Solar Beat Nuclear in 2025. What Energy Buyers Should Do Next

What the milestone actually measures

The headline comes from the World Nuclear Industry Status Report 2026, and pv magazine reported that solar generation surpassed nuclear generation globally for the first time in 2025. The word that matters is generation. This is about kilowatt-hours delivered, not about how many gigawatts of panels are installed.

According to the WNISR's own summary, 2025 was the first full year in which solar out-produced reactors worldwide. Solar generation rose about 30% that year, while output from operating reactors rose just 1.1%. The monthly crossover came earlier: wind passed nuclear in March 2025 and solar followed in April 2025.

The pattern has held since. From March 2025 through June 2026, either solar or wind individually generated more electricity than nuclear every month, and in March, April and May 2026 both did so independently. That is a sustained trend, not one sunny month.

The capacity figures and how firm they are

On capacity, the report puts global solar at 2,383 GW after 510 GW was added in 2025, up 27% year on year. Wind added a record 159 GW to reach 1,291 GW. Installed nuclear capacity rose just 1.5 GW to 399.5 GW, while operating nuclear capacity, the reactors actually running, fell 0.5 GW to 368.6 GW.

pv magazine describes WNISR's PV figures as conservative. IEA PVPS data put 2025 global PV additions at around 690 GW, about 180 GW above the figure WNISR used, and pv magazine says this suggests the gap between solar and nuclear is even wider. That comparison concerns capacity additions. It does not by itself tell us that the report's generation figures are wrong, and the size of the 2025 generation margin is not something this post can state.

On the nuclear side, the growth that exists is concentrated. Nuclear generation in China rose 7.6% in 2025, while outside China it fell 0.2% and remains well below its 2006 level. The WNISR site puts nuclear at 8.8% of global electricity production in 2025, with 413 reactors operating and a mean fleet age of 32.9 years.

Gigawatts are not kilowatt-hours

Solar needed roughly six times nuclear's installed capacity, and more than six times its operating capacity, to pass it in output. The reason is capacity factor. An International Solar Energy Society column gives typical figures of around 74% for nuclear, 40 to 60% for combined-cycle gas, 30 to 60% for wind and 12 to 25% for solar PV.

The same column, written in May 2024, forecast that solar would pass nuclear generation in 2026, wind in 2027, hydro in 2028, gas in 2030 and coal in 2032. On a full-year basis, solar arrived a year earlier than that forecast. If your energy plan rests on cautious renewable projections, it is worth revisiting.

Storage is the variable that changes solar's daily shape. A daily.dev summary of the pv magazine article reports that 104 GW / 257 GWh of grid-scale battery storage was added in 2025; treat that figure as reported via a summary rather than read from the primary report. The ratio implies a few hours of storage per unit of power on average. That can move midday output into the evening. It does not carry a plant through a cloudy week in winter.

A worked example for a factory roof

The capacity-factor range becomes useful once you apply it to your own site. The numbers below are illustrative arithmetic using only the typical 12 to 25% range cited above, not a yield estimate for any real location. A proper resource assessment by an engineer will narrow the range for your roof, orientation and climate.

Take 100 kW of rooftop PV. Running at full power for all 8,760 hours of a year would produce 876 MWh, which never happens. At a 12% capacity factor it yields about 105 MWh a year; at 25%, about 219 MWh. Now imagine a hypothetical factory consuming 1,000 MWh a year. That 100 kW system would cover roughly 10 to 22% of annual consumption, and for a plant using 2,000 MWh, roughly 5 to 11%.

Two lessons follow. First, rooftop solar on an energy-intensive site usually trims the bill rather than replacing grid supply, so set expectations accordingly. Second, annual totals hide timing. If the plant is closed on sunny weekends or runs mostly night shifts, a large share of that output does not line up with consumption, and what happens to it depends on local rules you need to check.

