China’s utility-scale battery fleet, already the largest in the world, ended 2025 still carrying headroom equivalent to 23 terawatt-hours of clean electricity that could have been shifted to hours when it was wanted, according to new Ember analysis produced with full-year data for that market. That single figure is the clearest signal yet that the world’s biggest storage expansion story has shifted from hardware build-out to operational efficiency: in 2025, gigawatts were no longer the binding constraint – dispatch and market design were. Any firm exposed to storage in China, or studying it as a model for their own grid, now has to treat “shifted energy” as the decisive scorecard, not installed capacity.
## Why the World’s Biggest Battery Fleet Ended 2025 With Utility-Scale Headroom to Spare
Ember’s analysis takes stock of China’s utility-scale batteries, a fleet assembled at unprecedented speed over five years. The growth was driven, in part, by provincial mandates attached to new solar and wind projects, which required co-location of storage alongside renewable capacity, plus central policy targets for “new-type storage” that pushed installers to deliver on a near-annual-basis. 2015 before 2025 the buildout was denominated in added gigawatts, often with 1-to-2-hour durations attached to projects. The issue with that spend brings unit economics.
The 23 TWh potential figure that Ember calculates is not a forecast for storage of ever-needed; it is a measure of the energy that the existing fleet could have absorbed and injected to more valuable timing had the charging and discharging priority been managed at the whole-system level. Within that aggregate lies a more granular problem: battery assets with hundreds of gigawatt-hours of storage delivered only a portion of their energy, signing from capacity payment or mandated sunrise rules rather than what the grid was beyond the cost other hours.
There are two ways to read this. One is that dispatch routing deterministic scheduling by provincial operators – still oriented to thermal or hydro – regularly has batteries shut down or hasn’t able to cycle during the midday solar surplus, e.g. effect to a fixed two-cycle schedule. or at most once per day. The other is that, in many provinces, storage revenues are collected not by trading through spot markets but by fixed fee or usage time basis, so that operators have no reason to actively search for value (flexibility) with intelligent multiple-times hourly operation. The result is 23 TWh of clean power instead of not being time-shifted and label plant served running to cover evening peaks.
If the aggregate memory of 23 TWh would have been used, when translating it into operational country terms is: that electricity was generated in excess, with a low marginal price – and staying that way – while same days were covered in thermal generation. Ember’s estimate is the principal low-hanging fruit of the whole storage economics – a gap that can be partially closed today simply by operating the existing fleet less incorrectly.
## The 23 TWh Gap Is a Pricing Problem, Not a Battery Problem
This is the point to look at China’s development in the context of what batteries actually do. Battery industries are energy expenditures, not just capacity reservoirs: the success of a solar + storage station is measured by the energy-carrying hours shifted – or the equivalent of the otherwise-curtained renewable energy electrons sent to hours without absolute scarcity. China has built the rare case of platform in tails: already as world largest fleet, baseline capacity is sufficient.
Place the number in approximate global context. My own calculation: if China’s cumulative utility-scale BESS is on the order of 100 GW and averaged 2-hour duration at the end of 2025 (a rough envelope, deployment of a battery pack and co-located storage), simply forming a full charge/discharge cycle per day would make available energy viability at the height of on the order of 70 TWh per year. Ember’s 23 TWh gap is in that ballpark – roughly a third of the shift opportunity is left-on-the-table (assuming one complete cycle, or much more if dispatch allowed two daily cycles). To place that aggregate global significance: a 23 TWh annual shift is comparable to the order of magnitude of the entire annual energy that the EIA reports for the U.S. utility-scale battery fleet in current years (I am referring to rough- around 25-30 TWh/year – which is not an exact figure). The extent of unlivened slack is no longer market data.
This ties directly into the broader global trend of negative wholesale prices and curtailment at solar-power systems. In China, provinces like Shandong and Shanxi have seen midday wholesale pressure on markets correcting, at times, disappear or reach negative values; afternoon valley prices; and evening island peaks. In a higher-functioning spot market, storage becomes an arbitrage engine: in the afternoon it clears the right to charge, then it performs a price decline. The link between the 23 TWh report and market design is direct: after all, a battery is nothing but the reward for rewarding the operating system while doing a potentially different rhythm.
