China’s Urban Energy Peak: Most Cities Haven’t Actually Peaked Yet

China’s cities have not yet reached a stable energy peak, according to a new study that challenges one of the most optimistic assumptions about the country’s climate transition. Using satellite nighttime-light data combined with official energy statistics, researchers reconstructed energy consumption for 336 Chinese prefecture-level cities from 2000 to 2022 and found that only 88 cities—roughly one in four—can be classified as having entered a stable post-peak phase. The remaining cities either recorded their highest energy use in the final year of the sample or remain close to their historical maxima, meaning that the urban energy peak remains unconfirmed for most of China’s urban system.

The study, published in Frontiers in Environmental Science, distinguishes between what researchers call observed peaks and effective peaks. An observed peak is simply the highest recorded energy consumption within the study period. An effective peak, by contrast, requires that the city’s average energy consumption over the final three years be at least 5 percent below its historical maximum. This distinction matters because a city that records its highest consumption in 2022—the final year of the sample—cannot be confirmed as having peaked. Its observation is right-censored rather than confirmed, a statistical problem that affects 202 of the 336 cities, or 60.12 percent of the sample.

The gap between observed and effective peaks is striking. While 202 cities recorded their historical maximum in 2022, only 97 cities had an observed peak no later than 2019. Applying the stricter effective-peak rule reduces the count further to 94 cities, or 27.98 percent of the sample. The researchers then classified cities into four transition stages: stable post-peak, rebound-risk, still-rising or near-peak expansion, and fluctuating transition. Only 88 cities (26.19 percent) are classified as stable post-peak, while 208 cities (61.90 percent) remain in the still-rising or near-peak expansion category. Seven cities are classified as rebound-risk cases—cities that peaked earlier but whose 2022 energy consumption had returned close to the historical maximum, with a median rebound index of 0.943, meaning approximately 94 percent of the decline from peak to post-peak minimum had been recovered by 2022.

The regional divide is equally pronounced. In eastern China, 46 of 87 cities satisfy the effective post-peak criterion, a share of 52.87 percent. By comparison, the corresponding figures are 25.61 percent for central China, 16.03 percent for western China, and 16.67 percent for the northeast. The right-censoring problem is especially severe in the west and northeast: 107 of 131 western cities (81.68 percent) and 29 of 36 northeastern cities (80.56 percent) recorded their historical maximum in 2022. Eastern cities were, on average, 21.61 percent below their historical peak in 2022, while the equivalent figures in central, western, and northeastern China were 4.99 percent, 3.12 percent, and 3.38 percent, respectively. Formal statistical tests confirm these differences are systematic, with a chi-square test rejecting equality in effective peak status across regions (χ² = 38.56, p < 0.001).

The study’s methodology is notable for its use of satellite nighttime-light data to reconstruct city-level energy consumption. Because consistent city-level energy statistics are not available for the full sample period, the researchers combined nighttime-light observations from the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) with officially reported provincial energy statistics. The approach matches provincial nighttime-light intensity with reported provincial energy consumption to estimate the relationship between light and energy use, then applies this relationship to city-level light totals. The resulting estimates are expressed in standard coal equivalent and capture total city-level energy consumption without distinguishing energy use by sector or fuel type.

The researchers conducted extensive robustness checks to ensure their findings are not driven by arbitrary parameter choices. Varying the tolerance threshold from 5 percent to 0, 3, or 10 percent changes the count of effective post-peak cities only modestly, from 94 to between 75 and 97. Extending the persistence window from three to four years lowers the count to 71. Applying a three-year moving average to reduce sensitivity to annual fluctuations increases the number of cities peaking in 2022 from 202 to 212 while reducing effective post-peak cities from 94 to 76. Across all alternative specifications that retain 2022 as the sample endpoint, the share of effective post-peak cities remains below one-third of the sample.

A pandemic-period stress test reveals how sensitive peak classifications are to the sample endpoint. Using 2019 as the endpoint, 155 cities satisfy the effective-peak criterion—substantially more than the 94 identified in the full-sample classification. The researchers then tracked these 155 pre-pandemic effective-peak cases through 2020–2022 using each city’s pre-2020 historical maximum as the benchmark. Only 60 of the 155 cities remained below that benchmark in all three years, whereas 95 exceeded it at least once. By 2022, 94 of these 155 pre-pandemic cases had returned to within 5 percent of their pre-pandemic historical maximum or had moved above it. The researchers treat this period as a stress test of peak durability rather than evidence of a causal pandemic effect.

Trend-break analysis provides complementary evidence about the nature of urban energy transitions. The researchers fitted a single-break trend model to the logarithm of energy consumption for each city, classifying cities as exhibiting acceleration, slowdown-but-still-growing, or reversal. The results show that 246 cities (73.21 percent) fall into the slowdown-but-still-growing category, while only 73 cities show a reversal pattern and 17 exhibit acceleration. This means the dominant urban energy-transition pattern in China is growth deceleration rather than absolute decline. Many cities are growing more slowly than before, but they have not yet entered a stable post-peak stage—a distinction with substantial policy implications, since cities with slower growth still need infrastructure, clean energy supply, and demand management, whereas stable post-peak cities need measures to consolidate decline and prevent rebound.

The distinction between energy peaks and carbon peaks is central to the study’s contribution. A city’s emissions can decline because of cleaner electricity, fuel substitution, industrial upgrading, or imported power—even while total energy use continues to rise. Conversely, a drop in absolute energy consumption may point to deeper structural shifts: reduced reliance on heavy industry, improved efficiency, more compact urban form, or demand saturation. The researchers argue that city-level carbon-emission peaking and energy-consumption peaking should be treated as complementary rather than interchangeable dimensions of urban transition. For energy security, infrastructure planning, and demand-side management, the absolute energy-consumption peak is a separate and policy-relevant object of inquiry.

The findings arrive at a critical moment for China’s climate commitments. The country’s updated nationally determined contribution commits to peaking carbon emissions before 2030 and achieving carbon neutrality before 2060, while the 14th Five-Year Plan targets a 13.5 percent reduction in energy intensity and an 18 percent cut in carbon intensity between 2021 and 2025. The researchers emphasize that their framework is diagnostic rather than causal—it identifies where cities stand in the energy transition without estimating the effects of particular economic or policy drivers. They suggest that the transition categories can guide demand-scenario selection in planning exercises: for stable post-peak cities, the recent post-peak demand level can serve as the central planning case; for rebound-risk cities, the historical peak should remain within the planning range; and for still-rising cities, the central scenario should not assume an imminent absolute decline. The study also notes that the four-region classification captures broad spatial gradients but cannot substitute for a functional classification based on industrial or resource characteristics, since cities at different transition stages coexist within each region.

Subject of Research: Urban energy peaking in China: timing, persistence, rebound risk, and regional heterogeneity

Article Title: Urban energy peaking in China: timing, persistence, rebound risk, and regional heterogeneity

Article References: Urban energy peaking in China: timing, persistence, rebound risk, and regional heterogeneity. (n.d.). https://doi.org/10.3389/fenvs.2026.1963897

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1963897

Keywords: China, energy transition, urban energy, carbon peak, nighttime lights, regional heterogeneity, rebound risk, climate policy, Urban, energy, peaking, timing