What does a reliable electricity system built on wind and solar actually need?

Why the question matters

The reliability problem in a variable-renewable system operates on two timescales at once. Bridging hours and bridging seasons are different engineering problems, and a technology sized for one is not automatically suited to the other.

Which storage technology deserves investment is decided by system-level modelling rather than by comparing costs per kilowatt-hour, because the value of a technology depends on the role it ends up playing in a least-cost system.

How many years of weather a model is optimized over changes its answer, since the rare extended shortfalls that drive long-duration storage requirements appear only in longer records.

What our research finds

• With gas at $50/MWh, concentrated solar thermal with molten salt storage can supply more than half of industrial process heat demand at a saving in most regions below 45° latitude (Wongel et al., 2026).
• For systems planned on a short weather record and then operated across decades of real variability, coordinated spending across both additional storage and additional generation improved resource adequacy more than spending the same amount on either category alone (Reich et al., 2026).
• Across 27 European countries, hydrogen made from surplus wind and solar electricity could substitute 30% of fossil-derived hydrogen — 1.9 million tonnes a year — cutting ammonia and refinery emissions by 18%, or 20 million tonnes of CO2 a year (Ganter et al., 2025).
• Long-duration storage sized for seasonal needs already has enough energy and power capacity to meet short-duration needs, so adding batteries as a second technology barely reduces system cost in any region examined (Li et al., 2024).
• The marginal cost of buying reliability with more weather data rises steeply — one extra planning year halves lost load, but beyond one or two years it exceeds a $10/kWh value of lost load in systems without long-duration storage (Ruggles et al., 2024).
• Across 243 combinations of published 2050 cost projections, which technology dominates a least-cost zero-carbon system cannot be predicted with confidence, and cutting one technology's cost does not reliably increase how much of it gets built (Duan and Caldeira, 2024).
• A wind, solar, battery and hydrogen system meeting every hour of Oahu's demand across 14 years of weather would cost $0.1673/kWh, below the $0.2126 to $0.2987 its petroleum-fired system actually cost in 2022-23, and would use less land than a battery-only system (Covelli et al., 2024).
• In wind and solar systems with abundant curtailed power, system cost is sensitive to the capital cost of hydrogen storage's discharge component and barely at all to round-trip efficiency — efficiency only starts to matter once capital costs have fallen (Dowling et al., 2024).
• Sequentially removing whichever firming technology is most valuable shows that reliable wind-and-solar systems do not depend on the feasibility of any particular one (Wongel and Caldeira, 2023).
• Lowering advanced nuclear capital cost from US$6,317 to US$4,000 per kilowatt-electric cuts the cost of a fully decarbonized system by 15 to 25 per cent in all 42 countries modelled, and thermal storage lets firm generation replace most of the storage such a system would otherwise need (Duan et al., 2022).
• Long-duration storage lowers total system cost relative to wind-solar-battery systems, and system cost is twice as sensitive to reductions in long-duration storage cost as to reductions in battery cost (Dowling et al., 2020).
• In least-cost systems batteries serve intra-day needs while long-duration storage handles inter-seasonal and multi-year variation, and dependence on the latter grows as more years of weather are modelled (Dowling et al., 2020).
• Storage costs would need to fall several hundred-fold, to about $1/kWh, for fully variable-renewable systems to be highly reliable without extensive curtailment (Tong et al., 2020).