Grid-Scale Battery Storage A Beginner's Q&A

It is a large battery installation connected directly to a utility's transmission network (the grid). Its role is to absorb the swings in solar and wind output and hold the frequency steady — to support the stability of the grid.

It does three things.

• Charges during the hours when electricity is in surplus
• Waits in a state where it can respond at any time
• Responds automatically within 10 seconds when the frequency is disturbed

This may come as a surprise, but the core of the earnings is not selling electricity. It is payment for staying on standby, ready to respond at any moment. Q3 covers this in detail.

The size that has become mainstream is a 2 MW high-voltage-connected system, and the site is as a rule around 1,000 sq m. Concretely, it is a single installation with an output of about 2,000 kW that can store about 8,000 kWh. These are two different units: 2,000 kW is the power it can deliver at one time, and 8,000 kWh is the amount it can hold. Do not mix them up.

You do not need an engineer stationed on site. At this scale, the rules allow the qualified electrical professional required by law to be outsourced to an outside party. The cooling fans do keep running, though, so there is noise (Q19).

Note that a system of this size does not count as a "power generation business operator" under the Electricity Business Act. You become one when the output is 1,000 kW or more and the combined maximum connection capacity for supplying retail electricity providers and the like exceeds 10,000 kW (10 MW). It is also far below the line where heavy procedures such as construction plan notifications begin (output of 10,000 kW or more, or capacity of 80,000 kWh or more). Notification to the fire department, on the other hand, is required. It applies to battery installations with a capacity of more than 20 kWh, so an 8,000 kWh-class system is obviously included.

The biggest difference is where the uncertainty sits.

With solar, the selling price was fixed for 20 years. The only thing that moved was output, and that could be estimated as a probability from solar radiation statistics. What the banks were checking, when you get down to it, was solar radiation.

Battery storage is the opposite. You can decide how much to run it, within limits, but the price is not fixed. Most of the revenue comes down to the market on any given day. The uncertainty does not disappear; it just moves from volume to price.

One more important point. Battery storage is not a replacement for solar. It is equipment that became necessary precisely because solar grew. The "duck curve," in which midday market prices fall to almost zero and then jump in the evening, has become the norm, and that price gap became the source of one of the three revenue streams (price-spread trading on JEPX). The gap between the highest and lowest price in a day has widened fivefold, from about 4 yen/kWh around 2020 to an average of about 20 yen/kWh in 2024.

If you carry your solar instincts straight over, there are places where you will trip. The most typical one is the grid connection study response. With solar, the response was effectively a confirmation that connection was settled; with battery storage it is not. Another contractual process follows after the response.

There is no scheme like the 20-year fixed purchase price, but there is a separate subsidy program for installation costs (Q12).

There are three places to sell, each with a different buyer and a different mechanism.

In practice, market participation, trade execution, and settlement are usually handled through an agent known as an aggregator. There is also an arrangement in which the owner becomes a trading member and outsources only the day-to-day operation.

The second one is the core. Because the amount of electricity consumed and the amount produced have to match at all times or the lights go out, money is paid for standing by as the adjuster. You receive it even if you never actually charge or discharge. Batteries, which respond fast, are structurally advantaged in this market. In the first half of fiscal 2025, primary frequency control paid 9.6 to 13.5 yen for batteries against 2.1 to 3.4 yen for thermal (yen/ΔkW per 30 minutes), a gap of three to six times. These results, however, come from a period when the price cap was 19.51 yen. The cap fell to 15 yen in March 2026, and it becomes 10 yen for delivery from September 1, 2026.

You cannot, though, earn from all three at once. It is a question of how you allocate one unit's capacity across the 48 blocks of a day (30 minutes each). Since March 13, 2026, the order has been to bid into JEPX first and offer whatever is left into the balancing market. And since you cannot offer into the market while charging, you cannot put all 48 blocks up for sale either.

The third one, the capacity market, pays for existing and standing by, regardless of whether the unit actually operated. The auction is held four years before delivery, and contracts run for one year as a rule. Prices differ by area: for fiscal 2028 delivery, Hokkaido, Tohoku, and Tokyo were 14,812 yen/kW, and Hokuriku, Kansai, Chugoku, and Shikoku were 8,785 yen/kW. Because contracts are for one year, they change every year.

