Shota Furuya

Pieces日本語版

Is battery technology good enough yet? What are the limits on cost and capacity?

The answer to “renewables are unstable because they depend on the weather” comes down, in the end, to “smooth it out with batteries”. So is battery technology genuinely ready? Is it too expensive, and is there enough capacity? This piece works through the question with current data.

The short answer. First, battery costs hit a record low in 2025. The levelised cost of four-hour storage fell to US$78/MWh (about ¥12,500/MWh), 27% below the previous year (BloombergNEF, 2026). Second, annual deployment reached 112 GW and 307 GWh in 2025, entering the 100-gigawatt era for the first time (Kikuma, 2026). Third, while smoothing variation over a few hours is now in practical use, storage over days and across seasons still faces problems of both technology and cost. Whether batteries are “good enough” depends on what you are asking them to do.

Cost: a long decline and a record in 2025

The economics have changed dramatically in a little over a decade. Lithium-ion pack prices fell from more than US$1,000/kWh (over ¥160,000/kWh) in 2010 to about US$140/kWh (about ¥22,400/kWh) in 2020 and 2023, and are projected at US$60–90/kWh (about ¥9,600–14,400/kWh) by 2030. That follows a learning curve — costs falling 17–21% with every doubling of deployment — the same structure solar panels moved along.

As a cost measure, the levelised cost of four-hour storage in 2025 was US$78/MWh (about ¥12,500/MWh), the lowest since BNEF began tracking in 2009 (Nalini, 2026). What stands out is the contrast: in the same year solar rose 6%, onshore wind 2%, offshore wind 12% and gas generation 16%, and only storage fell sharply. While supply constraints and higher interest rates pushed other costs up, competition between manufacturers and better system design kept driving storage down.

Why so cheap? Three main reasons. Fierce price competition among Chinese, Korean and American manufacturers. Continuous improvement in system design — thermal management, integrated inverters, control software. And the scale economics of LFP (lithium iron phosphate), which is safe and uses no cobalt. Together they have made storage the fastest cost decliner after solar. BNEF expects a further 25% reduction by 2035.

The symbolic figure is solar plus storage: 87 GW entered service worldwide in 2025 at a cost of supply of US$57/MWh (about ¥9,100/MWh), about half the US$102/MWh (about ¥16,300/MWh) of new gas generation. In California and Texas, developers increasingly choose solar and batteries over gas turbines to serve surging data centre demand. Storage is no longer an expensive accessory but generation that competes with thermal plants directly. BNEF calls the shift storage-led system balancing: batteries quietly taking over the evening peak that gas has covered.

Short duration is ready; long duration is the next problem

The issue is moving from cost to duration. Today’s mainstream four-hour systems charge on surplus midday solar and discharge into the evening peak. That is highly effective against the duck curve — the dip in net demand during the day and the sharp evening ramp — and against within-day variation generally. IRENA estimates that in regions with good resources, solar plus four-hour storage can deliver 95% reliability at US$54–82/MWh (about ¥8,600–13,100/MWh) (IRENA, 2026). Daily variation is approaching the point of being handled economically.

The difficulty is a Dunkelflaute — the German term for a dark, windless spell, when neither sun nor wind delivers for days. Filling a shortfall of several days or a week with four-hour batteries takes a quantity of storage that rises non-linearly in cost: it makes no economic sense to leave an enormous battery idle for an event that happens a few times a year. This is where long-duration energy storage (LDES), discharging for six hours or more, comes in.

LDES covers many approaches: flow batteries using vanadium and other chemistries, compressed air stored underground, gravity systems that raise and lower weights, thermal storage in rock or molten salt. Form Energy’s iron-air battery in the United States claims 100 hours of continuous discharge. Most use no lithium and specialise in discharging slowly over long periods with cheap materials. BNEF expects annual LDES deployment to quadruple to 2 GW in 2026 — still small in absolute terms, with wide deployment ahead. Spreading solar and wind geographically to average out weather risk, connecting regions with transmission, and shifting demand all have to be combined to cover long shortfalls. Batteries do not solve everything on their own; they are one instrument supporting the system.

Diversifying chemistry: sodium-ion and solid state

What is inside the battery is diversifying too. LFP accounts for more than 90% of stationary deployment in 2025 on safety and cost, but 2026 is seen as a turning point.

The leading alternative is sodium-ion. Replacing lithium with sodium, which comes from sea salt and is about a thousand times more abundant on Earth, it gives up some energy density (120–160 Wh/kg) in exchange for good low-temperature performance, and could fall to US$40–60/kWh (about ¥6,400–9,600/kWh) by 2030 — below LFP. Mass production for stationary use has begun, with CATL signing a 60 GWh supply agreement with Beijing HyperStrong in 2026. Because it reduces dependence on lithium-producing countries such as Australia and Chile, it also lowers geopolitical exposure.

