Why Grid Storage Is Hard
Photo: Ömer Gülen / Pexels
Electric grids have to match supply to demand within seconds, every hour of every day, and unlike oil or coal, electricity cannot simply be piled up until needed. As wind and solar, both inherently variable, take a growing share of generation, utilities need somewhere to park surplus power and release it later. That "somewhere" is grid storage, and building it economically at the scale a national grid requires has turned out to be one of the hardest unsolved problems in the energy transition.
Why electricity storage is fundamentally different from storing other resources
Storing oil, coal, or natural gas is straightforward. Fuel sits in tanks or piles until needed, and its stored chemical energy does not meaningfully degrade over time. Electricity itself cannot be stockpiled the same way; to "store" it, the grid has to convert it into some other form of energy, chemical energy in a battery, gravitational potential energy in a raised reservoir, compressed air in an underground cavern, and then convert it back to electricity later. Every conversion step loses energy as heat, friction, or resistance, so storage is never free.
Round-trip efficiency — the percentage of energy that comes back out of a storage system compared to what went in — is the metric that separates good grid storage options from bad ones. Lithium-ion batteries typically return roughly 85 to 95 percent of the electricity put into them, one of the best round-trip efficiencies of any storage technology, according to the U.S. Department of Energy's Storage Innovations 2030 roadmap. That efficiency is a major reason batteries have become the default choice for short-duration grid storage, even though they carry other tradeoffs.
Pumped hydro, the oldest large-scale storage technology, works differently. It pumps water uphill into a reservoir when electricity is abundant and releases it downhill through turbines when demand rises, converting electrical energy into gravitational potential energy and back. It is mechanically simple and durable, but efficiency losses at the pump and turbine stages, along with evaporation from open reservoirs, mean round-trip efficiency for pumped hydro sits somewhat below the best lithium-ion systems.
Compressed air energy storage, which pumps air into underground caverns, typically salt formations, at high pressure and releases it later through a turbine, trades an even larger efficiency hit for the ability to use existing geological formations instead of building new reservoirs. Each of these conversion detours costs energy, which is the core reason grid storage is a genuinely harder engineering problem than storing fuel.
The mismatch problem: renewables don't generate on demand
Solar panels only generate electricity during daylight hours, and even then output varies with cloud cover and season. Wind turbines generate power whenever wind conditions allow, which does not necessarily align with when people actually need electricity. Evenings are the clearest example: solar output has already dropped to zero by the time many households hit their peak demand for lighting, cooking, and air conditioning or heating.
Grid operators call this daily pattern the duck curve — a chart of net electricity demand across a day that dips in the afternoon, as solar output floods the grid, then rises sharply after sunset, tracing a shape that resembles a duck's body and neck. California's grid operator, CAISO, popularized the term to describe exactly this challenge on its own solar-heavy grid. The steeper that evening ramp gets, the more storage or flexible generation the grid needs to fill it without asking fossil fuel plants to spin up on short notice.
Pumped hydro storage still holds by far the most stored energy of any grid storage technology worldwide, precisely because it can store enormous amounts of energy relatively cheaply per unit once built. But it requires specific geography, an elevation change of at least a few hundred meters and a reliable water source, that is not available everywhere, and building new reservoirs today runs into land use, permitting, and environmental review timelines that can stretch a decade or longer.
Why can't batteries just handle all of this the way they handle short-term storage?
Battery cost scales directly with total energy capacity, so batteries excel at daily smoothing tasks, discharging in the evening what solar generated at midday, but scaling that same approach to cover multi-day lulls in wind and solar generation would require an enormous, currently cost-prohibitive amount of battery capacity. A four-hour battery large enough to smooth one evening peak is a very different, much smaller machine than one built to ride out several cloudy, windless days in a row.
How grid batteries actually work at scale, and where they run into limits
A grid-scale lithium-ion battery installation is, at its core, thousands to hundreds of thousands of individual cells wired together into racks, housed in shipping-container-sized enclosures, and managed by software that decides when to charge and discharge based on grid signals and price. The technology is the same chemistry found in electric vehicles and laptops, just deployed at a much larger physical footprint.
The clearest real-world proof that batteries can stabilize a renewable-heavy grid is the Hornsdale Power Reserve in South Australia. Developed by Neoen in partnership with Tesla and completed in November 2017 at 100 megawatts and 129 megawatt-hours, then expanded in September 2020 to a total of 150 megawatts and 194 megawatt-hours, Hornsdale responds to grid disturbances far faster than conventional thermal plants can, and its 2020 expansion added inverter-based inertia support services (Hornsdale Power Reserve). In its first two years of operation, the installation was credited with saving South Australian electricity consumers more than $150 million by reducing the market's reliance on expensive, slow-to-respond frequency control services (Hornsdale Power Reserve; Clean Energy Finance Corporation case study).
