains so limited that one Ohio facility is currently the only producer in the United States.
- Google, Microsoft, Amazon, and Meta signed contracts for more than 10 GW of new nuclear capacity in the United States over the past year, mostly to power data centers that cannot tolerate any interruption.
- Nuclear plants operate at a 92.7 percent capacity factor on average, nearly four times higher than solar's 24.8 percent, though that average can drop sharply during extreme weather.
Everyone said solar had already won. Prices dropped 90 percent. Panels spread across deserts and rooftops, and headlines cheered every new gigawatt. At the same time, quietly and with billions of dollars on the line, country after country kept building nuclear reactors, not instead of solar but alongside it. The reason is more specific and more interesting than most coverage of this debate lets on.
In this article
- Solar power in 2026: the numbers behind the record year
- The curtailment problem behind the installed capacity number
- The nuclear revival most coverage is underplaying
- Ten countries are still building nuclear reactors right now
- Why solar alone cannot close the reliability gap
- Why nuclear's reliability number has a seasonal blind spot
- The grid connection queue that slows down both technologies
- Why big tech is quietly turning to nuclear power
- The fuel bottleneck behind the next generation of reactors
- How grid planners use solar and nuclear together
- The real obstacles nuclear still has to overcome
- Checking five common claims against the data
- Solar and nuclear in the same grid
- DesiDaily take
Solar power in 2026: the numbers behind the record year
Solar's rise has been genuinely historic, the kind of growth that keeps forcing economists and energy analysts to revise their own forecasts upward, because almost no model predicted it would move this fast.
Solar panel prices fell by more than 90 percent between 2010 and 2025. A technology that once cost around 30 dollars per watt now costs under 0.50 dollars per watt, according to data tracked by the International Energy Agency. In 2025 alone, Ember Energy reported that 647 GW of new solar capacity was installed, an 11 percent jump from the previous year. Cumulative global solar capacity crossed close to 2,900 GW by late 2025.
A University of Surrey study published in late 2025 confirmed that solar energy now costs as little as 2 pence per kilowatt-hour in regions with strong sunlight, cheaper than coal, natural gas, or wind on a new-build basis. In many countries, building a new solar farm now costs less than simply running an existing coal plant.
Solar currently generates roughly 10 percent of the world's electricity and is projected to reach 20 percent by 2029, according to DNV's Energy Transition Outlook. That is a structural shift in how the world makes electricity, visible on rooftops, desert plains, and national energy balance sheets alike.
So the obvious question follows: if solar is this affordable and this scalable, why is nuclear making a meaningful comeback at the same time? Part of the answer involves a problem most solar headlines skip past, covered in the next section.
The curtailment problem behind the installed capacity number
The 2,900 GW figure measures what solar panels can produce under ideal sunlight, not what the grid actually lets them deliver. The gap between those two numbers is growing every year, and it explains part of why nuclear has not lost its case despite solar's price collapse.
In California, grid operator CAISO curtailed 3.4 million megawatt-hours of utility-scale wind and solar output in 2024, a 29 percent increase over 2023, according to the U.S. Energy Information Administration. Solar accounted for 93 percent of everything curtailed that year. Curtailment happens when supply exceeds what the grid needs at that moment, so the operator orders panels to produce less even though the sun is still shining.
The reason is timing. Average net demand in CAISO during the 9 a.m. to 3 p.m. window has fallen 45 percent since 2020, mostly because so much solar capacity already covers that window. On sunny spring afternoons, wholesale electricity prices in California regularly turn negative, meaning generators are effectively paid to produce less rather than more, a pattern often called the solar duck curve.
Battery storage is closing part of that gap. CAISO's battery capacity grew from about 500 megawatts in 2020 to more than 13 GW by early 2025, largely to absorb midday solar that would otherwise go to waste and release it during the evening peak. Texas and other solar-heavy grids are starting to see the same pattern develop.
