Solar is now the world's biggest source of new electricity, and in 2025,
with help from a mild year, it took coal power's growth. But coal barely fell, the fall is
already reversing in 2026, and China is still building coal plants
faster than it retires them. Solar has taken coal's growth; it has not
yet taken its place.
Solar is taking coal’s growth, not yet its place
Not yet. Solar met three-quarters of the growth in the world’s electricity use in 2025, and coal power fell for only the third time in a decade46.
But coal fell by only about half a percent, a mild year helped, and it is already climbing again in 202687. So solar took coal’s growth in 2025, with help from the weather. It has not pushed coal into a lasting decline: coal still made a third of the world’s electricity, down from 38.7% ten years earlier4.
That's the share of the growth in the world's electricity use that solar met in 2025, while coal power fell only half a percent
Solar added twice the UK’s electricity use in one year
Solar generated about 2,700 terawatt-hours (TWh) of electricity in 2025, a little under 9% of the world total8. A terawatt-hour is a billion kilowatt-hours, and the numbers get easier to hold on a country’s scale: solar’s total output is now the same size as the entire electricity demand of the EU’s 27 countries4.
It grew by about 600 TWh in the year alone8, twice what the whole of the UK uses4. That is the largest rise any power source has had from lasting growth rather than a bounce-back: only coal’s recovery after the Covid lockdowns, in 2021, added more in one year4.
The decade tells the story better than the year. In 2015 coal made about 36 times as much electricity as solar; in 2025, less than four times as much. Over those ten years solar added about 2,500 TWh, and coal still added about 1,2006. Both grew, because the world’s demand for electricity grew faster than solar alone could meet.
Add wind and hydropower, and renewables drew level with coal in 2025. By Ember’s count they passed it, 33.8% to 33.0%4; the International Energy Agency (IEA) calls it near parity and expects the crossover in 202611.
Renewables’ share of the world’s electricity, which drew level with coal’s in 2025.
New solar and wind usually beat new fossil plants on cost
For new power plants, usually yes. The cost of solar power has fallen 89% since 2010, and in 2025 more than 90% of new large-scale renewable projects, solar and wind together, produced electricity more cheaply than the cheapest new fossil-fuel plant would have17.
The fall followed a steady rule. Solar modules get about 20% cheaper every time the world’s installed capacity doubles1. Between 2010 and 2020 solar costs fell by about 15% a year, while nearly 3,000 energy-model projections had assumed 2.6%2. Roughly, that is the difference between costs falling about 80% in a decade and falling about a quarter.
Cheap new solar wins the race for new power plants. It doesn’t close the coal plants that already exist, and that is the main reason coal hangs on. In developing Asia the average coal plant is under 15 years old, against more than 40 in North America14, so most of Asia’s fleet has decades of life left. Owners keep running plants they have already paid for, and plans to close them slip: about 70% of the coal units due to retire in 2025 kept running18. New solar meets new demand first. Coal falls only when clean power grows faster than demand, and in most years it hasn’t.
Japan, Germany and China built the price curve
The curve was bought, country by country. Japan subsidised half the cost of rooftop solar from 1994 to 199615. Germany’s Renewable Energy Sources Act of 2000 then guaranteed solar owners a fixed price for their power for twenty years, and cut that price for new installations by 5% a year from 2002, pushing makers to get cheaper16.
China turned that demand into scale. By 2022 it held more than 80% of every stage of solar panel manufacturing, and its industrial policies had helped cut costs by more than 80%13.
China is building coal and solar at the same time
China keeps building coal plants for two reasons the record shows. Its rules pay coal plants for their capacity, for standing ready, not just for the power they sell. And its latest five-year plan softened the goal from “gradually phasing down” coal to “promoting a peak”18.
Paying for readiness explains a year that looks contradictory. In 2025 China commissioned about 78 gigawatts of new coal capacity (a gigawatt measures what a plant could produce at full power, not what it does produce), and between 2021 and 2025 it built seven times as much coal capacity as it retired18. Yet clean power met 94% of its growth in electricity demand, and its coal plants ran less: their average use fell from 56% of capacity to 52%18. Plants paid for standing ready can be built even while each one runs less. Global Energy Monitor, which tracks every coal plant, summed it up: “In 2025, the world built more coal and used it less.”18
Building both has a cost. In the first half of 2026 China threw away an estimated 360 TWh of wind and solar power its grid couldn’t absorb, 49% more than a year earlier: enough to cover all of its new power demand in that half-year19. Solar panels count for nothing against coal when the grid discards their power, and China’s coal generation rose 3.4% over the same months19.
