China’s carbon dioxide emissions fell 3.1 percent in the third quarter of 2026 from a year earlier, the steepest quarterly decline since the Covid lockdowns of 2022, according to an analysis published Nov. 6 in Carbon Brief by Lauri Myllyvirta, lead analyst at the Centre for Research on Energy and Clean Air, a think tank based in Helsinki.
That came three months after another first. Between January and June, coal produced 49.7 percent of China’s electricity, the first time its share had dropped below half, according to the National Energy Administration. Xiaoying You set out the milestone in five charts for BBC Future. Two decades ago the figure was about 80 percent.
Set those numbers beside the coal itself and the picture gets messier. Official statistics show China’s coal consumption rose 0.1 percent in 2025, which is to say it barely moved, while coal-fired generation climbed 3.4 percent in the first half of this year and then fell 3.3 percent in the third quarter. The fleet is what keeps growing: China switched on 30 gigawatts of new coal plants in the first six months of 2026 and had 204 gigawatts more under construction. Underneath the stalled total, the parts are moving in opposite directions. Steel and cement are shrinking with the property market. Oil is falling with electric vehicles and a shipping crisis. And coal-to-chemicals has been the one large source of new emissions.
Whether this adds up to the peak China promised for “before 2030” is a question the government has left open. It has never said what a peak is, how one would be verified or how long emissions must fall before it counts. This year it also changed the yardstick it uses to report progress, in a way that independent analysts say removes roughly 700 million metric tons of annual emissions growth from the official record for 2020 to 2025.
Share and tonnage
A share falls whenever the whole grows faster than the part. Chinese electricity demand rose so fast for so long that coal could lose ground in the mix while burning more in absolute terms, and for most of the past decade it did both. Coal’s share of Chinese power fell from 70 percent in 2015 to 54 percent in 2025, according to Ember, the energy think tank. Its plants still generated 5,757 terawatt-hours in 2025, 55 percent of all the coal-fired electricity in the world.
The tonnage finally turned last year. Coal generation fell 71 terawatt-hours, or 1.2 percent, and 2025 was the first year since 2015 in which China’s fossil generation did not grow, Ember found.
Then it turned back. In the first half of 2026, coal generation rose 3.4 percent, according to a joint review by CREA and Global Energy Monitor, a U.S. nonprofit that tracks coal plants. The problem was not a lack of clean power but a failure to use it. The two groups estimated that 360 terawatt-hours of wind and solar output was curtailed in the half, up 49 percent, and wrote that the wasted electricity “could have met all demand growth and allowed coal power generation to fall.” Weak wind made it worse. In his first-quarter analysis, Mr. Myllyvirta calculated that the average wind farm ran at 22 percent of capacity, down from 27 percent.
Part of the reason is written into contracts. Coal generators are still expected to lock in annual sales covering 60 to 70 percent of the electricity they delivered the year before, a floor equal to about half of China’s market-based power trading, the CREA and Global Energy Monitor report found. Its authors, Qi Qin and Christine Shearer, said the rule turns a risk-management tool into “a barrier to renewable integration.”
The third quarter undid the first half. Coal power fell 3.3 percent and total coal consumption about 3 percent, while solar, wind and nuclear plants added 110 terawatt-hours of generation year on year, more than twice the growth in demand. Mild weather helped, Mr. Myllyvirta wrote, but did not explain all of the slowdown.
The mines tell a parallel story. China produced 4.85 billion metric tons of coal in 2025, up 1.4 percent, according to the National Bureau of Statistics, even as consumption stayed flat and coal’s share of total energy use fell 1.8 percentage points to 51.4 percent. This year output has dropped. Mines produced 3.45 billion metric tons from January to September, 3.5 percent less than a year earlier, after safety inspections that followed two fatal accidents, according to official figures reported by Mining.com.au. One of them, an explosion at a mine in Shanxi in May, killed 82 people. Domestic coal prices have risen 24 percent since mid-July.
Capacity is the number that keeps climbing. By the end of July, China’s installed solar capacity had passed coal for the first time, 1,286 gigawatts to 1,285, according to administration data reported by Caixin. Mr. Myllyvirta called the crossover symbolic on X. Coal still produces far more electricity than solar, and the coal share is the figure he pointed to instead.