None of this erases solar's cost advantage. It does mean a new solar project's lifetime revenue depends heavily on how much storage and transmission capacity exists around it, a detail a single global capacity figure cannot show.
The nuclear revival most coverage is underplaying
Here is a figure most energy journalism buries deep in the article: more than 70 GW of new nuclear capacity is currently under active construction around the world. That marks one of the highest levels of nuclear construction activity in three decades, confirmed by the IEA in its landmark 2025 nuclear energy report.
IEA Executive Director Fatih Birol described the situation plainly in that report:
"It is clear today that the strong comeback for nuclear energy that the IEA predicted several years ago is well underway, with nuclear set to generate a record level of electricity in 2025."
Fatih Birol, Executive Director, International Energy Agency, 2025According to the World Economic Forum, approximately 30 countries are currently considering, planning, or starting nuclear power programs. At COP28, more than 20 nations signed a pledge to triple global nuclear capacity by 2050. By the time COP30 arrived, 33 nations had formally committed to that target.
Governments do not commit billions to a passing trend. You can read more about who actually controls the world's uranium supply and how those geopolitical realities shape every decision behind these numbers.
Three-quarters of all reactors currently under construction sit in emerging economies, with half of the global total in China alone. China currently operates 57 reactors and is building 29 more, on track to overtake the United States as the world's largest nuclear power producer before 2030, according to the IEA's 2025 projections.
Ten countries are still building nuclear reactors right now
These are not speculative proposals sitting on a government white paper. Each of these countries has signed contracts, committed public or private capital, and, in most cases, already broken ground.
All programme data above comes from the World Nuclear Association's 2025 Outlook Report, cross-referenced with IEA and IAEA figures. These are active programmes with allocated budgets, not aspirational statements.
Why solar alone cannot close the reliability gap
Here is the part that sounds simple but carries enormous consequences: the sun sets every day.
Solar panels generate electricity when sunlight is available. The deeper challenge is the mismatch between when solar produces the most power, around midday, and when households and industry need it most, which is early evenings, cold winter mornings, and heatwave nights when air conditioning runs nonstop.
Image Credit: Leonardo AI
According to the U.S. Department of Energy, nuclear plants operate at a 92.7 percent capacity factor, meaning they produce electricity at or near full output for more than 92 percent of every hour in a year. Solar averages 24.8 percent. Wind sits between 35 and 42 percent.
Spain and Portugal experienced one of Europe's worst electricity blackouts in decades on April 28, 2025. Approximately 15 GW of generating capacity disappeared from the Iberian grid in under five seconds. Emergency gas turbines had to be brought online to restore power across both countries.
Spain has a scheduled plan to phase out its nuclear reactors by 2035. Within days of the blackout, Spain's nuclear industry formally urged the government to reconsider that timeline.
To put the reliability difference in plain terms: replacing one 1 GW nuclear reactor with solar requires roughly four times more installed solar capacity to match the same annual electricity output, and even then, a grid still needs either large-scale battery storage or a backup power source for sunless stretches of days.
| Factor | Solar power | Nuclear power |
|---|---|---|
| Capacity factor | 24.8% | 92.7% |
| Levelized cost per kWh | $0.02 to $0.04 in sunny regions | $0.06 to $0.09 average |
| Typical build time | Months to 2 years | 10 to 20 years for large plants |
| 24/7 reliability | No, weather dependent | Yes, weather independent |
| Land use per GW output | High | Very low |
| CO2 lifecycle emissions | Approx. 40 g CO2 per kWh | Approx. 12 g CO2 per kWh |
| Grid stability role | Variable adds intermittency | Baseload anchors the grid |
| Home-scale deployment | Yes, rooftop viable | No, utility scale only |
Data is sourced from the U.S. Department of Energy, Science Times's 2026 energy analysis, and the University of Surrey cost study. Solar wins on cost and deployment speed. Nuclear wins on reliability, land efficiency, and grid stability. Neither technology wins on every metric, which is exactly why most serious grid plans use both.
