The Cheapest Renewable Electricity Source: It Isn't Solar

Forty-four dollars a megawatt-hour looks too cheap until you follow a single kilowatt of solar from the mounting rail to the meter. Installing it costs $667. A 16.5% capacity factor turns the 8,760 hours in a year into 1,445 kilowatt-hours, and IRENA amortises solar over 25 years, so that kilowatt produces roughly 36,100 kilowatt-hours before the accounting stops. Spread the hardware across all of it and the panel itself costs $18.50 per megawatt-hour.

Capital is not free, though. At a 7.5% real discount rate the $667 has to come back as $59.83 a year, which over 1,445 kilowatt-hours is $41.40. Add about $2 of operations and maintenance and the total is $43.50, the published $44. The panel is therefore $18.50 of solar's $44, and financing more than doubles it. That is the same mechanism as the country-by-country spread below, arriving from the opposite direction. The divisor is not the same for every source either: IRENA amortises hydropower over 30 years and bioenergy over 20, so the seven lifetimes being compared are not identical.

Hydropower was the cheapest source on earth in 2010 at $44 per megawatt-hour and led every year through 2019. It now costs $62, up 41%, the only large source to travel that far the wrong way. Most of that is construction: the 2025 installed cost of $2,079 per kilowatt is 36% above 2010 in real terms, and civil works, meaning dams, tunnels and powerhouses, run to between half and seven tenths of a project. Concrete poured into a gorge has no factory behind it and no learning curve.

The rest is one year of weather. The capacity factor of newly commissioned plants fell from 48% to 43% in 2025, after peaking at 53% in 2023, and a metric that divides by output punishes a dry cohort. Additions actually rose 150% to 18.4 GW, and 2025 was hydropower's cheapest build year since 2018. What ran out is the easy sites, not the engineering.

Offshore wind costs $2,931 per kilowatt, three times onshore, and returns a 41% capacity factor against onshore's 36%. Five percentage points of output do not pay for three times the capital, and $78 per megawatt-hour follows from it.

The world average also hides a split that has become two industries. Offshore wind costs $49 per megawatt-hour in China and $141 in the United States, and the global figure has held flat for five years only because Chinese declines offset European and North American increases.

Bioenergy is the one technology here that buys fuel every year, and that is the whole story of its flat line. Run the same capital-only arithmetic that reconstructed solar's $44 and bioenergy comes out at $52 against a published $86, which puts roughly two fifths of its cost outside capital altogether: feedstock, handling and operation.

A manufacturing boom that cut module prices by 89% cannot reach that share, so bioenergy's cost has moved 3% in fifteen years. The feedstock is what moves it: rice husk projects average $68 per megawatt-hour against $91 for wood waste. Its small 2025 fall came from output rather than cheaper plant, as its capacity factor reached 78%.

Geothermal's 53% rise since 2010 and its 53% rise over the past twelve months are the same rise. It cost $57.8 per megawatt-hour in 2010 and $57.8 in 2024, then $88.7 in 2025. Nothing drifted; one year moved. The cause is the composition of a very small sample: only 273 MW was commissioned worldwide in 2025, a fifth below 2024, with 105 MW in Indonesia and 106 MW in the Philippines, against a 2024 cohort that was mostly 225 MW in New Zealand.

This is not a story about producing less. Geothermal runs at an 85% capacity factor, the highest of the seven, and finishes sixth anyway because $5,997 per kilowatt is nine times what solar costs to build. It is also the most finance-sensitive source here, with the cost of capital at 45% of its levelised cost against 33% for the plant itself.

Concentrated solar hit a record-low installed cost in 2025, $2,418 per kilowatt, cheaper to build than offshore wind, and its levelised cost still rose 9% to $115. The reason is entirely the denominator: its capacity factor fell to 22% because Chinese developers are deliberately building smaller mirror fields per megawatt of turbine, so the plants work as flexible parts of hybrid wind and solar projects rather than as round-the-clock generators.

Every megawatt commissioned anywhere in 2025 was Chinese, 900 MW across nine projects. Chinese rules treat it as the add-on that lets a wind and solar project meet capacity and storage obligations without a central subsidy, so last place is partly a choice about what the plant is for.

Sort the seven by capital committed per kilowatt-hour of annual output instead of by cost and the ranking's logic surfaces. Onshore wind commits $0.31, solar PV $0.46, bioenergy $0.53, hydropower $0.55, geothermal $0.81, offshore wind $0.82 and concentrated solar $1.26. The middle reshuffles: bioenergy and geothermal look better on capital per unit of output than on levelised cost, and everything between the two views is fuel, maintenance and the cost of money.