Does new clean power actually push out fossil power?
Not automatically, and history says often not. A study of countries over fifty years found that each unit of electricity from non-fossil sources displaced less than a tenth of a unit from fossil fuels: most of it was simply added on top3.
That study ends before solar got cheap, and the headline estimates of what clean power has saved assume the opposite. Ember reckons fossil generation would have been 30% higher in 2025 without the wind and solar built since 20004, and the International Renewable Energy Agency (IRENA) puts the emissions avoided by all renewables that year at 8.4 billion tonnes of CO217. Both count every clean kilowatt-hour as one fossil kilowatt-hour avoided, so the real saving is probably somewhere between those figures and the historical pattern.
Where coal has fallen, solar often wasn’t the main reason. In the United States, cheaper natural gas explains 92% of the decline in coal production between 2008 and 20165, and US coal generation rose 13% in 2025 when gas got dearer4. Weather helped in 2025 too: the IEA estimates that coal demand, across all its uses, would have risen about 30% faster without a mild year8.
Coal power fell in 2025, and is rising again
Coal power is not yet declining worldwide. World coal generation grew 13% between 2015 and its record in 2024, then fell about 0.5% in 202568. Ember first put the drop at 63 TWh and has since revised it to about 4946. Solar’s rise that year was about twelve times as large as coal’s fall.
China’s coal generation fell by about 1% to 1.5%, India’s by nearly 3%, the EU’s by 5%84. In India a milder year helped: cooler weather avoided an estimated 32 TWh of demand4.
Coal demand is a different measure, and it moved the other way. Counting every use, from steel to chemicals, the world burned a record 8.84 billion tonnes of coal in 202510.
And the dip in power is already reversing. Ember’s preliminary monthly data put world coal generation about 1.4% higher in the first half of 2026 than a year earlier, by our own sum7, and the IEA now expects coal demand to grow 1.2% in 2026, to a new record, as a Middle East gas shock and El Niño push power stations back to coal10.
Demand is growing faster than solar can cover
The hardest problem is that the world keeps wanting more electricity. Demand grew 3% in 2025, and the IEA expects 3.6% in 2026 and 3.8% in 2027, driven by industry, air conditioning, electric cars, heat pumps and data centres11. In the United States, data centres alone made up about half of the growth in 20258.
Power-sector emissions barely moved in 2025; this is the total the world still emits.
What could stall the solar boom
Solar’s cost curve is flattening. The average cost of power from large solar farms stayed at $44 a megawatt-hour in 2025, and module prices had hit a floor by the end of the year17. Investment in clean-technology factories more than halved between 2023 and the end of 202517, and scale is what drove the cost curve in the first place, so slower factory building suggests slower cost cuts.
In April 2026 China also removed the tax rebate on exports of some solar products17, which is likely to make its panels dearer abroad. That matters because the supply is concentrated: China makes more than 80% of every part of a solar panel13, so any trade barrier or export change hits the whole world’s supply.
Politics can cut the other way too. After US tax credits for clean power were phased out early, and with new import restrictions, the IEA cut its forecast for US renewables by almost half12.
When solar could start cutting into coal
The IEA expects renewables to reach 37% of the world’s electricity by 202711. Whether that pushes coal down depends on the race between clean power and demand. In 2025 demand grew by 849 TWh and solar added 6364. If solar only matched that record each year while demand grew 3.6%, the world would need roughly 1,100 TWh of extra electricity a year, and solar would cover a little over half; the rest would have to come from wind, hydro, nuclear or gas before coal fell at all. That is a back-of-envelope sum, not a forecast.
Cheap panels are also reaching places the statistics barely track. In Pakistan, people and businesses installed an estimated 27 gigawatts of rooftop and small-scale solar in two years, as much capacity as every coal, gas and oil plant the country has ever built21. That power mostly replaces diesel and unreliable grid supply, not coal, but it suggests how quickly solar can spread once it is the cheapest option.
Solar has won the argument about cost for new power plants. Whether it wins the argument about coal will show in three places over the next few years: China’s coal output, how much clean power its grid wastes, and whether demand keeps outrunning supply.