Originally published May 27, 2015. Updated July 17, 2026.
A well-run grid can handle 60% to 80% wind and solar, according to the National Renewable Energy Laboratory’s own modeling of a reliable U.S. grid in 2035, far above the combined 14.1% wind and solar supplied in 2023. Whether your grid gets there depends on where the extra power goes once every turbine and panel produces at the same time. Grid operators call that midday glut, and the evening scramble after it, the duck curve.
Jesse Jenkins’ 2015 case for a ceiling
Wind and solar running into a ceiling before they run out of room isn’t a new worry. In a two-part piece published on Energy Collective on May 27, 2015, then-MIT researcher Jesse Jenkins argued that variable renewables hit an economic limit tied to a source’s capacity factor, the share of hours it can run at full output. Wind and solar cost almost nothing to run once built, so they get dispatched first, and Jenkins argued that pushes wholesale prices down hardest in the exact hours wind and solar generate the most. He called this the merit order effect, or value deflation. Jenkins put the ceiling for wind at roughly 25% to 35% of electricity supply, and for solar at 10% to 20% in most regions. Combined, he estimated variable renewables topping out around one-third to one-half of total electricity. He pointed to Ireland, where the grid operator was already capping instantaneous wind and solar output at about 50% of demand, with research at the time suggesting a ceiling near 55% to 60% of supply without a major jump in storage or grid flexibility.
What the duck curve looks like on a real grid
The duck curve is what that ceiling looks like on a chart. Plot how much power a grid still needs from other sources after wind and solar output is subtracted, and the line sags in the middle of the day, when solar output peaks, then climbs sharply in early evening as the sun goes down. Charted over 24 hours, the shape resembles a duck, with a saggy belly at midday and a rising neck at dusk. California’s grid operator, CAISO, coined the term in 2012 for this exact problem, a burst of cheap solar power arriving all at once. It’s more power than the region needs at that moment, and other generation has to come online fast once the sun sets. The dip is the same problem Jenkins described in 2015. Value falls as more of one source shows up in the same hour. Batteries are the main fix on the ground today. In California, batteries shifted about 6 gigawatt-hours of solar power a day from midday into the evening peak in 2022. That narrows the curve instead of wasting the power or shutting panels off.
Batteries are pushing the ceiling higher
Storage is the biggest thing that’s changed since Jenkins wrote in 2015. In Texas, the ERCOT grid’s battery storage capacity grew from about 2 gigawatts in 2022 to a projected 11 gigawatts by 2025. That gives operators somewhere to put a surplus instead of shutting it off. That grid’s wind farms alone generated 108,000 gigawatt-hours of electricity in 2023, on a system that also posted an all-time peak demand of 85,931 megawatts in August 2024.
Grids already testing the upper end
A handful of real grids are already close to, or briefly past, the numbers Jenkins was estimating in 2015. South Australia ran entirely on wind and solar for 6.5 consecutive days in December 2021, according to the state’s grid operator, and its rooftop and grid-scale solar together exceeded 100% of state demand for the first time in October 2020. Wind alone supplied 47% of South Australia’s electricity in 2022 and 2023, well past Jenkins’ top estimate for wind of 35%. Denmark’s wind turbines alone supplied 57.91% of the country’s domestic electricity in 2024, more than 20 percentage points above that same estimate. South Australia leans on the Hornsdale Power Reserve, a 150-megawatt, 194-megawatt-hour battery bank that was the world’s largest lithium-ion battery when it was built, to smooth out the gap between the windy, sunny hours and the rest of the day. Neither grid is anything like the size of the continental United States, and running on wind and solar for a stretch of days or a single year isn’t the same as running that way permanently. Still, the ceiling Jenkins described in 2015 has moved well past where he placed it, at least on grids small and flexible enough to test it.
What the math says for the whole U.S. grid
Scaling this up to the entire U.S. grid is a different problem than any one region proving it’s possible. The National Renewable Energy Laboratory modeled a reliable U.S. grid running on 60% to 80% wind and solar by 2035, four to six times the combined 14.1% those two sources supplied in 2023. Wind supplies most of that current total, 10.2% of U.S. generation in 2023, with solar contributing the remaining 3.9%. Including hydropower and the rest, total U.S. renewable generation came to 21.4% in 2023. In NREL’s modeling, getting to its target takes 120 to 350 gigawatts of new battery storage able to discharge for 2 to 12 hours at a stretch, plus another 100 to 680 gigawatts of longer-duration storage once wind and solar pass 80% to 95% of generation. NREL also put a price on the transition, $330 billion to $740 billion in added power-system costs against $920 billion to $1.2 trillion in net benefits to society through 2035. Those figures are the size of the gap between where Jenkins drew the line in 2015 and where the country’s own energy modelers think the grid can go.
What would change the ceiling next
The ceiling keeps moving because the underlying numbers keep moving, and a few of them are still unsettled. If you want to know whether it holds where you live, watch battery and transmission buildout more than any political announcement about a renewable target. NREL’s own scenario assumes storage capacity that doesn’t exist on the grid yet, built at a pace nobody has proven at that scale. The jump from 2 gigawatts to 11 gigawatts in ERCOT’s battery fleet is on pace to take about three years. If that pace holds nationally, and enough transmission gets built to move power between regions with different wind and solar patterns, the share of the grid running on wind and solar climbs well past where Jenkins placed the ceiling in 2015. If storage or transmission stall instead, the country ends up closer to his 2015 estimate than to NREL’s 2035 one.