Capacity factor also changes what the ranking means physically. Covering one American household's 10,380 kilowatt-hours a year takes roughly 3.3 kilowatts of onshore wind, 7.2 kilowatts of solar or 1.4 kilowatts of geothermal.

Onshore wind costs $27 per megawatt-hour in China, $40 in the United States, $52 in Germany and $109 in Japan, and solar runs from $35 in India to $107 in Poland. Almost none of that spread is equipment, which trades as a global commodity at one price. IRENA attributes about 56% of the variation in financing costs to country-level macroeconomic conditions against 24% to the technology.

The real cost of capital for a standardised, fully contracted project runs from about 3% in deep markets such as the United States and Australia to above 10% in the highest-risk economies, and that move alone raises solar's cost by about 92% on the FactsFigs arithmetic set out above. So the contract is the cheapest lever a government holds: a merchant project carries roughly double the weighted cost of capital of a contracted one, and analysis quoted by IRENA finds long-term contracts can cut levelised cost by 30% to 40% through financing alone. For emerging economies IRENA says it flatly, that the cost of capital rather than the cost of technology is the binding constraint.

The world average is substantially a measurement of Chinese construction. Take China out and onshore wind rises from $33 to $51, solar from $44 to $55 and hydropower from $62 to $70. China put up 315 GW of the 511 GW of solar built in 2025 and about 120 GW of the 159 GW of wind, so a developer anywhere else should read the headline figures as a floor rather than a quote. Volumes are uneven between the seven too: hydropower added 18.4 GW, bioenergy 3.4 GW, concentrated solar 900 MW and geothermal 273 MW.

The savings land unevenly as well. IRENA estimates renewables displaced about $480 billion of fossil fuel purchases worldwide in 2025 along with 8.4 gigatonnes of carbon dioxide, $177 billion of it in China. For Indonesia, Thailand and the Philippines the existing fleet displaced $5.7 billion of coal and gas, a saving that rises exactly when a fuel importer can least afford a price spike and needs no new investment, because the plant is already standing.

The first limit is the field itself. IRENA costs renewables and nothing else, so nothing above settles anything about coal, gas or nuclear. IRENA does offer a comparator in prose: a turbine shortage driven partly by data-centre demand roughly doubled the capital cost of a new combined-cycle gas plant in the United States, to about $2,400 per kilowatt, pushing gas toward $100 per megawatt-hour in Italy, Germany and Japan while it held at $50 to $60 where gas is cheap. More than 90% of utility-scale renewables commissioned in 2025 came in below the cheapest new fossil alternative in their market, the second year running.

A Fossil Comparison Needs Another Source · Lazard v19.0

For an unsubsidised like-for-like against fossil and nuclear plant, Lazard's LCOE+ v19.0 of July 2026 is the citable set, and it is a separate source, a separate method and a separate geography that should not be merged into the figures above. Its unsubsidised United States ranges put combined-cycle gas at $51 to $129 and nuclear at $175 to $255 against utility solar at $40 to $98. Lazard sits above IRENA largely because it assumes 60% debt at 8% and 40% equity at 12%.

It Is Not Your Bill · 3.3 to 11.5 Cents

These seven figures are 3.3 to 11.5 US cents per kilowatt-hour at the plant gate. The average American household paid 16.5 cents a kilowatt-hour in 2024 on about 865 kilowatt-hours a month, and the distance between the two covers transmission, distribution, taxes, retail margin and the existing fleet. IRENA also excludes subsidies, system balancing, carbon pricing and air-pollution costs, and calls these data points prices rather than costs.

Two caveats close it. Variable output has to be covered, and the cover is priced: Lazard's firming charge adds roughly $8 to $30 per megawatt-hour to wind and solar. And seven is not the whole set: hydropower's 30-year basis is deliberately conservative against plants that run 30 to 80 years, and IRENA separately tracks wave and tidal stream energy, left out of the ranking because they are pre-commercial.

Onshore wind, at $33 per megawatt-hour, and it wins on output rather than on price. Solar fell furthest, by 89% in fifteen years, and still finishes second because its capacity factor is less than half of wind's. The two sources that now cost more than they did in 2010 got there separately: hydropower through concrete that cannot be mass-produced and a dry commissioning year, geothermal through 273 megawatts of drilling in two countries.