Electricity Demand Peaks: When Grids Are Most Stressed

Electricity demand peaks when large numbers of people need heating, cooling and lighting at the same time, which usually means early evening on a weekday and, over the year, the hottest summer afternoons or the coldest winter mornings and evenings. Grids are most stressed when such a peak coincides with low wind or solar output, plant outages or limited imports, because that is when the margin between available supply and demand is thinnest.

This guide explains the daily and seasonal shape of demand, why extreme weather matters so much, and how grid operators prepare. It uses the operators’ and regulators’ own figures and, for the most recent days, our own data.

The daily shape: why the evening matters

On most grids demand is lowest in the early hours of the morning, rises as people wake and businesses open, and reaches its daily high in the late afternoon or early evening. In winter the evening peak is driven by lighting, cooking and heating; in summer it is driven by air conditioning that keeps running after the hottest part of the day.

Solar power has made the evening more important. Rooftop panels reduce the demand that the grid itself has to supply during the day, and AEMO notes that strong distributed solar output lowers operational demand in the middle of the day. When the sun sets, that support disappears just as households switch on appliances. Grid-scale solar falls away at the same time, so the hour after sunset can be the most demanding one even if it is not the hour of highest total consumption.

ERCOT’s all-time record shows the pattern. EIA reports that hourly load reached 91.1 GW on 22 July 2026, 6% above the previous peak of 85.5 GW set on 10 August 2023, and that the record peak occurred at 6:00 p.m. CT during a heat wave. At that hour, natural gas supplied 48% and solar 32% of ERCOT’s generation.

Seasonal peaks: heat waves and cold snaps

FERC staff put it simply in their 2025 State of the Markets report: “Peak load is typically caused by weather.” Whether a grid peaks in summer or winter depends on its climate and on how buildings are heated.

Summer peaks

The large U.S. grids on our pages all set their records in summer, and in 2026 several of them set new ones, according to EIA and the operators:

Grid operator	Peak hourly demand	Date	Context
PJM	162.6 GW	2 July 2026	1.3% above the pre-2026 peak of 160.6 GW (23 June 2025)
MISO	121.5 GW	30 June 2026	Below MISO’s 127.1 GW record of 20 July 2011
ERCOT	91.1 GW	22 July 2026	6% above the previous peak of 85.5 GW (10 August 2023)
SPP	58,181 MW	27 July 2026	SPP’s coincident peak; its winter peak was 48,142 MW (20 February 2025)
ISO New England	28,130 MW	2 August 2006	Summer record; winter record 22,818 MW (15 January 2004)

New England’s records date from 2004 and 2006 even though the region keeps growing. ISO New England counts about 400,000 distributed solar installations totalling about 7,500 MW, which reduce the demand its grid sees on sunny summer afternoons.

Demand growth makes these records more likely. NERC’s 2026 Summer Reliability Assessment found that load growth has increased by 11 GW since 2025, on top of 10 GW of growth in 2025, and pointed to early summer heat and drought as risks. It also found that fewer areas were at elevated risk of shortfalls in abnormal summer conditions: three regions and one locality in 2026, against six regions in 2025.

Winter peaks and cold snaps

For SPP and ISO New England the winter records are lower than the summer ones, but cold snaps can be more dangerous for the grid. Freezing weather raises heating demand while also knocking out generating units and the gas supply that feeds them.

Winter Storm Elliott in December 2022 is the clearest recent example. The joint FERC, NERC and Regional Entity staff report found that at the worst point of the event there were 90,500 MW of coincident unplanned generating unit outages, derates and failures to start in the Eastern Interconnection. Grid operators in the southeastern U.S. had to shed more than 5,400 MW of firm load in total, the largest controlled load shed recorded in the Eastern Interconnection. The same report notes that Winter Storm Uri in February 2021 led to the largest controlled firm load shed in U.S. history, more than 20,000 MW, and that more than 4.5 million people in Texas lost power.

How operators prepare for the highest hours

Reserve margins

A reserve margin is the amount of available capacity above expected peak demand, kept to cover forecast errors, outages and extreme weather. Operators and regulators set a target margin and check whether enough supply has been procured. PJM’s capacity auction for the 2027/2028 delivery year illustrates the challenge: PJM said committed supply was short of its reliability requirement by 6,623 MW, with a reserve margin of 14.8% against a 20% standard, and that nearly 5,100 MW of the increase in forecast peak load was data center demand.

Capacity markets and seasonal assessments

Four U.S. operators, PJM, MISO, NYISO and ISO-NE, rely on centralized capacity markets to secure enough resources for the peak, according to FERC. ERCOT has no centralized capacity market, so it relies on energy prices to reward plants that are available when demand is highest. Ahead of each summer and winter, NERC publishes reliability assessments that flag regions at risk, and operators run their own seasonal studies.

Operating the day itself

On a forecast peak day, operators schedule extra generation and reserves, delay planned maintenance where they can, coordinate imports with neighbours and call on demand response, where large customers agree to cut use for pay. If supply still runs short, they can issue conservation appeals and move through formal emergency levels. Controlled load shedding, short rotating outages, is the last step, used to prevent an uncontrolled collapse of the grid.

What our data shows in the days we have tracked so far

Our world electricity clock shows current demand for each market we cover. Our data starts on 23 October 2026, an autumn weekend in the north and spring in Australia, so none of these figures is a seasonal peak. They show the daily shape:

Market (our page)	Highest demand we hold	When (local time)	Lowest demand we hold	When (local time)
Great Britain	34,004 MW	Fri 23 Oct, 18:00 BST	17,725 MW	Sun 25 Oct, 04:30 GMT
New South Wales	9,199 MW	Sun 25 Oct, 17:40 AEDT	3,957 MW	Sat 24 Oct, 12:55 AEDT
Queensland	7,579 MW	Sun 25 Oct, 17:55 AEST	5,024 MW	Sun 25 Oct, 12:35 AEST
South Australia	1,419 MW	Sat 24 Oct, 20:10 ACDT	201 MW	Sun 25 Oct, 16:05 ACDT
ERCOT	58,813 MW	Sat 24 Oct, 19:00 CDT	44,116 MW	Sat 24 Oct, 04:00 CDT

The windows differ by market: Great Britain from 17:00 BST on 23 October, the NEM regions from the early morning of 24 October local time, and ERCOT from 20:00 CDT on 23 October, each up to the morning of 25 October 2026 (Australian evening). The ERCOT row uses EIA’s preliminary hourly data and has no data for 14:00–18:59 CDT on 24 October; all values are as we queried them on 25 October 2026 and may be revised.

Even in a few days the evening pattern is visible. Britain’s highest half-hours on both Friday and Saturday fell at 18:00 BST. In New South Wales and Queensland the highest five-minute demand came just before 18:00 on Sunday, while the lowest fell around midday, when rooftop solar covers much of household use. South Australia is the extreme case: grid demand fell to about 201 MW on Sunday afternoon, a seventh of its Saturday evening high. Market pages for PJM, MISO, SPP and ISO New England show the same hourly EIA data for those grids.

Why it matters

Peaks set the size of the system. Power plants, transmission lines and reserves are built for the highest hours, not the average one, which is why shifting even a little demand away from the evening peak has value. Our figures are provisional, come from the operators’ own publications and are not for operational decisions.