Solar won the price war. The battery is winning the second one

The bulk-energy price war is over and has been for years: on Lazard's 2025 numbers, utility-scale solar runs $38-78 per megawatt-hour and onshore wind $37-86, the cheapest new electricity in most of the world before subsidies¹. The build-out follows the price: the world added a record 647 gigawatts of solar in 2025², of which China alone installed 315, pushing its cumulative fleet past 1.2 terawatts and tipping its installed capacity majority non-fossil for the first time³.

The serious argument therefore moved house years ago. It lives in the hours the sun and wind miss: the January windless week, the evening ramp, the firming cost of guaranteeing a megawatt at 9pm from a resource that peaked at noon. That argument was winning against renewables in 2022 and 2023, when storage prices rose with the battery-materials squeeze and grid operators repriced capacity accreditation downward¹. Then the battery cost curve did what the solar cost curve had done a decade earlier.

The year storage broke the gas record

In 2025 the world added about 108 gigawatts of battery storage, up around 40% on 2024, and the IEA flagged the milestone hiding in that number: it exceeds the largest year gas-fired capacity ever managed, roughly 107 gigawatts back in 2002⁴. In energy terms deployment reached an estimated 250 gigawatt-hours, up 46%, while battery costs fell a further 45% in the single year⁴, driven by lithium-iron-phosphate oversupply out of a Chinese industry whose EV demand grew slower than its factories. About 80% of the new capacity was utility-scale; China took roughly 60% of global additions at 167 gigawatt-hours, with the United States closing its record year at 57.6 gigawatt-hours of new grid storage⁴⁵.

What the batteries do first is unglamorous and decisive: they eat the evening ramp. Four hours of storage moves solar's noon surplus into the dinnertime peak, which is where scarcity pricing lived and where gas peakers earned their keep. In California and Texas, the pioneering grids, batteries now set the evening price on many days and have flattened the duck curve their critics made famous. 'Anytime solar', Ember calls the pairing⁶, and at current module and cell prices, solar-plus-storage undercuts new fossil capacity for daily-cycle service across most of the sunbelt where most of humanity lives.

The demand side sharpens the timing. Electricity consumption in the rich world is growing again after two flat decades, with data centres alone over a fifth of advanced-economy demand growth to 2030 on the IEA's count, and the electrification of transport and heat stacking on top. Every incremental terawatt-hour has to come from somewhere, and the marginal economics now favour the pairing this piece describes: the question utilities actually face is not solar versus gas in the abstract but which resource can be contracted, permitted, and energised before the load arrives. Speed has become a cost category, and modular hardware delivered by the container is the fastest thing on the menu.

What firming still costs

The remaining problem should be stated plainly: four hours is not four days. Batteries solve the daily cycle brilliantly and the seasonal one not at all, and the deeper renewable penetration goes, the more the residual problem concentrates into rare, long, expensive events, the still winter fortnight, the cloudy monsoon week. Grid operators have responded by refining capacity accreditation, counting each successive gigawatt of solar or storage for less firm value than the one before, which is why Lazard reports firming costs generally rising even as hardware falls¹. The last 10% of reliability is where the cost lives, and every serious system plan prices some mix of long-duration storage, transmission, demand flexibility, and firm low-carbon capacity, with the nuclear revival one obvious claimant, to cover it.

The costs of the pieces keep moving, though, and in one direction. Lazard's 2025 storage supplement found US storage costs falling enough to erase the pandemic-era increases⁷; cell oversupply has cut contract prices below what 2022's models assumed for 2030. Firming is not free, and it is repricing downward on the same manufacturing curve that made the panels cheap. The argument that renewables' system costs would swamp their generation savings depended on storage staying dear. It did not.

Wind belongs in this account too, even if the title gives solar the victory. Onshore wind remains the cheapest resource in the windy latitudes and pairs with solar better than either pairs with itself, their outputs peaking in different seasons and hours; offshore wind, after a brutal repricing through 2023-24 when inflation broke its contract models, is rebuilding its pipeline at realistic prices. The energy transition's workhorse portfolio, solar, wind, batteries, wires, is boring by design: four commodities, each on a manufacturing curve, each substitutable at the margin for the others. Systems built from commodities converge on cost; systems built from projects converge on schedule slip. That, more than any single price point, is why the buildout keeps surprising forecasters to the upside.

The geography of the second war

Like the first price war, the second is being won in Chinese factories and fought everywhere else. China's 60% share of storage deployment mirrors its dominance of cell manufacturing, and the battery that firms a Texan solar farm was overwhelmingly likely to have been made by CATL, BYD, or their neighbours. Tariff walls change who pays the manufacturing margin, not where the manufacturing sits; the US record year was installed substantially from imported cells bought ahead of tariff deadlines. Energy security arguments for renewables are true at the fuel level, no cartel meters the sun, and incomplete at the hardware level, where one country makes most of the kit.

The sunbelt is where the pairing bites hardest and fastest. For sunbelt countries with big solar ambitions, Morocco among them, the 45% battery repricing changes the national arithmetic: a grid that couldn't afford firm renewables at 2022 storage prices can at 2025 prices, and the import bill shifts from fuel, recurring forever, to hardware, paid once a decade. That trade, operating expense for capital expense, is the fiscal shape of the energy transition in the global south, and it just got meaningfully better.

What the sceptics got right, and when

The firming critique was correct on the facts of its moment. In 2015, storage at scale did not exist; grids genuinely needed the fossil fleet for every evening; and early renewable subsidies paid for energy while the system quietly supplied the reliability free. The critique became wrong the way most correct analyses do: by holding still while its subject moved. The error was never the arithmetic; it was treating a manufacturing cost curve as a constant. Anyone still quoting 2015's system-cost studies is arguing with a grid that no longer exists, and anyone assuming 2025's battery price is final is making the same mistake in the other direction.

The residual truth the sceptics keep is seasonal: no plausible battery price firms a dark, still fortnight from lithium alone, and systems at very high penetration will pay real money, in long-duration storage, overbuild, transmission, or firm capacity, for the last reliability percentiles. The policy craft is to buy that insurance without letting its price be used as an argument against the cheap 90% that needs no insurance at all. Grids that keep those two accounts separate are decarbonising fastest and cheapest; grids that blend them are still holding conferences about system costs.

What to watch

Three gauges for whether the second war ends like the first. The 2026 battery price, because a 45% annual fall will not repeat forever and the trajectory sets every system plan's storage line⁴. The first commercial long-duration deployments at scale, iron-air, thermal, compressed gas, because the seasonal residual is the one problem the lithium curve does not solve. And the capacity accreditation rules in the big markets, because they are where the firming argument now actually happens: not in opinion pages, but in the committees that decide how many megawatts a battery is worth at 7pm in January. The first price war took fifteen years from expensive curiosity to cheapest source on earth. The second is running the same course faster: it has already passed the 2002 gas record, and in the pioneering grids it has already taken the evening peak.