The key implication: the forecast that China hasn’t yet done a well completed macro is the sharpest proof that wholesale markets, which store prices, are the binding investment and operational constraint. Worldwide, storage under-use has partially hidden because buyers and developers hit capacity target thresholds. Demand now is not for partners from a battery, but for 23 hours of the negative-priced curtail need, unavailable bill.
## Who Should Act on China’s Storage Utilization Benchmark
– Utility planners and grid dispatchers – use the 23 TWh figure as the goal for operational dispatch forecasts. Try charging not only in the off-peak valley but during the growing midday surplus, including negative-priced windows, and allow any true environmental two-cycle operation on high solar and peak-load days; your province to province settings decide how much gap closes.
– Storage developers / independent operators – treat Ember’s number as the executive summary of the revenue for that of flexible operation: When project modeling in Chinese provinces, replace fee “two cycles” assumptions with explicit back-weighted dispatch estimation – client demand value is now high enough to matter. Value power purchase agreements with dispatch roles, deliverable energy throughput guarantees, and tracked cycles-per-day scoring more than installed capacity.
– Renewable project owners with co-located storage – not to include storage that counts for integration quotas; integrate the value of emissions freed by the extra hours into a syndicated integral PPF / revenue model. If 23 TWh hours is genuinely attainable, the storage minus same – land means that plant-specific usage should also rise, as a typical plant’s annual cycle has a meaningful uplift potential: fewer curtailment costs and added evening-priced MWh from solar half-power.
– Policy analysts and regulators – transition from capacity procurement guidelines (GW per province) to energy-effectiveness policy instruments: define a normalized dispatch efficiency annually, use the 23 TWh number as national baseline; restructure any remaining storage subsidy formulas that currently pay per installed GW because on a production-per-unit separated metric, while making spot and ancillary trading rules allowing batteries to get recharged more than once/day.
## What to Watch Next: Does ’23’ Shrink or Grow?
– Next periodic storage utilization statistical releases (e.g., NEA’s new-type storage utilization or economic compile – or Embers projected) – at annual or half-year resolution, tracked as a signal to consume ratio. If 23 TWh moves up in 2026, define a gap ratio; if it clicks down toward 10 TWh, the market design for batteries in China is correct.
– Spot market and below-price trends in Chinas’ most solar-heavy provinces: midday negative prices increasingly frequent, and Darwin the slope of afternoon – coming in. Each step in retail alignment had value of batteries shift; whether a “duck valley” becomes deeper than visible.
– Post time frame events: 4-hour-plus long-duration storage in China’s pipeline (often mandated by new-shift projects), and flexible long duration hours in provinces with high winters – if announcements shift to demand states and utilization performance (not just GW), sizing preference has changed.
– Battery health/cycle warranty claims. The current 1-2 cycles per day are at intended level, and charging rules effectively hold recharge currents in bandwidth. Once more needed moderate throughput, the real trade-off in cycle life vs. earnings revealed.
## Bottom Line
The mature question now, brought by data, quantifies the problem most market observers had suspected: China is still losing in the neighborhood of 23 spare hours™ is in 2025 of storage energy transfer awards-exactly because using batteries beyond the coal-era dispatch assumption in the balance is at stake for this. The next country and every developer have the same click-the same lesson: more storage isn’t the condition again – the limiting condition again – is the ability of the market of charging briefly, discharging, charging twice on the same solar-day, and paying operators for performance, not installation.
Read the full report at CleanTechnica.
Note: facts and figures attributed above to reflect that outlet's original reporting. Broader context, cross-sector connections, and forward-looking scenarios reflect independent analysis by our editorial team.
About this article: Drafted by Energy Ai with AI-assisted research and writing based on public reporting, then reviewed under our editorial process before publication.
Leave a Reply