This headline is of course entirely symbolic - coal still generates almost four times as much electricity and about 8 times as much total energy as solar in China.
The much bigger milestone is that coal generated less than 50% of China’s power in the first half of 2026.
The plants keep coming
The current building wave began after the power shortages of 2021 and 2022, The Associated Press reported. The National Development and Reform Commission, the state planner, has said coal should “play an important underpinning and balancing role” for years to come.
Construction started on 94.5 gigawatts of coal capacity in 2024, about 93 percent of all new coal construction worldwide that year, Yale Environment 360 reported. About 78 gigawatts came online in 2025, the most in a decade. In the first half of 2026, 30 gigawatts started up, 43 percent more than a year earlier and the most for any first half since 2016, while 2.7 gigawatts were retired. For every gigawatt that closed, 10 opened.
Behind them are 204 gigawatts under construction and 70 more permitted, together about 22 percent of the operating fleet. Approvals have slowed, to 8.6 gigawatts in the half. Construction has not: 25.4 gigawatts broke ground. From January to August China added 58.7 gigawatts of thermal capacity, mostly coal, the most for that period in 15 years, CREA’s monthly tracker shows.
Each plant runs less. Coal units averaged 1,998 operating hours in the first half, the lowest since 2021, and Climate Action Tracker, a research project that rates government climate policies, puts average utilization below 50 percent.
Officials call the plants insurance. In a written statement to Inside Climate News in June, the energy administration described coal power as the “backbone” and “stabilizing force” of the grid. Yu Aiqun, a research analyst at Global Energy Monitor, told the BBC that Beijing treats coal as the safety net under its energy transition. Qi Qin, the CREA analyst, told Inside Climate News that the motive is also local: provinces in the high-demand east want generation under their own control, and with it the investment, tax revenue and jobs.
In our view, it is a competition, and coal plants have an institutional advantage.
Daniel M. Kammen, a professor at Johns Hopkins, told the same publication that the security case was out of date. “So the old argument that you need coal for security is definitely incorrect,” he said. In February, Mr. Myllyvirta told Yale Environment 360 that Beijing would soon have to “put brakes on coal power plant construction and start closing down older coal power plants” or else slow its clean-energy expansion, and that the choice would come within a year or two. Jeremy Wallace of the Johns Hopkins School of Advanced International Studies said in comments to Carbon Brief last month that “China’s renewables buildout has slowed dramatically in 2026.” Solar additions from January to August were 58 percent below the same months of 2025.
Steel, cement and oil
Cement did more than anything else to hold China’s emissions down in 2025. Fossil-fuel emissions grew an estimated 0.1 percent, but carbon dioxide from cement fell 7 percent, leaving the total down 0.3 percent, Mr. Myllyvirta calculated. Cement output fell 6.9 percent to 1.69 billion metric tons and crude steel 4.4 percent to 960.8 million, according to the statistics bureau.
The slide has sped up. Cement production in the first nine months of 2026 was 1,131 million metric tons, down 10.2 percent and the lowest for the period in 18 years, according to industry data reported by CemNet. CREA puts the decline at 8.6 percent. Crude steel output fell 2.9 percent to 724.77 million metric tons, according to the bureau. Cement and clinker exports rose 181.6 percent.
No climate policy produced that. The collapse of China’s real estate sector did, Belinda Schäpe, who leads CREA’s China team, told the BBC.
Oil was the other surprise. After the Strait of Hormuz closed, China cut oil imports 32 percent in the second quarter, and oil consumption fell 9 percent overall and 16 percent in transport, Mr. Myllyvirta found. It was the first time falling oil use had pulled China’s total emissions down, though about 60 percent of the import drop came from drawing on stockpiles. In September, electric vehicles took 67 percent of passenger car sales and 50 percent of heavy-truck sales. Mao Shiyue of the International Council on Clean Transportation told the BBC that a mandate drives the trucks: companies in dozens of industries must run a set share of them or face production cuts on heavily polluted days.