There is also history behind why nuclear, despite its well-documented flaws, remains a strategic safeguard that most energy-importing nations are unwilling to abandon, and the choice of uranium over thorium as the world's dominant nuclear fuel decades ago locked in infrastructure that still shapes today's decisions.
Why nuclear's reliability number has a seasonal blind spot
The 92.7 percent capacity factor cited above is a fleet-wide annual average. It does not mean every reactor produces full output every week of the year, and two real events from the past five years show where that average can break down.
France's nuclear fleet sits mostly along inland rivers, and French regulation requires reactors to reduce output when river water gets too warm or river flow drops too low to protect downstream ecosystems. In 2022, that heat-related curtailment combined with a separate and larger problem, a stress corrosion issue discovered across the fleet's newer reactors, pushed national nuclear output down to 279 terawatt hours, the lowest level since 1988 and roughly 30 percent below the 20-year average, according to grid operator RTE's 2022 annual electricity review. Availability briefly fell to around 40 percent of capacity for about a month. Heat curtailment alone was a smaller piece of that story. Research from the Clean Air Task Force found that heat-related output reductions have cut France's total nuclear generation by only about 0.15 percent since 2000, with the corrosion repairs responsible for most of the 2022 shortfall.
Cold weather can cause the same kind of disruption. During Winter Storm Uri in February 2021, Unit 1 of the South Texas Project nuclear plant automatically tripped offline after a cold-weather-related failure in a pressure-sensing line tied to its feedwater pumps, according to World Nuclear News. The plant lost about 1,280 megawatts of capacity for nearly three days during the exact stretch when Texas needed every megawatt it could get. The other three Texas reactors kept running at full power throughout the storm.
Neither event changes the underlying math. Nuclear still runs far more consistently than solar or wind across a full year. What these cases show is that the 92.7 percent figure describes a fleet's annual performance, not a guarantee for any single plant during the exact extreme weather week when the grid needs it most, which is the same vulnerability often raised against renewables.
The grid connection queue that slows down both technologies
Build-time comparisons between solar and nuclear usually stop at construction. They skip the step that happens before either one can sell a single watt: getting permission to connect to the transmission grid.
As of the end of 2024, roughly 2,290 GW of generation and storage capacity sat in U.S. interconnection queues, according to Lawrence Berkeley National Laboratory's annual Queued Up report, nearly double the capacity of the entire existing U.S. power fleet. Total queued capacity had topped 2,600 GW earlier in 2025 before a wave of project withdrawals pulled the number back down. Solar and storage make up roughly 80 percent of new capacity entering these queues, and hybrid solar-plus-storage projects, which require the most complex interconnection studies, now account for more than half of all active solar and storage capacity waiting in line.
The wait itself has grown steadily longer. The median time from an interconnection request to commercial operation now runs close to five years, up from under two years for projects completed between 2000 and 2007. Berkeley Lab's data shows that only about 19 percent of projects entering U.S. queues between 2000 and 2018 ever reached commercial operation. Most are eventually withdrawn.
The Federal Energy Regulatory Commission's Order 2023 introduced a first-ready, first-served cluster study process along with stricter site-control and deposit requirements, aimed at clearing speculative projects out of the line so that serious ones move faster. Queue volume began falling for the first time in years, though clearing a backlog this size still takes time.
Nuclear projects move through a smaller queue since far fewer of them apply at any one time, but they depend on the same scarce transmission corridors that solar and storage projects are waiting years to access. A new reactor with a finished construction schedule can still sit waiting for the transmission upgrade needed to carry its power to market. Both technologies share a bottleneck that has nothing to do with reactors or panels and everything to do with the wires connecting them to everyone else.
Why big tech is quietly turning to nuclear power
Here is the angle most mainstream energy coverage misses: artificial intelligence is accelerating the nuclear revival for reasons that have little to do with climate pledges or government targets.
