Baseload vs. baseload

The claim comes up constantly: solar provides zero baseload capacity. It must be backed up with coal, natural gas, nuclear, or hydro. Without those, the lights go out.

Worth looking at more carefully. Because "baseload" is doing two different jobs in that sentence, and the argument depends on nobody noticing.

Two meanings, one word

Baseload demand is real. Every grid has a floor — some irreducible minimum that shifts with the season, the time of day, and how much the heating or cooling is running, but never drops to zero. Even at 3am on a mild Tuesday, hospitals run, data centres hum, and lights stay on. That part's not in dispute.

To service demand, a grid has a portfolio of sources — some running continuously, some ramping with the load, some held in reserve for peaks. Each contributes a share; the mix covers the load. It's not unlike an orchestra: strings carry certain parts, brass others, woodwind others still. No section is expected to play every note; the performance works because the ensemble does.

What came to be called "baseload plants" are a specific component of that portfolio. Coal and nuclear run flat-out, continuously, because they're slow and expensive to ramp up and down. That operating characteristic got labelled "baseload generation," and somewhere along the way the label started doing extra work: if there's baseload demand, the argument goes, it must be met by baseload plants.

That's not an argument. It's a definition doing the work of an argument. "We need baseload. Baseload means coal. Therefore we need coal." The conclusion is already in the premise. Conversely, "intermittent" gets applied to solar and wind as if the label alone settles the argument. It doesn't.

How grids actually plan

Grid operators don't think in terms of "baseload plants" anymore. The planning metric is ELCC — effective load carrying capability — which is simply: how much does this resource contribute when the grid needs it most? It's a number, expressed as a percentage of capacity, and every resource gets one. The grid's job is for the portfolio to cover the load in every hour, with each source contributing its measured share.

Solar's ELCC isn't zero, and pairing it with batteries raises it further — storage makes solar dispatchable, in the sense grid operators care about: available when called upon. Wind's ELCC isn't zero either, and the two tend to complement each other, producing more combined than either does alone. Criticising solar for not covering the whole load is a bit like expecting the strings section to play every part — the wind section has a contribution to make too.

Everything needs backup

The "must be backed up" claim carries an implicit assumption: that coal, gas, and nuclear simply run — on, all the time, 24/7/365, no exceptions. They don't. Everything needs backup. Nuclear plants trip offline unexpectedly, and go down for weeks during refuelling. Coal plants need scheduled maintenance. Gas plants fail. Grid operators carry reserve margins for all of them, regardless of the mix. Variable renewables add a specific, well-studied integration challenge — and that's real — but it's one cost item among many.

Renewables need backup in a different way from thermal plants — more routine, but also more predictable. The challenge is variable output: solar doesn't generate at night, wind drops in still weather. Sunset isn't a surprise; wind patterns are forecastable days in advance. An unscheduled coal or nuclear outage isn't. The solution to variable output is storage — capture surplus generation when it's available, dispatch it when it isn't.

Storage has taken several forms over the years. The technology dominating the conversation now — and increasingly the grid itself — is the lithium-ion battery. Its success over the past decade has been extraordinary: costs have fallen by over 90%, transforming grid-scale battery storage from a niche proposition into a mainstream one.

Batteries can't cover five-day droughts — and they don't need to

Lithium-ion batteries are designed for daily charge-discharge cycles: shifting midday solar into the evening peak, smoothing short-term frequency variations. They do that job well. There's a real limitation, though: scaling lithium-ion to cover five consecutive days of low wind and weak sun (a phenomenon grid planners call dunkelflaute) would be, at best, a serious engineering and economic challenge.

But "batteries" and "storage" aren't the same thing, and lithium-ion is one chemistry among many. Pumped hydro has handled days-long balancing for over a century — China added 18 GW of new pumped hydro capacity in 2025 alone, bringing its national total over 70 GW. A growing range of newer technologies is moving from development into commercial production: iron-air batteries, built from iron and water, designed for 100-hour durations; sodium-ion batteries, a lithium-free alternative already in commercial manufacture; compressed air storage, using underground geological formations; vanadium flow batteries; and thermal storage — each suited to different durations and use cases.

It's worth being honest about where things stand: long-duration storage currently costs more than lithium-ion, and many technologies beyond pumped hydro are at early commercial scale. It's a genuine engineering challenge, not a solved problem. But it's a different challenge than "we need more lithium-ion batteries" — and the grids already managing multi-day events are mostly reaching for pumped hydro first, not waiting for lithium-ion to do a job it was never designed for.

The price claim doesn't hold where renewables are strongest

A common corollary: electricity costs are highest where renewables are most concentrated. The correlation is real in some places; the causation isn't what it looks like.

Texas — the largest electricity market in the US — generates over a third of its electricity from wind and solar, and its residential electricity prices run around 9% below the national average. In California, the utility PG&E has certainly increased rates over the past several years — but those increases are mostly explained by transmission and distribution spending for wildfire mitigation, not the generation mix. The "renewables cause high prices" argument doesn't survive being tested in the places where renewables are actually dominant.

The experiment has already been run

Denmark now generates over 90% of its electricity from renewables — mostly wind. For a sense of how far that is to travel: Denmark used to get 90% of its energy from oil. The 1970s oil crisis prompted the country to pioneer wind power, and fifty years later it's running an advanced industrial economy almost entirely on it. Reliably.

Uruguay isn't a country that comes up much in energy debates, which is what makes it useful. Starting from almost nothing in 2010, it built wind power from 1% of its electricity supply to about a third in under a decade — without being wealthy, without being a technology superpower. Electricity production costs dropped by nearly half in the process.

Australia makes the point another way: rooftop solar alone — panels on houses and businesses, not utility-scale farms — now stands at 28.3 GW of installed capacity. That's larger than the country's entire coal-fired generator fleet. The speed of the shift is as striking as the scale. In 2010, renewables provided around 10% of Australia's electricity; coal and gas together were close to 70% as recently as 2021. By the final quarter of 2025, renewables and storage were supplying over half of the National Electricity Market for the first time, while coal-fired generation hit an all-time quarterly low. Australia was one of the most coal-dependent grids in the developed world. That's no longer true.

The money has already voted

Solar is now the single largest investment category in global energy — bigger than oil production. Clean energy overall attracts twice the capital of fossil fuels. A decade ago, fossil fuel investment was 30% higher than electricity investment. Now electricity investment runs 50% higher than all fossil fuels combined.

The "baseload" framing was always a description of how grids worked with the technology available at the time. It was never a law of physics. The technology has moved on.

Researched and drafted in collaboration with Claude (Anthropic).