The chemical exception
One sector kept growing through all of it. Carbon dioxide from China’s chemical industry rose 12 percent in 2025 as its coal use climbed 15 percent and its oil use 10 percent, Mr. Myllyvirta’s full-year analysis found. From 2020 to 2025 the industry accounted for more than half of the growth in China’s fossil-fuel use, including 40 percent of the coal growth and all of the oil growth, and fuels used as raw material rose from 7 percent of fossil consumption to 13 percent, he calculated in May.
The chemistry is expensive in carbon. Coal-to-olefins plants, which make plastics feedstock from coal, emit about 10 times as much as oil or gas routes, and only about 30 percent of the coal’s carbon ends up in the product. China mines more than 90 percent of the coal it uses but imports about three-quarters of its oil, Inside Climate News reported, which makes coal chemicals a security project as much as a commercial one. Climate Action Tracker wrote that excess coal supply has been partly absorbed by the industry’s expansion. A 2022 rule change exempted coal used as industrial feedstock from some energy-consumption controls.
The Hormuz crisis was expected to set off a boom. When oil and gas prices jumped at the start of the Iran war, China broke ground on three coal-to-gas and coal-to-chemicals projects worth about $10 billion, Ms. Yu and two Global Energy Monitor colleagues wrote in Climate Home News in June. They counted 34 projects under active consideration, more than 1 trillion yuan ($150 billion) of planned investment, that could add about 300 million metric tons a year of coal demand, roughly South Africa’s entire coal mining capacity.
Output has not followed, so far. Coal use for chemicals grew 20 percent in the first quarter and slowed to 8 percent in the second, because plants were already running near capacity, Mr. Myllyvirta wrote. Growth slowed again in the third quarter as expensive coal squeezed margins and higher product prices cut demand.
The last pause
China has been here before. Coal consumption peaked at 4.24 billion metric tons in 2013, by a Brookings estimate, after the State Council’s air pollution plan and bans on new coal plants in three industrial regions. Use fell about 3 percent in 2014 and kept falling through 2016, and China’s energy-related emissions dropped in 2015 and 2016 after more than a decade of rapid growth. Brookings argued that the top had passed.
It hadn’t. A construction stimulus that began in the autumn of 2016 revived steel and cement, and coal use grew 3 percent in 2018, the biggest jump in Chinese carbon dioxide since 2011, Greenpeace’s Unearthed reported. Even after the 2017 rebound, coal use was 8 percent below its 2013 level, the Rhodium Group estimated.
Ms. Schäpe says this time is different: the 2014-16 fall came from cutting heavy industry, while today’s rests also on an “enormous expansion” of clean energy and the shift of machines and vehicles to electricity. The third quarter supports her in part, since clean generation grew more than twice as fast as demand. But the property slump, a weak economy and a mild summer did some of the work. Peter Hannam wrote in RenewEconomy that some of the drop “might be attributable to China’s economic slowdown,” and Mr. Myllyvirta wrote that weaker growth as well as weather slowed demand.
The stimulus risk looks smaller than in 2016, on his reading. Reported GDP growth is within target, which reduces pressure for more support, and the measures announced so far favor “six networks” such as power grids and transport that already fall under green and zero-carbon policies.
An undefined peak
President Xi Jinping pledged in 2020 that China would peak carbon dioxide emissions before 2030 and reach carbon neutrality before 2060. In 2021 he said coal consumption would be gradually reduced from 2026 to 2030. The 15th five-year plan, approved in March, set a different course: coal and oil use are to peak, with no national cap, no peak level and no year. Mr. Myllyvirta and Ms. Schäpe wrote at the time that “this is clearly walking back the commitment to gradually reduce coal consumption made by President Xi back in 2021,” and that the wording leaves room for coal use in power and chemicals to keep growing past the targeted peak. The carbon peaking action plan the State Council issued in July sets a 17 percent cut in carbon intensity from 2025 to 2030 and a 25 percent non-fossil share of energy use, and promises the “timely achievement” of the peaking goal without defining it.
Mr. Myllyvirta described the gap to the BBC.
The government has not defined what it means for emissions to have peaked ... and how this would be verified.