Artificial intelligence data centers operate around the clock, seven days a week, with no tolerance for power interruptions. A solar farm delivering intermittent output during a cloudy fortnight is not a workable clean energy source for a large language model. It is a liability. And the energy demands of AI infrastructure are growing faster than even optimistic grid planning models expected.
According to Introl's December 2025 analysis of nuclear deals in the technology sector, major technology companies signed agreements for more than 10 GW of new nuclear capacity in the United States during 2024 and 2025 alone.
Microsoft signed a 20-year, 16 billion agreement to restart the Crane Clean Energy Center at Three Mile Island, providing 835 MW dedicated to its AI data centers across the northeastern United States.
Google backed Kairos Power with a 500 MW small modular reactor agreement, targeting first electricity generation by 2030, the first corporate small modular reactor deployment of its kind at this scale.
Amazon committed more than 20 billion dollars to convert its Susquehanna facility into a nuclear-powered AI campus, with multiple reactor units planned across a phased build-out.
Meta signed a 20-year, 1.1 GW nuclear power purchase agreement with Constellation Energy's Clinton Clean Energy Center in Illinois, one of the largest single corporate nuclear deals in American history.
The IEA projects electricity consumption rising six times faster than total energy use in the coming decades, with artificial intelligence, data centers, electric vehicles, and industrial electrification accounting for most of that new demand. Solar can cover large portions of daytime requirements. Nuclear handles the rest continuously, without checking a weather forecast.
It is worth tracking closely. Understanding some of the regulatory gaps that still exist in the nuclear fuel supply chain matters, as this commercial expansion accelerates beyond traditional government-led programs, a gap explored further below.
Image Credit: Leonardo AI
The fuel bottleneck behind the next generation of reactors
Most of the advanced reactor designs named above, including those built around newer fuel choices, need a type of fuel called HALEU, short for high-assay low-enriched uranium, enriched to between 5 and 20 percent uranium-235. That is a meaningfully higher concentration than the fuel running today's existing reactor fleet, and it requires a different part of the enrichment supply chain entirely.
Until 2024, Russia's state nuclear company Rosatom, through its subsidiary TENEX, was the only commercial supplier of HALEU anywhere in the world, according to the World Nuclear Association. The United States banned imports of Russian enriched uranium in May 2024 under the Prohibiting Russian Uranium Imports Act, with limited waivers running through 2027 while domestic producers ramp up, according to the U.S. Department of Energy.
That left a gap that has not yet closed. As of mid-2025, Centrus Energy's demonstration plant in Piketon, Ohio, remained the only HALEU producer in the United States, having delivered just over 900 kilograms in total since its cascade started running in October 2023, a small fraction of the multi-ton annual quantities reactors such as TerraPower's Natrium, X-energy's Xe-100, and Kairos Power's Hermes will eventually need. The U.S. government has committed more than 2.7 billion dollars to Centrus, General Matter, and Orano to build out domestic enrichment capacity, but new enrichment facilities take years to license and build.
This is why several of the headline big tech nuclear deals mentioned above involve restarting or extending existing large reactors rather than commissioning new small modular ones. Microsoft's Three Mile Island deal and Meta's Clinton plant agreement both run on conventional fuel that is already available. Google's Kairos Power deal, the most advanced corporate small modular reactor commitment announced so far, still targets first power only by 2030. Construction timelines are no longer the only constraint worth tracking for this industry. Fuel availability may end up setting the real pace.
How grid planners use solar and nuclear together
Energy commentary often frames solar and nuclear as rivals competing for the same budget. In practice, grid planners use them to solve different problems on the same system.
Solar provides cheap, distributed, rapidly deployable power that scales from a single suburban rooftop to a 5 GW desert installation. Nuclear provides dense, continuous, weather-independent electricity that anchors a national grid and runs for 60 or more years without interruption. Smart energy systems use both, and the countries that have figured this out are already pulling ahead.
