Clean Grid, Gas Boiler or Electric Heat: How Ember's China Data Rewrites the Factory Energy Calculation

By Michael Torres | WorldIndustrialWeekly — General

Short answer: the default assumption behind most industrial energy budgets on the mainland is eroding, but not uniformly. Ember's tracking shows clean additions covering more than the country's entire demand rise last year, coal generation slipping only marginally, and fuel use already past its apex in eight of 11 industrial branches. For owners and operators, that converts a policy headline into a rebuilt comparison table for siting, power procurement and process-heat capital spending.

Image credit: An electronics factory in Zhuhai, China; Chris from Shenzhen, China via Wikipedia.

What follows is a side-by-side reading of the dataset: what the province-level and sector-level numbers actually prove, what they do not prove, and which energy asset wins in each specific thermal duty. Column position: we are comparing options and trade-offs, not promoting a technology vendor.

Is coal-fired generation really falling in China, or only decelerating?

Conclusion first: both, and the two signals carry different planning implications. Ember separates the picture into eight provinces where coal output is now contracting — Hunan and Shandong appear in that list — plus nine more where expansion has slowed markedly. At the national level the energy think tank's country tracker records a 71 TWh drop in coal generation, its first absolute fall since 2015, with fossil fuels supplying 58% of the electricity mix in 2025 against roughly 62% in 2024.

Scale check that most coverage skips: those 17 regions are drawn from the 26 coal-heavy provincial-level units Ember monitors, and together they hold more than half of national thermal capacity. Turnover is therefore partial, not unanimous — some provinces still commission coal for local reliability. And at national level, clean sources produced 42% of the electricity in 2025, statistically the same as the 43% global average quoted in Ember's own country data. Treat this as a shifting cost curve, not a completed switch.

Signal in the data	What it changes for an industrial buyer	What it does not mean
Coal generation contracting in eight provinces (incl. Hunan, Shandong)	Average and marginal emissions intensity of site power falls; scope 2 reporting gets easier	Not security of supply — balancing resources may still be fossil
Coal growth flattening in nine additional provinces	New coal capacity becomes harder to justify on utilisation economics	Not capacity retirement; units remain built and callable
National coal generation down 71 TWh (first fall since 2015)	Spot-price volatility increasingly driven by hydro and renewable output, not demand	Not a structural collapse — new approvals continued through 2025
Clean share 42%, versus 43% globally	China's grid is at world parity, so clean-power siting advantages are provincial, not national	Not a lead over the EU or US grids on a like-for-like basis

Which provinces belong on a siting shortlist now?

Conclusion: provincial grid trajectory now matters as much as labour cost and logistics, because it sets both future tariff exposure and the credibility of a customer-facing emissions claim. Comparing the archetypes visible in Ember's provincial work is more useful than chasing a single "greenest" province.

Province archetype	Examples in the reporting	What to expect over 2026-2030	Buyer action
Post-peak manufacturing provinces	Shandong, Hunan, Shanxi, Anhui, Henan	Clean buildout steadily eroding coal dispatch in large industrial demand centres	Good candidate sites for electrified process heat; confirm 3-year dispatch trend before committing capex
Large renewable export bases	Hebei, Inner Mongolia	Abundant, comparatively cheap bulk power, with grid-congestion risk at peak transfer periods	Negotiate flexible connection and interruptibility terms; check curtailment patterns
Coal-linked balancing provinces	Gansu — one provincial analysis noted its clean share slipped by roughly 1.3 percentage points as new coal was relied on for system support, with coal and wind/solar rising together	Improving structure can stall when integration capacity binds	Do not extrapolate national averages here; verify the province's own coal-plus-renewables build plan
Hydro-dominated western provinces	Qinghai, Tibet	Headline clean share swings with rainfall, masking strong wind and solar growth	Assume dry-year volatility; contract firm, dispatchable volumes where needed

The archetypes above are WorldIndustrialWeekly's synthesis of Ember's provincial findings plus published provincial breakdowns; they are intended as a screening layer, not a substitute for site-specific due diligence.

Which industrial sectors have crossed the fossil-fuel peak, and which have not?

Conclusion: the peak is concentrated in light manufacturing, and it arrived without output contraction. Ember finds fuel use already past its apex in eight of 11 tracked branches, textiles, food, pharmaceuticals and machinery among them, and its review states that electricity now supplies about three-quarters of final energy in light manufacturing such as machinery, electronics and textiles.

What has not peaked is the high-temperature cluster: chemicals, non-ferrous metals, steel and cement, plus mining and non-metallic mineral products — a group that carries roughly 40% of industrial final energy use, where fuels double as feedstock and equipment lives for decades. Ember's forward reading is that progress now hinges on these sectors, a conclusion echoed in Agora Energiewende's 2026 China status review.