The yardstick moved too. A footnote in the statistics bureau’s February communiqué defines carbon intensity as emissions from “energy activities and industrial production,” a scope that adds process emissions such as those from cement kilns and appears to leave out fossil fuels used as chemical feedstock. Under the old measure China cut carbon intensity 12.4 percent from 2020 to 2025, short of its 18 percent target. The five-year plan reported 17.7 percent. By Mr. Myllyvirta’s calculation, the new figures imply emissions rose about 7 percent over those five years, roughly half the 14 percent implied before, a difference of about 700 million metric tons in annual emissions, comparable to the yearly output of Germany or South Korea. Falling cement output and carbon locked into products explain part of it. About 380 million metric tons he could not account for.
He spelled out what that means for the peak pledge in the same analysis.
This target can now be met even if CO2 emissions go up between 2025 and 2030
The other targets leave the same room. CREA estimated that the 17 percent intensity target allows emissions to grow 3 to 6 percent by 2030 if GDP grows 4.5 to 5 percent a year. China’s 2035 pledge, which Mr. Xi announced at the United Nations on Sept. 24, 2025, is to cut net greenhouse gas emissions 7 to 10 percent “from peak levels,” according to the Foreign Ministry’s account. A higher peak makes that cut easier to reach. Carbon Brief has warned that unless provinces get binding carbon targets, local governments and industries have an incentive to push emissions up early and lock in a higher baseline, the problem known as “storming the peak.”
Mr. Myllyvirta posted his findings on the metric on X in May, putting the size of the revision in plain terms.
New from me: China made a dramatic change to its carbon emissions numbers in the 15th five-year plan, erasing half of the previously reported emissions growth from 2020 to 2025. In absolute terms, the change is equal to the total emissions of Germany or South Korea.
Weighing the objections
The strongest defense of the new accounting came in August from Hu Min, Li Xindi and Yang Li of the Institute for Global Decarbonization Progress, writing for Dialogue Earth. Bringing cement and other process emissions under a mandatory intensity target is “a step forward in governance,” they wrote, and progress on the 2030 pledge will be judged on national inventory data reported to the U.N., not on a five-year-plan indicator. On whether the change makes the target easier to meet, their answer was: “In an accounting sense, no.”
the scope change narrows the gap between domestic and international indicators, but it does not rewrite the NDC.
On the formal point they are right. The pledge China filed under the Paris Agreement is assessed through its national inventory, and a footnote in a domestic communiqué cannot change it. But the inventory has an open question of its own. A U.N. technical expert review of China’s first transparency report, published this fall, encouraged China to state clearly how feedstocks and non-energy use of fuels are accounted for. Neither document fixes the level from which the 2035 cut is measured. And the three authors conceded that the timing flatters the 2021-25 result, because cement output fell sharply during the property slump.
The second objection runs the other way: the peak has already happened, and the definition is a technicality. Climate Action Tracker’s June assessment says China’s emissions “may have peaked in 2025” at 14.8 billion metric tons of carbon dioxide equivalent if the downward trend continues, and projects 11.2 to 14.2 billion metric tons in 2035. The BBC described the tracker as finding a 2024 peak and a fall of about 18 percent by 2035, figures that do not appear in its current assessment. CREA’s series puts the top in early 2024, measured as a rolling 12-month total. In February, Mr. Myllyvirta told Agence France-Presse: “Because the relative drop is so small, we can’t say with certainty yet that it’s a fall.” This year’s emissions are down an estimated 0.5 percent so far. On the record to date, that is a plateau. It is not yet a decline.
What comes next
COP31 opened in Antalya, Turkey, on Nov. 9 and runs through Nov. 20. Vice Minister Li Gao heads China’s delegation, and Vice President Han Zheng is to attend the leaders’ summit on Nov. 11 and 12 as Mr. Xi’s special representative, Carbon Brief reported, citing Xinhua. Beijing’s stated priorities for the talks are implementation and finance from developed countries. It has not declared a peak.
Ms. Yu has argued that the shape of the descent matters more than the date of the summit, and that even a 1 percent cut would weigh heavily on global climate action. The statistics bureau’s annual communiqué, which sets the official coal and carbon-intensity figures, came out on Feb. 28 this year.
Before then, China owes the U.N. its second biennial transparency report, due by Dec. 31 under Paris Agreement rules. Mr. Myllyvirta wrote in May that the report should make clear which emissions China is counting.