Fast to deploy, months rather than decades. Among the cheapest new electricity in most markets on a per kWh basis. Scales from individual homes to utility-scale farms. Produces no radioactive waste. Improves energy access in remote regions. Can be installed and generate power before a nuclear plant has finished its environmental impact assessment.
Runs day and night, rain or shine, winter or summer, for six decades or more. Requires the smallest land footprint of any energy source per gigawatt of output. Among the lowest carbon lifecycle emissions of any generation technology studied. An essential grid anchor for countries with harsh winters, heavy industrial loads, or round-the-clock AI infrastructure demand. Reduces dependence on fuel imports from politically volatile regions.
France offers the clearest long-run evidence. It generates approximately 67 percent of its electricity from nuclear, the highest share in the world, while expanding solar capacity in its southern regions. The result is one of Europe's most stable grids and one of the lowest domestic electricity carbon intensities on the continent.
Germany took a different path. After Fukushima in 2011, it moved to phase out nuclear entirely and invested heavily in renewables. By 2024, Germany had among the highest household electricity prices in Europe, at times importing power from nuclear-generating neighbors including France. The geopolitical dimensions of fuel dependency are explored further in this analysis of what happens when traditional fuel supply routes are disrupted, which connect to these decisions more closely than most coverage admits.
Uranium supply concentration carries its own risk. A single country now sits at the center of global uranium supply, and the world's largest uranium-importing nations, ranked among the 10 countries importing the most uranium, are acutely aware that concentration creates strategic vulnerability.
The real obstacles nuclear still has to overcome
The nuclear revival is real, but it is not without genuine friction. Acknowledging those obstacles is what separates credible energy analysis from promotional copy.
Radioactive waste management remains unresolved over the long term. High-level nuclear waste needs isolation from the environment for tens of thousands of years. Finland is the only country to complete licensing for a permanent deep geological repository, at Onkalo, with Sweden close behind. Every other major nuclear nation is still working through the political and technical process of finding a permanent solution.
Construction timelines and cost overruns have historically been severe. The Vogtle Units 3 and 4 project in Georgia, the only new large nuclear reactors completed in the United States in decades, came in at approximately 35 billion dollars, roughly double the original estimate, and ran seven years behind schedule. Finland's Olkiluoto 3 reactor took 17 years to complete against an original four-year build window. These are not isolated cases. They represent a systemic pattern in large reactor construction across Western countries.
Small modular reactors are widely cited as the fix, with over 80 different designs currently in development globally. The argument that smaller, factory-built units can cut construction time and cost is credible in theory. As of 2026, no SMR design has demonstrated full commercial deployment at scale in a Western market, and as covered above, fuel supply now adds another layer to that timeline. The first true test cases, including NuScale's revised program and Rolls-Royce's UK SMR design, remain in regulatory review.
Public perception remains a barrier in many countries, shaped by Chernobyl in 1986 and Fukushima in 2011, both of which caused lasting shifts in attitudes that policy alone cannot quickly reverse. Countries without a history of nuclear accidents, or those where enough time has passed for the narrative to shift, are moving faster.
The workforce challenge is significant too. Questions around nuclear expertise and the pipeline of trained specialists deserve attention as this expansion accelerates. Building 70-plus GW of new capacity needs a generation of engineers, regulators, and project managers who do not currently exist in sufficient numbers in most countries, a gap that takes a decade or more to close.
Decommissioning costs add a final layer rarely discussed alongside construction figures. The Nuclear Regulatory Commission estimates decommissioning costs between 280 million and 612 million dollars per reactor, and the NRC's financial assurance data shows the process can legally stretch out for up to 60 years under the SAFSTOR method, during which a retired plant sits in protective storage before final dismantling.