Battery costs have also fallen sharply enough to make projects like Hornsdale increasingly common rather than exceptional. Global lithium-ion battery pack prices fell to a record low of $108 per kilowatt-hour in 2025, and prices for stationary storage applications specifically, the category grid batteries fall into, dropped to roughly $70 per kilowatt-hour, 45 percent lower than the year before, according to BloombergNEF's survey published 9 December 2025. For how the cells themselves store charge, see how lithium-ion batteries work. That decline is a major reason four-hour grid batteries have become one of the fastest-growing categories of new power generation capacity added to the U.S. grid in recent years.
Scale is not the only lesson from real deployments, though. In January 2025, a fire broke out at Vistra Energy's Moss Landing battery storage facility in Monterey County, California, one of the largest lithium-ion grid battery installations in the world at roughly 300 megawatts in the affected building, ultimately damaging about 55 percent of that building's roughly 100,000 battery modules (EPA, 2026). The fire began on 16 January 2025, burned until 18 January and flared up again on 18 February; EPA announced the final cleanup phase on 8 September 2026. Investigators have pointed to the facility's indoor, converted-power-plant layout and an apparent compromise to its fire suppression system as likely contributing factors, rather than a fundamental flaw in lithium-ion chemistry itself, and the incident is still under formal investigation. The episode is a genuine caution about siting, ventilation, and fire suppression design for indoor battery installations, not proof that grid batteries are inherently unsafe, but it is a fair reason for skepticism about any claim that grid-scale battery storage is a fully solved, risk-free technology.
The seasonal storage problem nobody has fully solved
Daily storage, smoothing supply and demand within a 24-hour cycle, is a largely solved engineering problem using batteries and pumped hydro. Seasonal storage, storing summer solar surplus for use in winter months, or storing wind surplus from a windy season for use during a calmer one, remains a much harder unsolved problem at meaningful scale, since no current technology stores energy cheaply enough over months at the volumes a full national grid would require.
Some regions have explored hydrogen produced from surplus renewable electricity, often called green hydrogen when made through electrolysis powered by renewable sources, as a form of long-duration or seasonal storage. Hydrogen can be stored for extended periods in tanks or underground caverns and converted back to electricity later through fuel cells or turbines. But the round-trip efficiency losses at each conversion step, electricity to hydrogen through electrolysis, then hydrogen back to electricity, are substantial, often leaving only 30 to 40 percent of the original electricity recoverable, and this approach remains at an early, largely pilot-scale stage rather than proven at full grid scale. For readers who want the mechanics of that first conversion step, how green hydrogen is made walks through the electrolysis process and where its own efficiency losses come from.
A useful analogy: daily battery storage is like a checking account, money goes in during the day and comes back out that evening, with a small transaction fee along the way. Seasonal storage is more like trying to save a summer's harvest to eat all winter without refrigeration; the "food" (energy) degrades, spoils, or costs enormous resources to preserve, and nobody has yet built a cheap, large enough pantry for an entire grid's worth of energy.
Why grid storage is a genuine bottleneck for the energy transition
Grid operators cite storage limitations as a key reason renewable-heavy grids still sometimes rely on fossil fuel peaker plants — natural gas plants built specifically to ramp up quickly and cover gaps when wind and solar output drops unexpectedly, rather than to run continuously like baseload generation. The more variable renewable generation a grid adds, the more valuable flexible, fast-responding storage becomes for maintaining reliability without falling back on fossil fuel backup capacity.
This is why the Department of Energy and international energy agencies have identified long-duration energy storage specifically, not just any storage, as one of the most consequential unsolved technology gaps standing between current grids and a largely renewable electricity system. That distinction matters: short-duration battery storage, the kind that shifts a few hours of solar output into the evening, has already seen substantial cost declines and rapid deployment growth, as the Hornsdale and Moss Landing examples show. Storage that needs to bridge multiple cloudy, windless days, let alone entire seasons, has not.
Other storage technologies sit somewhere between these two extremes on cost, maturity, and duration. Flow batteries, which store energy in liquid electrolyte tanks rather than solid cells, can be scaled to longer durations by simply building bigger tanks, decoupling power capacity from energy capacity in a way lithium-ion cannot easily match, though the technology remains less mature and more expensive per unit than lithium-ion at current deployment volumes.
