Notebook - Inquiry 08
Below Half Is Not Gone
Coal fell below half of China's electricity mix while its output and fleet still grew.
Working thesis
Coal's 49.7 percent generation share is a real structural milestone, but share, output, installed capacity, and utilization answer different questions: coal can lose share while a larger fleet still generates more electricity, and integration constraints can coexist with rapid renewable growth.
- China electricity
- coal power
- renewable energy
- energy transition
- grid integration
- curtailment
The four-layer frame
Bounded frame - January through June 2026
Below half is one layer, not the whole system
A percentage is a fraction, not a physical quantity. Coal can supply a smaller fraction of a larger electricity system while still producing more kilowatt-hours. The below-half headline becomes useful only after the denominator and the other layers are kept visible.
This Notebook therefore asks four separate questions: what share each source supplied, how much electricity coal produced, what capacity existed, and how intensively the system used that capacity. No shared axis combines them, because their units and evidence kinds are not commensurate.
Non-additive figure - four units of analysis
One system, four different questions
01
Generation mix
What share of total electricity came from each source?
Coal-fired generation share
Official measurement49.7%
percent of total electricity output - January-June 2026
Reading: Non-coal sources supplied slightly more than half of first-half electricity output.
Counter-reading: Below half does not mean coal disappeared or that its absolute output declined.
Comparison, method, boundary, and sources
Comparison: First half-year below 50 percent
Basis: NEA total-generation share
Boundary: Not a full-year share, capacity value, fuel-consumption total, or emissions measure.
Renewable generation share
Official measurement41.2%
percent of total electricity generation - January-June 2026
Reading: Renewables served a large and growing share of the system.
Counter-reading: Coal and renewables do not exhaust the generation mix; nuclear, gas, and other sources remain outside this pair.
Comparison, method, boundary, and sources
Comparison: About 9 percent more renewable generation year over year
Basis: NEA renewable-generation share
Boundary: Generation share is not installed capacity and does not reveal hourly availability or location.
02
Generation volume
How much coal-fired electricity was produced?
Coal-fired generation
Official measurement2.5 trillion kWh
trillion kilowatt-hours - January-June 2026
Reading: Coal remained an enormous source of electricity even after its share fell below half.
Counter-reading: An absolute output level alone does not show whether coal is gaining or losing relative importance.
Comparison, method, boundary, and sources
Comparison: Absolute first-half output
Basis: NEA coal-fired generation total
Boundary: Not installed capacity, thermal generation, coal production, consumption, or emissions.
Coal-fired generation change
Independent analysis+3.4%
percent change - H1 2026 versus H1 2025
Reading: Coal generation grew, but more slowly than the total system, allowing its share to fall.
Counter-reading: The analysts' series is not interchangeable with official thermal-power data or the IEA's annual 2025 comparison.
Comparison, method, boundary, and sources
Comparison: Year over year
Basis: CREA/GEM bounded coal-generation series
Boundary: An analyst-calculated first-half change, not an official full-year observation.
- China electricity-demand growth forecast - 5.5%
- Forecast - Full-year 2026 - IEA forecast versus 5.2 percent in 2025. Not an observed first-half total and not a sufficient causal explanation for the 3.4 percent coal change.
03
Installed capacity
What generating equipment was connected or commissioned?
Combined wind and solar capacity
Official measurement1.95 billion kW
billion kilowatts - End of June 2026
Reading: Wind and solar capacity expanded rapidly enough to reshape the fleet.
Counter-reading: Nameplate capacity cannot be read as the same amount of generated electricity.
Comparison, method, boundary, and sources
Comparison: +16.8 percent year over year
Basis: Installed nameplate capacity
Boundary: Does not encode utilization, curtailment, dispatch priority, transmission, or storage.
Coal capacity commissioned
Independent analysis30 GW
gigawatts - January-June 2026
Reading: A large previously approved pipeline continued to add operating coal capacity.
Counter-reading: Gross commissioning is not net fleet growth and does not establish future output or motive.
Comparison, method, boundary, and sources
Comparison: +43 percent year over year
Basis: Gross units entering operation
Boundary: Separate from retirements, construction starts, proposals, permits, utilization, and generation.