Magnitude, with a caveat: in Ember's sector tracking, absolute fuel use sits between 26% and 71% below each sector's peak, depending on branch. Those figures rest on the think tank's own sample construction and are not reconciled against official consumption statistics — a limitation raised in Chinese analyst reviews of the report. Use them as direction of travel, not as a procurement benchmark.

Sector group	Electrification readiness today	Typical heat duty	Decision implication
Machinery, electronics, textiles, food, pharma (part of the eight of 11 already peaked)	High — commercial electric boilers and heat pumps available	Low to mid temperature, batch or continuous	Electrify now where tariff arithmetic works; capture efficiency first
Rubber, plastics, man-made fibre	Medium	Thermal oil, steam at moderate pressure	Hybrid systems; stage retrofits around equipment replacement cycles
Metals smelting, non-metallic minerals, chemicals, mining	Early — partial penetration of electrification	High temperature, continuous, capital-locked	Watch feedstock substitution and green power access; do not plan around near-term fuel elimination

Gas boiler, electric steam boiler or heat pump: how to choose by duty, not by slogan

Conclusion: pick by temperature band, duty cycle and tariff structure. The fuel itself is a secondary variable, and in several cases it is currently the cheaper input on a cost-per-GJ basis even as it loses on site-emissions accounting.

Option	Practical temperature range	Efficiency logic	Where it wins	Watch-outs
Retained gas boiler	Full range, incl. high-pressure steam	85-95% combustion efficiency	Intermittent high-temperature duties above ~200 C; sites with firm gas supply and old electrical infrastructure	Site emissions exposure, fuel-price volatility, future permit tightening
Electric steam boiler	Any steam duty, incl. short spikes	Near-100% conversion at the asset; no flue losses	Small or distributed loads, startup-heavy lines, sites needing rapid turndown	Transformer and connection capacity; average electricity tariffs can exceed fuel cost per GJ
Industrial heat pump	Generally hot water and low-pressure steam, up to roughly 90-120 C	COP above 2 divides purchased electricity	Wet processing with available waste heat — rinse baths, effluent, exhaust air	Capex, water quality, ceiling on supply temperature
Heat pump plus solar thermal or recovered heat	Pre-heating to mid temperature	Offsets top-of-stack electricity	Greenfield plants with roof or land area	Integration complexity, longer payback

Two evidence points anchor that comparison. First, the Lawrence Berkeley National Laboratory team behind Electrifying Industrial Heating in China (2023) and the companion techno-economic study of textile steam and thermal-oil boilers estimated that electric steam boiler adoption in China's textile sector could save on the order of 84 PJ of final energy annually — while warning that swapping combustion boilers for electric ones can raise energy costs, because average Chinese electricity prices exceed fossil fuel prices per unit of delivered energy. Second, a 2026 preprint on electrified textile steam generation places high-temperature heat pumps at the lowest levelised cost of heat, roughly USD 12-30 per GJ under current grid conditions — attractive, but only where a soft waste-heat source exists. Both readings point the same way: electrify the easy degrees first.

Case in point: squeeze the duty before switching the fuel

A dyeing operation in Changzhou, Jiangsu, illustrates the sequencing logic that the Ember data rewards. The plant adopted equipment that atomises dye liquor with high-pressure air before it meets the fabric; "This new machine greatly reduced energy consumption, which was very high in traditional dyeing. With this new technology, the factory could save 200,000 tons of water and three million kWh of electricity per year," said Wang Cunshan, the factory's president, as reported by China Science and Technology Daily. The relevant lesson for a comparably sized textile or food site today: process-side efficiency gains arrive faster and cheaper than a full fuel switch, and they lower the electrical load that any later boiler or heat-pump retrofit must serve.

What Ember's analysts say — and where the report draws its own limits

Muyi Yang, a senior analyst at Ember, framed the structural shift directly: "The clean electricity system is gaining both strength and scope, and the fossil fuel system is already registering the shift," adding that "You can see fossil fuel use flattening out, sector by sector and province by province." The same work stresses timing rather than unanimity: provinces do not turn at once, some still raise coal output, and the electricity dimension of the substitution is far more advanced than total energy. That asymmetry is the single most important qualifier for anyone reading the story as a finished fuel transition.

Six misreadings that produce poor energy capex

• Reading "clean met demand growth" as "clean met energy demand". The finding is power-sector specific; process heat, kilns and furnaces move far more slowly.
• Treating 42% national clean share as a site input. The variance is provincial, which is why the archetype screen above beats the national headline.
• Assuming coal decline equals coal retirement. Capacity kept rising while output plateaued, and since 2024 a capacity-payment mechanism pays units for availability rather than energy delivered — plants persist, re-tasked as balancing resources.
• Equating lower fuel use with shrinking industry. Declines occurred while output grew; rising per-capita industrial value added is consistent with substitution by electricity, not deindustrialisation.
• Assuming electrification automatically lowers bills. Where tariffs exceed fuel cost per GJ, the meter cost rises even as on-site emissions fall.
• Buying certificates instead of building flexibility. Environmental instruments hedge reporting risk, not curtailment, congestion or outage risk.