Comparing the major grid storage technologies
No single storage technology wins on every metric that matters to a grid operator: how much energy it can hold, how long it can discharge, how mature and bankable the technology is, and where it can physically be built. The table below lays out how the leading options stack up against each other.
| Technology | Typical Duration | Round-Trip Efficiency | Maturity | Key Constraint |
|---|---|---|---|---|
| Lithium-ion battery | 1–4 hours (up to ~8 in newer projects) | 85–95% (NREL models 85%) | Commercially mature, fastest-growing segment | Cost scales directly with total energy capacity; multi-day storage is currently cost-prohibitive |
| Pumped hydro | 6–20+ hours | ~70–85% | Oldest, most-deployed technology (largest installed capacity worldwide) | Requires specific elevation change and water availability; new sites face long permitting timelines |
| Compressed air energy storage | 4–24+ hours | ~40–70% depending on design | Early commercial deployment, few operating projects | Requires suitable underground geology, typically salt caverns |
| Flow batteries | 4–12+ hours | ~65–80% | Early-stage commercial, limited deployment history | Energy capacity scales with tank size, not cell count, but current cost per kilowatt-hour exceeds lithium-ion |
| Green hydrogen | Days to seasonal | ~30–40% (electricity-to-electricity) | Pilot-scale, not proven at full grid scale | Large efficiency losses across electrolysis and reconversion steps |
Lithium-ion dominates short-duration grid storage because of its efficiency and falling cost, while pumped hydro still holds the largest total installed capacity worldwide; nothing on this list yet solves storage measured in months rather than hours.
What people get wrong about grid storage
A common misconception is that building enough batteries will eventually solve every part of the storage problem, the same way batteries solved the daily solar-to-evening shift. That assumption breaks down once the question shifts from hours to weeks. Battery cost is roughly proportional to how much total energy the system needs to hold, so a battery bank sized to survive a multi-day windless, cloudy stretch, sometimes called a "dunkelflaute" in the case of prolonged calm, overcast weather, would need to be dramatically larger and more expensive than anything deployed today, purely because of how the cost math scales.
A second misconception is that any single technology will "win" and replace the others. In practice, grid operators increasingly deploy a mix: lithium-ion batteries for fast, short-duration balancing, pumped hydro where geography allows for larger, longer-duration capacity, and gas peaker plants as backup for the rare multi-day gaps that neither can economically cover yet. Treating grid storage as a single problem with a single solution misses how differently each technology performs across duration, cost, and geography.
A third misconception, reinforced by incidents like the Moss Landing fire, is that grid batteries are either perfectly safe or fundamentally dangerous. Neither framing holds up. Indoor, densely packed installations built inside older infrastructure carry different risk profiles than newer, purpose-built outdoor container systems with modern fire suppression and cell-level monitoring, and regulators and utilities have already begun tightening siting and safety standards in response to real incidents rather than treating the technology as static.
How much storage is actually being built
The growth numbers for short-duration batteries are steep. The U.S. Energy Information Administration reported that 10.3 GW of utility-scale battery capacity was added to the U.S. grid in 2024, and developers planned a record 18.2 GW for 2025 (EIA, 24 February 2025). By mid-2025, about 5.9 GW had come online in the first half of the year, roughly 26 percent of all new U.S. capacity and second only to solar, with about half in Arizona and California (EIA, 20 August 2025). Texas developers alone expected 7.0 GW of batteries in 2025.
Globally, the picture depends on which unit you use. Measured in power (gigawatts), batteries have been catching pumped hydro quickly. Measured in energy (gigawatt-hours), pumped hydro is still far ahead, because a typical battery discharges for one to four hours while a pumped hydro plant often runs for many hours. The International Hydropower Association estimated about 161 GW of pumped storage worldwide holding up to roughly 9,000 GWh of energy, more than 94 percent of installed storage at the time of its estimate (IHA). The IEA's 2021 forecast likewise expected global storage capability to reach almost 12 TWh by 2026 with pumped hydro providing nearly all of it (IEA, 1 December 2021). The countries that lead on that front largely mirror the ranking of countries with the most hydropower.
| Metric | Figure | Source and date |
|---|---|---|
| U.S. utility-scale battery additions, 2024 | 10.3 GW | EIA, Feb 2025 |
| U.S. planned battery additions, 2025 | 18.2 GW | EIA, Feb 2025 |
| U.S. battery additions, first half of 2025 | 5.9 GW (~26% of new capacity) | EIA, Aug 2025 |
| Stationary storage battery pack price, 2025 | ~$70/kWh (down 45% year on year) | BloombergNEF, Dec 2025 |
| Global pumped storage, power and energy | ~161 GW; up to ~9,000 GWh | IHA estimate |
| Global storage capability forecast, 2026 | almost 12 TWh, mostly pumped hydro | IEA, Dec 2021 |
Power (GW) and energy (GWh) measure different things: power is how fast a system can deliver; energy is how long it can keep delivering.