Checking five common claims against the data
Coverage of this topic tends to repeat the same lines from either side without checking them against current numbers. Here is what the data actually shows for five of the most common claims.
| Claim | What the data shows |
|---|---|
| Big tech's nuclear deals prove small modular reactors already work. | Most of the major announced deals, including Microsoft's Three Mile Island restart, Meta's Clinton plant agreement, and Amazon's Susquehanna investment, involve existing large reactors rather than new small modular designs. Google's Kairos Power deal is the most advanced small modular commitment, and it still targets first power in 2030. |
| Nuclear plants are fully funded for cleanup once they close. | The NRC estimates decommissioning costs between 280 million and 612 million dollars per reactor, and the IAEA puts the range as high as 2 billion dollars for some designs. Several utilities have reported funding shortfalls, and cleanup can legally stretch out for up to 60 years. |
| Solar produces no waste, unlike nuclear. | IRENA projects that 78 million tonnes of solar panel waste will accumulate worldwide by 2050. Most retired panels in the United States currently go to landfill, since recovering silicon and silver is not yet economical at scale and there is no federal recycling mandate comparable to the European Union's rules. |
| Adding more solar capacity always lowers electricity bills. | It depends on the hour. Midday oversupply has pushed wholesale prices negative in California, but evening shortfalls still need gas peaker plants, and ratepayers ultimately cover the cost of both the negative-price hours and the backup capacity. |
| Nuclear always costs more than solar. | The comparison changes with the grid. A system with constant industrial demand and little storage values nuclear's steady output very differently than a system with abundant cheap storage and flexible demand. Levelized cost figures from either side of that divide rarely use the same assumptions. |
Solar and nuclear in the same grid
The IEA's 2025 outlook lays out a few different paths for nuclear investment depending on how fast governments move over the next five years. Under its Announced Pledges Scenario, nuclear investment needs to reach 120 billion dollars a year by 2030, nearly double current levels. Under its Net Zero Emissions pathway, that figure climbs above 150 billion dollars annually. Both scenarios treat nuclear and renewables as parts of the same system rather than as substitutes for each other.
What is happening right now has no real precedent. Solar costs fell faster than almost anyone forecast. Battery storage is arriving fast enough to start smoothing out the gaps solar leaves behind. Artificial intelligence has created a new category of round-the-clock electricity demand that barely existed five years ago. A new generation of nuclear technology is trying to avoid the construction delays that defined the last one, while running into a fuel supply problem few expected to be the bottleneck. The 2030s grid will likely look different from anything currently modeled.
Framing solar and nuclear as a binary choice misses how grid planners are actually using them. Solar delivers the cheapest new electricity in most markets and can be installed in months. Nuclear delivers steady, weather-independent output that can anchor a grid for 60 years or more. The evidence from France, China, the United Arab Emirates, and now the data centers of America's largest technology companies points in the same direction: most serious grid plans now include both.
The useful question is whether a single country's grid can afford to depend on only one of these technologies, given how differently solar and nuclear fail.
DesiDaily take
Solar and nuclear power are expanding at the same time because they solve different parts of the same problem. Solar has become the cheapest way to add new electricity in most parts of the world, and its growth in 2025 was historic by any measure. That growth comes with a real limitation: solar only produces when the sun is out, and grids with high solar penetration are already curtailing meaningful amounts of that output during oversupply hours.
Nuclear power's case rests on its ability to produce electricity continuously for decades, which is why artificial intelligence companies with round-the-clock power needs have signed some of the largest nuclear contracts in recent history. That reliability is not absolute. Extreme heat, drought, and extreme cold have all interrupted nuclear output in specific, documented cases, and the next generation of reactors depends on a fuel supply chain still being rebuilt after decades of relying on a single foreign supplier.
Neither technology is free of obstacles, and neither one is winning a race against the other. The data from grid operators in California, France, Texas, and the U.S. Department of Energy point toward the same conclusion: countries serious about reliable, low-carbon electricity are building both, and the ones that leaned too heavily on either one alone have already had to adjust course.