LDES vs lithium-ion batteries

Why this comparison matters in 2026

Grid-scale storage stopped being a single question the moment renewables passed 30% of generation in leading markets. Lithium-ion BESS solved the short-duration problem — frequency response, intra-day arbitrage, capacity firming — and did it so successfully that global deployments passed 315 GWh of additions in 2025 with 450+ GWh forecast for 2026. But the economics of stacking more lithium-ion cells stop making sense somewhere between 6 and 10 hours of duration, and that is exactly where the next problem starts: multi-day wind droughts, overnight solar gaps and seasonal balancing.

Long Duration Energy Storage — pumped hydro, flow batteries, compressed and liquid air, thermal, gravity — is the class of technology that takes over from there. This guide compares the two on the numbers utilities, developers and investors actually use: duration, round-trip efficiency, LCOS, siting and revenue stacks.

The one-table summary

Duration: the axis that separates them

A lithium-ion BESS is a power-first asset. You buy a power block (inverters, transformers, controls) and add cells until you hit the duration you want. Doubling the duration roughly doubles the cell count, which roughly doubles the cost. That linear relationship is fine at 2 or 4 hours, uncomfortable at 8, and non-competitive at 12+.

LDES technologies decouple power from energy. A flow battery adds duration by adding electrolyte to a tank — the stack (which sets power) is unchanged. Pumped hydro adds duration by making the upper reservoir bigger. Liquid air adds duration by scaling the storage vessel. The marginal cost of another hour of storage is a fraction of what it costs on lithium-ion, so LCOS curves cross somewhere between 6 and 12 hours depending on the technology and site.

Round-trip efficiency: LDES pays a real energy tax

Lithium-ion BESS delivers 85–92% AC round-trip efficiency in the field. That means for every 100 MWh you charge, you get 85–92 MWh back. On price-arbitrage duty this is decisive: even a 50 €/MWh spread barely covers the losses of a low-efficiency asset.

LDES efficiency varies enormously by technology. Pumped hydro is 70–85%. Vanadium redox flow is 65–80%. CAES sits at 45–70% in first-generation adiabatic designs, LAES around 55–70%. Hydrogen-cycle and thermal-only pathways are typically 30–55%. In each case, the LCOS calculation only works when the discharge is long enough — and the underlying energy cheap enough (curtailed renewables, off-peak) — that the round-trip loss becomes a rounding error versus the cost of the alternative.

Cost: separate the $/kW and $/kWh conversations

The lithium-ion cost story is well known — BloombergNEF's 2024 pack-price survey put the global average at $115/kWh, with LFP packs around $95/kWh. At full-system (installed) level NREL benchmarks a 4-hour utility BESS around $350–450/kWh. That number is dominated by cells, which is why the marginal cost of another hour of duration is high.

LDES flips the ratio. A flow battery might install at $400–600/kW for the power block, but each additional hour of tank storage adds only $40–80/kWh — because tank plus electrolyte is cheap compared with cells. Pumped hydro, once built, adds duration essentially for the cost of a bigger dam. That is why:

• At 2–4 hours, lithium-ion is almost always the lowest LCOS.
• At 6–8 hours, LFP-based BESS and the cheapest LDES technologies are within a few dollars per MWh of each other.
• At 10+ hours, LDES pulls decisively ahead on LCOS.
• At 24–100+ hours, lithium-ion is not a serious commercial option — LDES is the only viable path.

Cycle life, calendar life and degradation

Lithium-ion BESS is a consumable. Modern LFP cells warranty 4,000–6,000 cycles to 80% state of health, with best-in-class BESS-grade cells now reaching 10,000+ cycles. Calendar life is 15–20 years. Most 20-year contracts include a mid-life augmentation or replacement.

Most LDES technologies do not degrade in the same way. Pumped hydro and CAES are civil-works assets with 40–80 year lives. Flow-battery electrolytes are effectively infinite; the stacks need periodic membrane refurbishment but not wholesale replacement. Thermal and gravity systems have mechanical wear but no chemistry decay. For a 30- to 50-year power-system asset, this changes the NPV materially.

Siting and permitting

Lithium-ion BESS is deliberately boring to site: containerised, dispatchable in 12–24 months, brownfield-ready, small footprint. That optionality is a real economic advantage — and one of the reasons lithium-ion won the 1–4 hour category despite theoretically higher LCOS than some alternatives.

LDES is often site-specific. Pumped hydro needs topography and water. CAES needs suitable geology or purpose-built caverns. Liquid air, flow and thermal systems are more flexible but still have footprint and civil-works implications that lithium-ion does not. Permitting timelines are typically 4–8 years for large LDES versus 1–2 years for BESS, which affects developer economics as much as CAPEX does.

Revenue stacks and merchant risk

Lithium-ion BESS operators in mature markets (GB, Texas, Australia, Germany) stack frequency response, wholesale arbitrage, capacity payments and increasingly ancillary services. As frequency-response markets saturate — GB is the textbook example — new BESS is being underwritten on merchant arbitrage plus capacity, with real revenue compression year-on-year. LFP's long cycle life is what keeps that model economic.

LDES revenue models are less mature. Today most projects are underwritten on capacity payments, reliability contracts, and public procurement (DOE Loan Programs, EU Innovation Fund, California LDES solicitations). Merchant LDES is emerging in markets where multi-day wind-drought risk creates persistent price spreads — GB Winter 2024–25 is the reference case. The 2030s revenue thesis is that as short-duration BESS saturates the arbitrage envelope, the remaining value shifts to the longer-duration edge of the curve, where LDES sits.

Which technology for which grid problem

• Frequency response, voltage support, black start. Lithium-ion BESS. Response times below 100 ms and high round-trip efficiency are decisive.
• Intra-day arbitrage (peak shifting, evening peak, solar shift).Lithium-ion at 2–4 h, hybrid or LDES at 6–8 h.
• Multi-day balancing (wind droughts, winter cold snaps). LDES — flow, LAES, CAES, iron-air.
• Seasonal storage. Pumped hydro and hydrogen-cycle. Nothing else is commercial at that duration yet.
• Off-grid, microgrid, backup. Lithium-ion for sub-24 h; LDES (usually flow or thermal) for anything that has to ride a multi-day outage.
• Data-centre backup and behind-the-meter reliability. Lithium-ion remains dominant; flow batteries are being trialled for AI-load resilience where 8 h+ duration matters.

How to think about the decision commercially

Frame it as a two-step question. First: what duration does the use case actually require, and how firm does the discharge have to be? Second: over the asset's economic life, which technology minimises LCOS at that duration and collects a defensible revenue stack in the market you are bidding into.

For most projects being financed in 2026, the honest answer is that lithium-ion BESS wins below roughly 6 hours and LDES wins above roughly 10 hours, with a contested middle. As solar and wind penetration rises through the late 2020s, the value pool moves toward the LDES end of that curve. That is why utilities, developers and investors are quietly building LDES pipelines now, even where lithium-ion still dominates today's deployment numbers.

The technology choice is a commercial choice about duration, market design and asset life — not a beauty contest between chemistries. That is the fluency battery-industry professionals need to move confidently between BESS and LDES conversations.

Informational and educational content only. Not professional, financial, legal, or engineering advice.