Two ledgers, four layers
Power capacity and energy capacity are priced separately because they scale with completely different physics.
Storage medium
Electrochemical material, water, air, heat, hydrogen, or another medium holds energy with a duration-dependent cost and loss rate.
- Measure
- $/kWh · retention · lifetime
- Failure boundary
- Reservoir cost and geography bound which medium is viable where.
Where the frontier moves
Cheaper long-duration reservoirs (thermal, chemical, or mechanical) that can wait without leaking value.
Power block
Cells, turbines, compressors, pumps, heat engines, inverters, and transformers set charge and discharge rate.
- Measure
- $/kW · efficiency
- Failure boundary
- Conversion cost and durability set a floor that duration alone cannot dilute.
Where the frontier moves
Power electronics and conversion machinery that keep costing less per kilowatt.
Site and interconnection
Land, caverns, reservoirs, foundations, safety systems, studies, substations, and permits turn equipment into a grid asset.
- Measure
- Time to energize · availability
- Failure boundary
- Location and permission can bind before any component cost does.
Where the frontier moves
Faster interconnection studies and siting that don't gate cheap hardware behind years of queueing.
Operations and contract
Dispatch, charging, degradation, augmentation, maintenance, financing, and market rules determine recoverable value.
- Measure
- Delivered MWh · dependable capacity
- Failure boundary
- Utilization and revenue certainty decide whether cheap components ever get paid for.
Where the frontier moves
Contracts that pay for readiness and avoided system cost, not just arbitrage.
Buy power once; add energy as cheaply as the duration requires
Stored energy must occupy a physical state and reversible machinery must move it. No design escapes conversion losses, leakage, maintenance, and capital tied up while waiting.
Why is cost per megawatt-hour a surface, not a line?
The model separates power-block cost (divided across duration) from energy-reservoir cost, then levelizes both against lifetime cycles and round-trip efficiency. It is a screening frame, not a bankable project forecast.
4-hour duration, 350 cycles/year over a 15-year life
- Installed cost
- $225/kWh
- Power block share
- $75/kWh · 33%
- Energy reservoir share
- $150/kWh · 67%
- Lifetime delivered energy
- 4463 kWh per installed kWh
NREL's 2024 Annual Technology Baseline uses 85% as a representative lithium-ion round-trip efficiency and assumes roughly one cycle per day for short-duration systems. The 168-hour example assumes four full discharges a year; its $70/kWh reservoir is illustrative, not an observed installed week-long system. Rare reserves need a separate capacity payment because energy-only revenue may not recover standby cost.
Calculation and boundaries
Installed cost per kWh = energy-reservoir cost + power-block cost ÷ duration. Lifetime delivered energy per installed kWh = cycles/year × 15-year assumed life × round-trip efficiency. Levelized cost = installed cost ÷ lifetime delivered energy. Excludes financing structure, augmentation schedules, calendar (non-cycle) degradation, curtailment losses, interconnection cost, and any capacity or resilience payment for readiness.
Cheap components expose the missing market product
Four-hour batteries can stack frequent services; week-long reserves exist to protect rare conditions. Financing the latter requires contracts that value dependable readiness and avoided system cost, not energy arbitrage alone.
Specify the duty cycle
Procure duration, response, ambient conditions, availability, throughput, and end-of-life capacity together, not as afterthoughts.
Separate power and energy
Scale the reservoir without duplicating every expensive conversion component that sets the power rating.
Price degradation and waiting
Include calendar life, cycles, augmentation, leakage, and idle capital tied up while the asset waits to discharge.
Pay for reliability
Give rarely used duration a bankable capacity or resilience value instead of pricing it only on energy arbitrage.
Who is building what
Iron-air multi-day batteries, redox flow cells, compressed air caverns, and liquid CO₂ thermodynamic cycles decouple energy capital from power rating. Search the record, or filter by storage mechanism.
Form EnergyIron-Air Multi-Day BatteryReversible rust technology: oxidizing metallic iron to rust with ambient oxygen during discharge, reducing rust back to iron with electricity during charge
- Reported evidence
- Form Factory 1 manufacturing opened in West Virginia; the first 100-hour demonstration was deployed in 2025, with commercial projects contracted but not yet a fleet of proven operating plants.
- Announced next step
- A 100-hour discharge product; cost and fossil-peaker substitution remain commercial targets, not verified installed costs.
- Unresolved risk
- Low round-trip electrical efficiency (~40–45%), large geographic footprint, and managing hydrogen evolution side-reactions during charging.
ESS IncEnergy Warehouse & Energy CenterAll-iron redox flow battery utilizing abundant, non-toxic iron chloride electrolyte with proprietary proton-pump plating chemistry
- Reported evidence
- Commercial units operating across utility and commercial microgrid sites; partnership with Honeywell and Sacramento Municipal Utility District (SMUD).