- Coal capacity retired - 2.7 GW
- Independent analysis - January-June 2026 - Retirements during the same period. A retirement total does not prove when replacement capacity generated electricity.
- Coal capacity starting construction - 25.4 GW
- Independent analysis - January-June 2026 - New construction starts. Construction start is not completion, operation, or future utilization.
04
System use
How intensively were assets used, and what clean output was not absorbed?
Average coal utilization
Independent analysis1,998 hours
utilization hours - January-June 2026
Reading: The coal fleet grew faster than the electricity produced from it, lowering average use per unit of capacity.
Counter-reading: Lower utilization does not by itself measure reliability value, profitability, retirement timing, or emissions.
Comparison, method, boundary, and sources
Comparison: Down 2.9 percent year over year
Basis: CREA/GEM average utilization calculation
Boundary: A first-half average, not a full-year capacity factor or plant-by-plant dispatch record.
- Prior first-half utilization - 2,056 hours
- Independent analysis - January-June 2025 - 58 more hours than H1 2026. The comparison shares the report's series and does not substitute for official plant-level data.
Estimated wind and solar curtailment
Modeled estimate360 TWh
terawatt-hours - January-June 2026
Reading: Clean electricity availability exceeded what the system absorbed in some places and hours.
Counter-reading: The national estimate cannot prove that every curtailed unit could have displaced coal at the same time and place.
Comparison, method, boundary, and sources
Comparison: +49 percent year over year
Basis: Reported values plus modeled unreported curtailment
Boundary: Not an official observed total; depends on provincial data availability, weather regression, and spatial and temporal constraints.
- Electricity-demand growth - 258 TWh
- Independent analysis - January-June 2026 - CREA/GEM national-scale comparison. National arithmetic is not proof of a feasible one-for-one dispatch counterfactual.
Share - official first-half measurement
Coal lost majority share
The NEA's 49.7 percent coal share marks the first half-year below 50 percent. The same official record places renewable generation at 41.2 percent. These figures support a structural reading without implying that coal and renewables are the whole mix.
Kilowatt-hours - level and change
Coal generation still increased
Coal generation still totaled 2.5 trillion kWh. CREA/GEM's bounded series estimates that this was 3.4 percent above the first half of 2025. The apparent contradiction disappears once total electricity demand and non-coal generation grow faster than coal output.
Fleet - renewable and coal additions
Capacity expanded on both sides
Wind and solar nameplate capacity reached 1.95 billion kW, up 16.8 percent year over year. At the same time, 30 GW of coal capacity entered operation, only 2.7 GW retired, and another 25.4 GW started construction. These are fleet measures, not generation totals.
Use - utilization and modeled curtailment
Integration became the binding question
CREA/GEM reports that average coal utilization fell from 2,056 to 1,998 hours as the fleet expanded. Its 360 TWh curtailment estimate combines reported values with modeled unreported wind-and-solar output. Both measures point toward an integration problem, but neither proves a simple counterfactual.
Interpretation - four compatible lenses
Four readings the same system can support
Reading 01
InterpretationA structural milestone
Coal below half is meaningful because non-coal sources served a larger share of a growing electricity system.
Boundary: The milestone is a first-half generation-share observation, not a full-year emissions result.
Reading 02
InterpretationAbsolute levels still matter
Coal's share fell while its electricity output rose and new capacity entered operation.
Boundary: The record does not establish a durable future direction from one half-year.
Reading 03
ScenarioA portfolio and incentive problem
Reliability planning, provincial incentives, capacity payments, long-term contracts, and legacy approvals may overlap in sustaining coal investment.
Boundary: The reviewed sources do not isolate one motive or estimate each factor's causal share.
Reading 04
InterpretationAn integration constraint
Curtailment and falling coal utilization place grid, dispatch, storage, and market design beside renewable construction in the transition story.
Boundary: The 360 TWh magnitude is modeled, and national availability does not prove local substitutability.
The four layers can all be true at once: coal loses share, coal produces more electricity, both renewable and coal capacity expand, and the system uses coal capacity less intensively while curtailing clean output. The transition is a rebalancing of a growing system, not a single finish line.