A worked example: why days of storage break the budget
The cost problem becomes concrete with a simple sizing exercise. Take a hypothetical region whose average demand is 10 GW. Covering one four-hour evening peak at that level needs about 40 GWh of storage. Riding out a five-day windless, overcast stretch with no other supply would need 10 GW × 120 hours = 1,200 GWh — thirty times more energy capacity.
At the BloombergNEF 2025 stationary pack price of about $70 per kWh, the four-hour battery's packs cost roughly $2.8 billion. The five-day version's packs alone would cost about $84 billion, before inverters, installation, land, grid connection and replacement over time — and most of that capacity would sit idle for most of the year, earning nothing. That is the heart of the issue: a battery's revenue comes from cycling often, while long-duration storage is valuable precisely because it is used rarely.
This is why system planners talk about cost per kWh of capacity for long-duration technologies, not efficiency. A technology with poor round-trip efficiency can still win the multi-day role if its energy capacity is very cheap — which is the logic behind hydrogen in salt caverns, compressed air, and low-cost chemistries such as iron-based batteries now in early commercial deployment.
What storage actually does on a grid
Grid storage is not one service but several, and each pays differently:
- Frequency response. Grids must hold frequency close to 50 or 60 Hz. Batteries can inject or absorb power in fractions of a second, which is where Hornsdale earned much of its early value.
- Energy arbitrage. Charging when power is cheap, often midday solar hours, and discharging when it is expensive in the evening.
- Capacity and peak shaving. Guaranteeing output during the highest-demand hours so utilities can avoid building new peaker plants.
- Transmission relief. Storing power near congested lines so that new wires are needed later or not at all.
- Backup and black start. Restarting parts of the grid after an outage.
A single battery can stack several of these revenue streams, which is a large part of why short-duration storage is now bankable. A seasonal store, used a few times a year, cannot. Demand-side flexibility competes in the same space: managed electric vehicle charging, discussed in our article on how EVs affect the power grid, can shift load in a similar way without any dedicated storage hardware.
Battery fleets also degrade with use and time, which affects the economics of every one of these services. Our explainer on why lithium-ion batteries age covers the chemistry behind capacity fade, and why operators often oversize systems at installation to meet a guaranteed capacity years later.
Open questions
Several questions will decide how quickly the long-duration gap closes. Can cheap chemistries prove their cycle life and safety at scale? Will market rules pay for rarely used, multi-day capacity, the way capacity markets pay for standby generators? How much of the problem can be solved by wider grids that move power across weather systems, overbuilt wind and solar, and flexible demand, rather than by storage alone? For background on why the underlying generation is variable in the first place, see renewable energy explained, and for the history of how today's centrally balanced grids came to be, see the invention of the electric power grid.
The takeaway
Grid storage is hard because electricity cannot simply be piled up like coal or oil; it must be converted into another energy form and converted back, and every available conversion technology, batteries, pumped hydro, compressed air, hydrogen, has a different sweet spot and a different hard limit on cost, duration, and geography. Batteries have largely solved short-duration, daily storage, as real projects like the Hornsdale Power Reserve and Vistra's Moss Landing facility demonstrate at gigawatt-hour scale, even as Moss Landing's 2025 fire is a reminder that deployment details still matter. The harder problem, storing energy cheaply across days, weeks, or entire seasons, remains one of the most significant unresolved technical bottlenecks standing between today's grid and one running primarily on variable renewable power.
Read next:
FAQ
Q: Why can't electricity just be stored the way coal or oil is stored? A: Electricity has to be converted into another energy form, chemical energy in a battery or gravitational potential energy in pumped hydro, and then converted back later, and every conversion step loses some energy, unlike coal or oil, which sit stably in storage without needing conversion.
Q: Why are batteries good for daily storage but not for weeks-long storage? A: Battery cost scales directly with total energy capacity, so while batteries handle short-duration tasks like shifting midday solar output to evening demand efficiently, scaling that same approach to cover multi-day or weeks-long gaps in renewable generation would require a currently cost-prohibitive amount of battery capacity.
Q: What is pumped hydro storage and why is it still the largest grid storage technology? A: Pumped hydro pumps water uphill into a reservoir when electricity is abundant and releases it downhill through turbines to generate power later, and it still holds most of the world's stored grid energy, measured in gigawatt-hours, because it can store large amounts of energy relatively cheaply, even as batteries catch up in gigawatts, though it requires specific geography not available everywhere.