- Announced next step
- Providing 4- to 12-hour duration storage with zero degradation over 20,000+ cycles and 25-year service lifetimes.
- Unresolved risk
- Electrolyte pump parasitic power losses, iron plating morphology dendritic growth, and system balance-of-plant capital cost.
Energy DomeCO2 BatteryThermodynamic closed-loop cycle compressing and liquefying carbon dioxide using off-peak power, expanding gaseous CO₂ through a turbine during discharge
- Reported evidence
- Commercial 2.5 MW / 4 MWh demonstration plant operating in Sardinia, Italy; secured project financing for multi-hundred-MWh commercial projects globally.
- Announced next step
- Delivering 8- to 24-hour long-duration storage with >75% round-trip efficiency using standard, off-the-shelf industrial turbomachinery.
- Unresolved risk
- Large flexible gasholder dome footprint, thermal insulation losses across multi-day storage intervals, and high-pressure liquid CO₂ storage tanks.
HydrostorAdvanced Compressed Air (A-CAES)Compressing air into unlined deep rock caverns using water compensation for constant pressure, storing thermal heat of compression to reheat air during expansion
- Reported evidence
- Goderich, Ontario demonstration: 1.75 MW discharge and more than 10 MWh; much larger California and Australian projects are in development.
- Announced next step
- Providing 8- to 24+ hour clean capacity competing directly with fossil peakers and pumped storage hydro without surface reservoirs.
- Unresolved risk
- Underground geological cavern availability, mining and cavern excavation capital costs, and permitting lead times.
Highview PowerCRYOBattery (LAES)Liquid Air Energy Storage (LAES) chilling ambient air down to -196°C to store as liquid at low pressure, re-gasifying and expanding through turbines during peak demand
- Reported evidence
- Pilots operated in the UK; secured £300M in commercial funding to construct a 50 MW / 300 MWh commercial plant in Manchester.
- Announced next step
- GWh-scale long-duration grid storage with zero geographic constraints and 30- to 40-year equipment operational life.
- Unresolved risk
- Round-trip efficiency (~50–60%) requiring co-located waste heat/cold integration to reach economic parity with batteries.
AmbriLiquid Metal BatteryHigh-temperature stationary battery cells with liquid calcium-antimony metal electrodes and molten salt electrolyte, operating at ~500°C
- Reported evidence
- Prototype testing preceded a Chapter 11 filing and asset sale in July 2024; commercial scale and lifetime claims remain unverified in an operating fleet.
- Announced next step
- Low-cost 4- to 24-hour stationary storage with zero thermal runaway risk and immune to ambient weather extremes.
- Unresolved risk
- Thermal management maintaining 500°C internal operating temperatures during prolonged standby, and manufacturing scale-up challenges.
Eos Energy EnterprisesZnyth Aqueous Zinc BatteryAqueous zinc-halide stationary battery utilizing non-flammable electrolyte and low-cost earth-abundant materials for 3- to 12-hour discharge
- Reported evidence
- Manufactured and deployed multi-megawatt systems in the US; secured multi-hundred-million-dollar DOE loan guarantee to scale automated manufacturing.
- Announced next step
- Safe, non-lithium stationary storage manufactured domestically with 100% recyclable materials.
- Unresolved risk
- Lower round-trip efficiency (~65–75%) than lithium-ion and zinc dendrite formation requiring automated chemical maintenance cycles.
Long Duration Energy Storage Council / US DOELDES EarthshotGlobal CEO-led industrial council and US Department of Energy initiative targeting a 90% reduction in LDES levelized cost within ten years for 10+ hour systems
- Reported evidence
- Published scenarios estimating up to 8 TW of long-duration capacity by 2040; the figure is modeled demand under assumptions, not installed capacity or a measured requirement.
- Announced next step
- Establishing commercial capacity payments and regulatory market mechanisms recognizing long-duration reliability over short-term power arbitrage.
- Unresolved risk
- Capacity market rules in deregulated grids failing to reward multi-day resiliency, leaving long-duration storage without financing mechanisms.
Round-trip efficiencies (RTE) vary widely across storage technologies (from ~40% for thermal/air to >75% for batteries); duration-matched economics depend on low capital cost per kWh rather than high efficiency.
The optimistic view, with conditions
Storage becomes a portfolio organized by duration
Lithium-ion can dominate frequent short-duration work while pumped, thermal, compressed, chemical, flow, and other systems compete where cheap capacity and long waiting matter more than compactness and peak efficiency.
Lithium-ion keeps winning frequent cycling
Falling pack prices and high utilization make short-duration storage the easiest case to finance today.
The power block starts to matter more
As duration grows, the energy-reservoir share of cost rises and chemistry choice widens.
Readiness value decides bankability
Rare-discharge reserves need contracts that pay for avoided system cost, not just the energy delivered.