Action checklist for the next four quarters

1. Sub-meter steam, hot water and compressed air by temperature band; without that, no option comparison is defensible.
2. Run the tariff arithmetic locally before committing: electricity-versus-fuel cost per delivered GJ, including demand charges and any connection upgrade.
3. Electrify the lowest-temperature, highest-recovery duties first — pre-heating, wash baths, low-pressure steam — where heat pumps divide purchased power.
4. Apply a provincial clean-share and dispatch screen to any new site, and re-test it annually rather than at signature.
5. Use the formal market rather than bilaterals only: China's green power trades reached 328.5 billion kWh in 2025, up 38.3% year on year, within a record 6.64 trillion kWh of traded electricity equal to 64% of consumption, per National Energy Administration figures reported by China Daily.
6. Budget for rising environmental-attribute costs. Certificate prices steepened: 2025-vintage volume averaged about RMB 5.57 per certificate against RMB 2.12 in 2024, per the regulator's green certificate development report.
7. Retain gas or other dispatchable heat for high-temperature intermittent duty; plan it as peaking, not baseload.
8. Monitor whether new local coal is being built as flexibility or balancing support — it changes outage risk and capacity-charge pass-through exposure in your tariff.
9. Keep supplier and customer claims scoped honestly: sectoral peaks in eight of 11 branches are not a corporate-wide fuel exit.

FAQ

Does declining coal use mean Chinese factories are producing less?

No. Ember's analysis records the declines occurring while industrial output continued to grow, and the pattern is consistent with electrification of final energy rather than contraction — electricity supplies about three-quarters of final energy in light manufacturing.

Is it cheaper to electrify process heat today?

Sector-specific. Studies of China's textile steam systems show large technical energy savings from electric boilers, while the same analyses caution that average electricity tariffs can exceed fossil fuel cost per unit of heat. Efficiency, load shape and connection capacity usually decide before fuel choice does.

Which provinces are the safest bets for a clean-power plant?

The strongest screen is coal generation trending down or flattening alongside large renewable buildout — a group that includes Shandong and Hunan, with Shanxi, Anhui and Henan showing clean power beginning to displace coal in coal-linked economies. Avoid inferring from national averages in balancing provinces such as Gansu, or in hydro-dependent western regions.

Can exporters use this shift in customer emissions claims?

Partly, and only with instruments. Grid intensity is improving, but a supplier's scope 2 position still depends on contracted green power or certificates, whose pricing has been rising. Verify with actual trade or certificate records rather than regional averages.

Is China's grid now cleaner than the world average?

Essentially at parity: 42% clean generation in 2025 against a 43% global average, per Ember — with coal still the swing fuel and its decline measured in tens rather than hundreds of terawatt-hours.

Sources and further reading

• Ember — China Energy Transition Review 2026: https://ember-energy.org/latest-insights/china-energy-transition-review-2026/
• Ember — China energy trends (generation, clean share, coal output change): https://ember-energy.org/countries-and-regions/china/
• China Government Net / NEA data coverage — 2025 power consumption tops 10 trillion kWh, non-fossil capacity above 60%: https://english.www.gov.cn/archive/statistics/202601/17/content_WS696b548ec6d00ca5f9a08a0e.html
• China Daily — record 6.64 trillion kWh traded in 2025, green power trading up 38.3% to 328.5 billion kWh: https://www.chinadaily.com.cn/a/202602/10/WS698a8ee1a310d6866eb386a0.html
• LBNL / Hasanbeigi et al. — Electrifying Industrial Heating in China: https://eta-publications.lbl.gov/sites/default/files/chinaindustrialelectrification_final.pdf
• Energies (MDPI) — Electrified process heating in textile wet processing: https://www.mdpi.com/1996-1073-15-23/8939
• Agora Energiewende — China's energy transition and climate status report 2026: https://www.agora-energiewende.org/fileadmin/Projekte/2026/2026-10_CN_china_data_annual_review/A-EW_415_China%E2%80%99s_energy_transition_and_climate_status_report_2026_WEB.pdf
• Carbon Brief — new coal approvals in 2025 and the capacity-payment mechanism: https://www.carbonbrief.org/rush-for-new-coal-in-china-hits-record-high-in-2025-as-climate-deadline-looms
• China Science and Technology Daily — A Decade of Glory: China's industrial development (Changzhou dyeing case): https://digitalpaper.stdaily.com/http_www.kjrb.com/ywtk/html/2022-06/23/content_537492.htm

Bottom line for readers of this column: the China story is no longer only about installing solar. It is about which factory duty moves first, which province pays the difference and which fuel keeps a justified role as back-up. For WorldIndustrialWeekly, published by Industrial Weekly, the operating takeaway is that fossil-linked energy budgets need re-testing this year — province by province, degree Celsius by degree Celsius — rather than re-written wholesale. Corrections, dataset requests and editorial contact: news@industrialweekly.com.