Q: Has anyone solved seasonal energy storage, like storing summer solar for winter use? A: Not at meaningful scale yet. Some regions are exploring hydrogen produced from surplus renewable electricity as a seasonal storage option, but round-trip efficiency losses of roughly 60 to 70 percent are substantial, and the approach remains largely at an early, pilot-scale stage rather than proven at full grid scale.
Q: Why do renewable-heavy grids still sometimes rely on natural gas plants? A: Grid operators use fast-ramping natural gas "peaker plants" to cover gaps when wind and solar output drops unexpectedly and storage cannot fill the shortfall, which is why long-duration storage is considered one of the key remaining technology gaps for a largely renewable grid.
Q: Are grid-scale lithium-ion batteries safe? A: Grid batteries operate safely at scale in most deployments, including large installations like the Hornsdale Power Reserve, but the January 2025 fire at Vistra's Moss Landing facility in California showed that older indoor layouts and compromised fire suppression systems can create real risk, prompting tighter siting and safety standards for newer projects.
Q: What is the duck curve and why does it matter for grid storage? A: The duck curve is a chart of daily net electricity demand on solar-heavy grids that dips sharply in the afternoon as solar output peaks and then rises steeply after sunset, a shape CAISO popularized to describe California's grid; the steeper that evening ramp, the more storage or flexible generation is needed to fill it without relying on fossil fuel plants.
Sources
- U.S. Energy Information Administration, "Solar, battery storage to lead new U.S. generating capacity additions in 2025," 24 February 2025. https://www.eia.gov/todayinenergy/detail.php?id=64586
- U.S. Energy Information Administration, Today in Energy, 20 August 2025. https://www.eia.gov/todayinenergy/detail.php?id=65964
- International Energy Agency, "How rapidly will the global electricity storage market grow by 2026?", 1 December 2021. https://www.iea.org/articles/how-rapidly-will-the-global-electricity-storage-market-grow-by-2026
- U.S. Department of Energy, Long Duration Storage Shot announcement, 14 July 2021. https://www.energy.gov/articles/secretary-granholm-announces-new-goal-cut-costs-long-duration-energy-storage-90-percent
- BloombergNEF, lithium-ion battery pack price survey, 9 December 2025. https://about.bnef.com/insights/clean-transport/lithium-ion-battery-pack-prices-fall-to-108-per-kilowatt-hour
- International Hydropower Association, "The world's water battery." https://www.hydropower.org/publications/the-world-e2-80-99s-water-battery-pumped-hydropower-storage-and-the-clean-energy-transition
- U.S. EPA, Moss Landing cleanup final phase release, September 2026. https://www.epa.gov/newsreleases/epa-begins-final-phase-post-fire-battery-cleanup-moss-landing
- Hornsdale Power Reserve project site. https://hornsdalepowerreserve.com.au/
- NREL Annual Technology Baseline, utility-scale battery storage (85% representative round-trip efficiency assumption).
Follow new articles
Add SciHub101 to your feed reader to see new science and technology stories.
About the Author
SciHub101 is an independent science and technology publication founded and edited by Buğra Sözeri, a software engineer and app developer based in Türkiye. AI tools assist with drafting and research. Source documentation in the existing archive is uneven, and articles are being reviewed and corrected. The Editorial Team byline identifies editorial responsibility; it does not certify specialist or independent human review. Where we compare products, anything built by our own team is labelled as such — see the editorial guidelines and About page for how corrections and disclosures work.
Related Articles
Countries That Consume the Most Electricity (Ranked)
China used about 10,070 TWh of electricity in 2024, roughly a third of the world's total, followed by the United States and India. This ranking uses Ember data for 2024, adds per-person figures, and explains what drives the gaps.
Renewable Energy Explained — Solar, Wind, and Beyond
How solar panels and wind turbines work, why their costs fell about 90%, how fast they are spreading, and what storage and grids must still solve.
Countries That Run on the Most Renewable Electricity (Ember Data)
A handful of countries already generate essentially all of their electricity from hydro, wind, solar, geothermal and bioenergy. This ranking uses Ember's generation data, with the year for every row, and explains why most leaders are hydro countries and what that means for everyone else.
Countries with the Most Wind Power Capacity (2025 Ranked)
China has about 641 GW of installed wind capacity, nearly half of the world's roughly 1,291 GW. The United States, Germany, India, and Brazil follow. Denmark ranks 17th by capacity but gets about 59% of its electricity from wind. Here is the 2025 ranking, what it measures, and what it leaves out.