What a duration-matched storage portfolio actually needs
- Separated pricingPower capacity and energy capacity quoted and financed as distinct ledgers.
- Duty-cycle specificationDuration, response, availability, and throughput procured together, not assumed.
- Honest degradation accountingCalendar life, cycles, and augmentation priced into lifetime delivered service.
- Readiness revenueMarket products that pay rarely discharged reserves for the capacity they hold.
- Site and interconnection speedPermitting and grid connection that don't erase a cheap component's advantage.
What actual lulls and operating systems tell us
Germany’s network regulator defines a Dunkelflaute for its 2024 price-spike investigation as at least 48 hours when wind and solar together produce below 15% of their installed capacity. It identified the 5–7 November and 11–12 December 2024 events; 34 of 35 hours above €300/MWh occurred in those windows. The same investigation found unavailable conventional plants also mattered. A weather lull is therefore a stress condition, not a direct measure of storage needed: imports, dispatchable generation and demand response change the residual shortage. German regulator, 2024 price-spike investigation.
For scale, a constant 1 GW shortfall over seven days is 168 GWh of delivered electricity before charging losses; a 10 GW shortfall is 1.68 TWh. These are arithmetic examples, not forecasts for Germany or any other grid. A 2026 peer-reviewed European weather and power-system study finds that extreme, correlated events can imply several hundred TWh of long-duration supply in modeled 100% renewable cases. That outcome depends heavily on weather years, transmission, generation mix and allowed firm capacity. System-cost modeling by Sepulveda and colleagues finds firm low-carbon resources can reduce the cost of deeply decarbonized systems. Neither study establishes that every grid should buy that amount of storage.
Short-duration deployment is already observable. The US EIA counted 11.7 GW of battery capacity in CAISO and 8.1 GW in ERCOT at the end of 2024. In California, average battery output during the evening net peak rose from less than 1 GW in May–June 2022 to 4.9 GW in 2025. Those are power figures, not evidence of week-long energy supply. The January 2025 Moss Landing battery fire also shows why siting, fire protection and emergency response remain part of installed cost and reliability. EIA deployment comparison; EIA evening output.
Costs must keep the same boundary
IRENA’s installed-system series fell from $2,571/kWh in 2010 to $192/kWh in 2024. BNEF’s 2025 averages of $108/kWh for lithium-ion packs and $70/kWh for stationary packs are component prices. A $70 pack cannot stand in for a $70 installed reservoir with a funded power block, site and connection. DOE’s 2022 cost assessment compares lithium-ion, flow batteries, pumped hydro, compressed air and hydrogen on stated durations and system boundaries; no single $/kWh quote applies across them. Pumped hydro’s physical density illustrates one limit: at a 100-metre head, 1 m³ of water contains just 0.272 kWh of gravitational energy before turbine losses, so land and elevation become material constraints.
A deliberately simple 168-hour example shows the standby problem. At an assumed installed energy cost of $150/kWh, the energy component alone is $25,200/kW of discharge rating. A 10% annual capital-recovery factor would require $2,520/kW-year before operation, charging losses or the power block. This is an illustrative assumption, not a market quote; its implied cost is why the calculator’s low-cost reservoir input must be validated for an actual project. Rarely discharged storage also cannot recover that carrying cost from energy sales alone. A gas peaker comparison must include fuel, emissions, availability and equivalent reliability service, rather than comparing equipment prices only.
The LDES Council’s multi-terawatt 2040 estimates are scenario outputs, not operating capacity or a procurement requirement. DOE’s Storage Shot targets 5¢/kWh levelized cost for ten or more hours, likewise a target rather than observed plant performance. The near-term empirical case is strong for daily batteries; week-long economics still require demonstrated project costs, utilization and capacity contracts.
Deployment is moving faster than multi-day proof
The IEA reports 108 GW of battery capacity added worldwide in 2025, 40% above 2024, with installed capacity eleven times its 2021 level. This is a power-capacity curve; neither GW nor pack $/kWh specifies delivered energy over a week. The public cost studies cited above use different heads, cavern geology, reservoir sizing and lifetime assumptions for pumped hydro, compressed air, hydrogen and iron-air batteries. Until a project publishes installed cost, duration, cycle life and annual availability together, a single technology-specific $/kWh comparison would imply false precision.
Sources, method, and boundaries
Component prices, installed cost, LCOS, and reliability value retain separate boundaries throughout. The duration equation and interactive model are a screening frame, not a project forecast; bankable analysis requires hourly dispatch, financing, degradation, replacement, and market rules.
- LCOS
- Levelized cost of storage: lifetime cost divided by lifetime delivered energy, analogous to LCOE for generation.
- Round-trip efficiency
- The share of charged energy that is later delivered on discharge, after conversion losses.
- Augmentation
- Adding capacity over a project's life to offset degradation and maintain contracted output.
