Market consensus often misinterprets the surface-level outbreak of an industry as the starting point of a trend. It is argued that the development of long-duration energy storage (LDES, typically defined as continuous discharge of 4 hours or more) in China did not emerge out of thin air in 2026. Rather, following policy incubation in 2023-2024 and scaled initiation in 2025, it has reached an absolute inflection point in 2026, characterized by a closed-loop business model and an explosion in grid-connected capacity. According to the latest authoritative data released by the National Energy Administration (NEA) and the China Energy Storage Alliance (CNESA) in January 2026, the cumulative installed capacity of novel energy storage in China reached 144.7 GW by the end of December 2025, representing a significant year-on-year surge of 85%.
The more critical data lies in the structural mutation: among newly commissioned projects in 2025, the number of LDES projects (4 hours and above) soared by 44% year-on-year. Entering 2026, the increase in average storage duration has notably accelerated. This indicates that the LDES trend was established well in advance, and 2026 merely presents a qualitative leap triggered by previous quantitative accumulation (the landing of hundred-megawatt-scale projects and GW-scale independent energy storage stations). In the past, LDES failed to scale comprehensively due to a distorted business model; today, the rewriting of market rules has lifted this restriction.
Prior to the substantive phase-out of the "mandatory paired storage" policy in 2025, the commercial logic of LDES was fundamentally fractured:
First, the sunk cost constraints of the "mandatory paired storage" era: In the 1.0 era, energy storage served merely as a "compliance accessory" for renewable energy grid connection (typically requiring 10%-20% capacity and a 2-hour duration). Lacking market-oriented dispatch revenues, developers adopted extremely low capital expenditure (CAPEX) as their sole procurement criterion. Consequently, LDES was driven out by inferior alternatives due to its prohibitive upfront costs.
Second, the inability of solitary "peak-valley arbitrage" to cover high costs: During the phase without a capacity compensation mechanism, energy storage survived solely on energy arbitrage in the spot market. The exorbitant costs of large-capacity lithium-ion or all-vanadium flow batteries simply could not yield a viable internal rate of return (IRR) in a highly volatile spot market, prompting collective capital flight.
Therefore, the explosive growth of LDES in 2026 is the result of a resonance among policy mechanisms, underlying costs, and novel demand-side requirements converging at this exact moment.
1. The Revenue Side
The year 2026 marks the watershed where energy storage assets completely transition from "cost items" to "profit-generating tools." As multiple provinces explicitly incorporate grid-side independent energy storage into the capacity pricing mechanism, the revenue model shifts from solitary spot market arbitrage to a structure of "guaranteed capacity pricing + spot market arbitrage / ancillary services." Since compensation levels are strictly pegged to peak discharge duration, only systems with a long-duration discharge capability of over 4 hours can secure full capacity pricing and cover the "long-valley, short-peak" spot arbitrage intervals. Consequently, their bankability is fundamentally restored in 2026.
2. The Cost Side
The underlying physical advantage of LDES lies in the decoupling of energy and power—extending the discharge duration only requires increasing the storage medium (e.g., electrolyte or compressed air volume) without the need to add expensive power conversion equipment, yielding a significant diminishing marginal cost of expansion effect. Entering 2026, the large-scale delivery of large-capacity battery cells (500Ah+), the continuous downward trajectory of the Levelized Cost of Energy (LCOE) for all-vanadium flow batteries, and the lifecycle cycle advantages (30-50 years for mechanical storage, 20,000+ cycles for flow batteries) have been validated by financial models. The lifecycle LCOE of LDES has crossed the grid-parity threshold.
3. The Demand Side
The incremental variable in 2026 lies in the development of artificial intelligence and data centers. In 2026, the stringent demand from Artificial Intelligence Data Centers (AIDCs) for 24/7 uninterrupted green power is erupting. Intermittent energy sources like wind and solar must be paired with LDES to transform into the stable "baseload power" required by computing centers. High-energy-consumption intelligent computing scenarios have directly become the core battleground for LDES enterprises in 2026.
Furthermore, short-duration and long-duration energy storage do not exist in a zero-sum, substitutive relationship; rather, they represent rigid demand configurations based on different dimensions of the power grid.
- Short-Duration Energy Storage (< 2 hours, e.g., high-rate lithium-ion, flywheels): The core lies in power support. Acting as the grid's "shock absorbers," they primarily address millisecond-to-second frequency fluctuations and are active in ancillary service markets such as primary/secondary frequency regulation (AGC) and instantaneous reactive power compensation.
- Long-Duration Energy Storage (≥ 4 hours, e.g., flow batteries, compressed air): The core lies in energy shifting. Acting as the grid's "auxiliary fuel tanks," they primarily address extreme net load fluctuations ("duck curves") caused by the high penetration of renewable energy grid connections across peak-valley, cross-day, or even cross-seasonal cycles. They provide system-level capacity support in extreme scenarios such as smoothing output at mega-bases in desert/Gobi regions or during prolonged periods without wind or sunlight.
Similarly, why can't the timeline be delayed? The reason lies in the massive pressure that would be exerted on the physical limits of the grid if the scaling of LDES were postponed beyond 2026. As renewable energy penetration approaches the critical threshold of "system instability," short-term frequency-domain regulation is completely inadequate to resolve massive time-domain energy mismatches. The outbreak of LDES in 2026 is both the inevitable choice of capital following the validation of commercial rules, and the final infrastructure window to prevent systemic grid risks and avoid large-scale renewable energy curtailment.