Limitations - no transition score
What these measures do not settle
- The 49.7 percent result covers January through June 2026, not the full year.
- Generation share, generation volume, installed capacity, utilization, fuel production, consumption, and emissions are not interchangeable.
- The official English summary repeats NEA data and is not independent corroboration.
- CREA/GEM's 3.4 percent generation change uses its bounded series rather than an official full-year comparison.
- The 360 TWh curtailment total combines reported values with modeled unreported output.
- National curtailment arithmetic cannot prove local and hourly substitutability with coal generation.
- The IEA 5.5 percent demand-growth value is a forecast, not an observed result.
- The reviewed record does not isolate the causal contribution of reliability planning, provincial incentives, contracts, capacity payments, or legacy approvals.
Source trail - roles and evidence kinds preserved
Eight source records, ten direct URLs
The publication refresh adds the exact NEA item while preserving its hub and both CREA/GEM report URLs.
Open the source trail
01 - Primary official measurement
National Energy Administration press conference on first-half 2026 energy conditions
National Energy Administration
Controls the official Chinese-language 49.7 percent share, 2.5 trillion kWh coal generation, and 1.95 billion kW wind-and-solar capacity figures.
Limit: Official measurement and characterization, not independent validation; the hub contains multiple records with distinct definitions.
02 - Official English summary
China's coal-fired power output share falls below 50 pct for first time in H1
State Council / Xinhua
Provides the official English-language share, generation, renewable output, and capacity summary.
Limit: Repeats official NEA data and does not independently validate the figures or make unlike measures interchangeable.
03 - Official later-period context
Energy Production in July 2026
National Bureau of Statistics
Supplies July and January-July coal-production and thermal-generation context after the H1 window.
Limit: Raw coal, thermal generation, coal-fired generation, capacity, consumption, and emissions are different series; July does not replace H1.
04 - Independent analysis and modeled estimate
Built for backup, contracted to run
CREA and Global Energy Monitor
Controls the 3.4 percent coal-generation change, coal fleet flows, utilization, and 360 TWh modeled curtailment estimate.
Limit: CREA is an advocacy/research organization and GEM supplies tracker data; curtailment combines reported and modeled values and is not locally substitutable one for one.
05 - Independent unit-level dataset
Global Coal Plant Tracker
Global Energy Monitor
Supplies the unit-level coal project status data used in the CREA/GEM report.
Limit: Project classifications depend on sourced records and periodic updates; proposal, permit, construction, operation, and retirement are distinct states.
06 - Independent annual historical analysis
Global Energy Review 2026 - Coal
International Energy Agency
Reports that China coal-fired generation fell about 1.5 percent in 2025 while almost 80 GW of coal capacity was commissioned.
Limit: Annual 2025 context, not an H1 2026 measurement and not interchangeable with the CREA/GEM first-half change.
07 - Independent forecast
Electricity Mid-Year Update 2026 - Executive summary
International Energy Agency
Forecasts 5.5 percent China electricity-demand growth in 2026 compared with 5.2 percent in 2025.
Limit: A forecast, not an observed full-year result and not a causal estimate for the H1 mix.
08 - Independent prior-year analysis
Built to peak? Coal power expansion runs out of room in China
CREA and Global Energy Monitor
Provides prior-year context for the coal construction pipeline and commissioning surge.
Limit: Proposals are not operating capacity, gross commissioning is not net fleet growth, and neither proves future utilization.
Publication refresh - September 1, 2026
What changed after the August 30 ledger
The source set was refreshed without rewriting the dated research snapshot or changing evidence labels.
Read the publication refresh
The September 1 publication refresh added the exact NEA conference item rather than relying only on its hub, rechecked every admitted source and both CREA/GEM URLs, and preserved the H1 frame because no correction or superseding full-year release was found.
The refresh did not convert access into verification. Official summaries remain official, the CREA/GEM calculations remain independent analysis or modeled estimates, and the IEA demand figure remains a forecast.
Open - comparable full-year series
What would change the reading?
What comparable full-year official series will show whether coal share, output, utilization, capacity, curtailment, and electricity-sector emissions converge or continue to diverge?