Solid-state batteries, on which Japan pins its hopes, replace the liquid electrolyte with a solid for safety and energy density. Toyota and others aim at mass production between 2027 and 2030, but yields on large cells are low, and costs are expected to run 1.5–2 times lithium-ion even in 2030. Promising for EV range, they are unlikely to displace cheaper LFP and sodium-ion in stationary storage for some time.

Where the limits on capacity and materials lie

“Will the materials run out?” is a common worry. Known lithium reserves are about 22 million tonnes, and some estimates find that with 100% recycling, world EV demand beyond 2050 could be met from them. China is already building arrangements that recover 96.5% of lithium and 99.6% of cobalt, nickel and other metals from used cells, so the resource constraint is shifting from how much is mined to how used batteries are returned into a formal collection system — a question of institutional design.

On capacity, deployment is faster than for any generation technology before it. Going from 10 GW to 100 GW a year took about eight years for solar and fifteen for wind; batteries did it in four (Kikuma, 2026). China accounts for 54% of the world total and the United States 16%, while Australia grew about sixfold in a single year. Storage is demonstrating the fastest S-curve of any technology.

Japan: market design as the central problem

Japan’s situation has its own features. The country holds about 28 GW of pumped hydro, among the largest long-duration storage assets in the world. That should make it a leader in long-duration storage; what is less clear is the direction of its investment in new batteries.

Grid-scale batteries are being deployed faster as solar grows. But their revenue depends on arbitrage — charging when prices are low and discharging when they are high — which requires a market with large enough price spreads. India, having built the rules, saw grid batteries deployed in 2025 achieve internal rates of return as high as 24%. In Japan, the design of the capacity and balancing markets is still developing, and entry has run ahead of the spreads available, with warnings of a bubble. As in the early years of large-scale solar, entrants without careful revenue modelling may find themselves unable to cover their costs. The economics of storage depend less on the technology than on how well the market around it is designed.

Objections and limits

In fairness, the limits. First, a battery is not a generator: about 15% is lost in each cycle (round-trip efficiency is typically about 85%). It moves electricity rather than making it, so renewable generation has to be there in the first place. Second, as above, long-duration and seasonal storage remain commercially immature, and it is premature to assume that batteries alone make 100% renewables easy. Third, the concentration of the supply chain in China lowers costs while creating economic security risk. Storage is maturing quickly, but not everything is settled. The accurate description is: within-day balancing is in practical use; long-duration storage and market design are still ahead.

Summary

  • Costs hit a record low in 2025: four-hour storage at US$78/MWh (about ¥12,500/MWh), 27% below the previous year — the exception in a year when other technologies rose.
  • Solar plus storage at US$57/MWh (about ¥9,100/MWh) is about half the US$102/MWh of new gas generation. It competes with thermal directly.
  • Annual deployment passed 100 GW in 2025. Going from 10 GW to 100 GW took batteries four years, solar eight and wind fifteen — the fastest ever.
  • Variation over hours is close to solved. What remains is long-duration storage over days and seasons.
  • Chemistry is diversifying, with sodium-ion emerging as the option with less resource risk.
  • The resource question is shifting from mining to recycling systems. In Japan, the binding constraint is less capacity than the lag in market design.

Asked whether battery technology is good enough, the honest answer depends on the application. For smoothing variation over hours, cost and performance are in practical use, and under the right conditions cheaper than thermal generation. For storage over days and seasons, promising technologies are lining up but deployment is only beginning. Japan’s constraint is not only the performance of the batteries but the design of the market that would draw out their value. The technology is advancing quickly. The question is not whether to wait, but what institutions will let us use what is already arriving.

A note on currency conversion

Yen figures use an approximate ¥160 to the dollar, based on the central rate of ¥160.12 in the Bank of Japan’s foreign exchange quotations for 15 June 2026.

References and data sources

Costs and markets

Technology, capacity and long-duration storage

  • International Renewable Energy Agency. (2026). 24/7 renewables: Firming solar and wind.
  • Ember. (2025). The age of storage. https://ember-climate.org/

Lithium-ion and next-generation batteries

  • BloombergNEF. (2024). Lithium-ion battery price survey.
  • World Intellectual Property Organization. (2025). WIPO technology trends: The future of transportation.

Japan and its institutions

Exchange rate

Bank of Japan. (2026, June 15). Foreign exchange quotations (15 June) [in Japanese]. https://www.boj.or.jp/statistics/market/forex/fxdaily/fxlist/fx260615.pdf

← Back to list