• 100 kW × 8,760 h = 876 MWh theoretical maximum per year
• At 12%: about 105 MWh per year
• At 25%: about 219 MWh per year
• Hypothetical 1,000 MWh/year factory: about 10 to 22% covered
• Hypothetical 2,000 MWh/year factory: about 5 to 11% covered

What it means for energy costs and procurement

For buyers, the more useful finding is not the crossover itself. A pv magazine and ISES analysis found that globally, in 2025, new solar and wind supplied nearly all of the additional electricity needed to meet demand growth, including electrification of transport, heating and industry, while net nuclear growth was roughly zero. That is a global aggregate, and individual grids will differ.

For a single company, the practical question is timing. If your tariff is time-of-use and midday prices are lower, shifting load into those hours captures value without buying a panel. That is a condition to check on your own tariff and market, not a trend this post can report. Compressors, furnace pre-heating, battery charging and chilling are typical candidates for shifting where production allows it.

None of this guarantees a lower bill. Grid fees, taxes, the cost of supply in hours solar does not cover, and local market design all sit between a global generation statistic and your invoice. The research behind this post did not include verified 2026 corporate PPA prices for Europe or Türkiye, so ask for a dated source when anyone quotes you a market price.

On-site solar or a PPA: how to decide

The two routes solve different problems. On-site solar reduces what you draw from the grid during daylight and gives you an asset you control. A power purchase agreement lets you contract renewable output from a larger plant elsewhere, which suits companies with limited roof space, leased buildings, or loads far bigger than any roof can serve, as the worked example above suggests.

Size to the consumption curve, not the roof. Filling every square metre with panels makes sense only if your demand absorbs the output when it is produced, or if the rules for excess generation make it worthwhile. Answer the questions below before talking to an installer or developer.

On-site solar is not for everyone. With a short remaining lease, a weak roof, or mostly night-time operation, the economics can be poor. A PPA is not automatically simpler either: contract length, price indexation and the handling of underperformance all need careful legal reading.

• What does your hourly load curve look like across a full year, not just an average month?
• How much of your consumption falls in daylight hours, including weekends and holidays?
• Do you own the roof and the building, and how long will you stay?
• Can the roof carry panels structurally, and when does it need re-roofing?
• Would a modest battery move enough midday output into your evening peak to matter?
• For a PPA, who carries the risk if contracted output falls short, and how is the price indexed?

AI, data centres and the reliability question

The capacity-factor contrast frames this debate: around 74% for nuclear against 12 to 25% for solar PV. A data centre needs steady power around the clock, while solar delivers a daily curve. Solar can therefore serve such a load only alongside storage, grid supply or other firm generation.

Nuclear's high capacity factor looks attractive on paper, but delivery time matters. The daily.dev summary of the pv magazine article reports that none of the 16 reactors connected in 2023 to 2025 was finished on schedule, with construction averaging 10.9 years. Again, that figure comes via a summary rather than the primary report. The research for this post did not include verified data on how data centres are actually being supplied, so claims in either direction deserve scrutiny.

For most businesses the relevant point is indirect. If you rely on cloud and AI services, your provider's power sourcing affects the reliability of those services and potentially the supply-chain emissions you report. Asking providers how they source and firm their power is a reasonable procurement question.

What we could not verify, and what to do anyway

This post does not cover current Turkish rules for commercial rooftop solar, self-consumption or unlicensed generation, nor the current scope and timelines of EU CBAM and sustainability reporting rules. Verify connection limits, approval steps and the treatment of excess output with your distribution company and a qualified engineering firm, and check EU requirements against official guidance or an advisor.

The steps below hold regardless of how those rules read today. They are the same for a manufacturer in Izmir and an SME anywhere else: measure first, then size, then contract.

• Sub-meter your biggest loads so you know where electricity actually goes.
• Map hourly consumption against daylight hours before sizing any system.
• Run the capacity-factor arithmetic on your own consumption to set realistic expectations.
• Shift flexible, energy-heavy processes only where your tariff rewards it.
• Get a structural roof survey and a current regulatory check before requesting quotes.
• Keep invoices, meter readings and contracts organised so they can support any emissions questions from customers.