Do not, however, simply multiply that unit price by the 2,000 kW of output. What gets contracted is installed capacity times an adjustment coefficient, and a four-hour battery is discounted by that coefficient (about 70.8% on an annual average in Tokyo). With the coefficient included, the actual receipt for a 2 MW unit is roughly 20 to 26 million yen a year.

Q8 covers what this adds up to over a year.

Solar and wind have grown, and the amount of electricity now swings widely with the weather. There is not enough capacity to absorb those swings — and that is where batteries come in.

"Not enough" shows up in the numbers. As of the end of December 2025, grid-scale battery storage actually connected to the grid came to about 640,000 kW. Meanwhile, the volume piled up at the grid connection study window was about 172 million kW. A gap of roughly 270 times.

But it is a mistake to read that gap purely as proof of popularity. A large share of it consists of low-feasibility placeholder applications on land such as disaster-prevention parks or plots with someone else's building on them. That is why the government began tightening up in 2026. From January, land survey results must be submitted with the application, and from April the deposit rose from 5% to 10% of the estimated grid connection charge. Applicants are also required to prove their right to use the land within two months of connection approval.

And to be straight about it: the upside is thinning.

• Market headroom has already narrowed in the actual figures. The shortfall rate for offline primary frequency control went from about 96.6% in fiscal 2025 to 67.9% in April–June 2026
• The ceiling on revenue has been lowered by regulation. The price cap went from 19.51 yen to 15 yen for trades from March 13, 2026 (a cut of about 23%), and a further cut to 10 yen has been decided for delivery from September 1, 2026. The procurement volume (the amount the market buys) has also been squeezed sharply, and exchange fees have doubled
• A level of 7.21 yen has been floated beyond that, but that one is still under discussion

Meanwhile, capacity market prices are on an upward trend, driven by rising electricity demand from data centers and semiconductor plants. The view closest to reality is that the center of gravity of the earnings is shifting from payment for standing by to payment for existing.

As for how long it lasts, we do not have an answer in years. We know only three things. Headroom has already narrowed. The price cap is headed further down. And the bar for securing new grid capacity went up in 2026. Turned around, that means the value of projects that have already obtained a grid connection study response is relatively rising.

The names that come up are all companies.

The first group is listed companies. In July and August 2026, three companies of quite different character disclosed acquisitions or arrangements within 48 hours of one another: one that paid about 700 million yen for a completed unit, one that put debt on a 100 MW-class project to be built over three years, and one that bought an already operating 2 MW unit and raised its earnings forecast for the period.

The second group is companies from other industries — neither utilities nor generators. Sumitomo Corporation, Kamigumi (port transport), Japan Petroleum Exploration, Idemitsu Kosan, Tokyu Land, West Holdings: the sectors are all over the place. Because construction, maintenance, and market operation can all be outsourced, entries from outside the power industry keep coming.

The third group is leasing companies and funds. Major lessors have announced large investment allocations: press reports put Tokyo Century at 100 billion yen through the fiscal year ending March 2030 and Sumitomo Mitsui Finance and Leasing at around 200 billion yen through the fiscal year ending March 2032. These are multi-year investment plans, not the price of a single project. Itochu set up Japan's first fund dedicated to grid-scale battery storage (over 8 billion yen) in 2024.

Overseas investors are looking as well. But the electricity business is a designated sector under the Foreign Exchange and Foreign Trade Act, so prior notification is likely to be required; the review takes 30 days as a rule, and can run to as long as five months if there are national security questions. The closing schedule has to be set to match.

This may come as a surprise, but in this market being able to close in cash is a strength. Sellers often want settlement in about a month, while a buyer using borrowed money needs one and a half to two months for internal approval and bank review, and six months to a year for project finance. It is not that financing is unavailable; it is that it does not arrive in time.

One more thing. In a transaction for a completed unit, what actually gets credit-checked is not the buyer but the seller and the construction company. Since you place several hundred million yen before handover, quite a few deals stall at internal approval right there. If the seller is a group company of a listed firm, the same asset at the same price clears approval more easily.

The figures on review periods, the dynamic that makes closing in cash a strength, and the point about approvals stalling on credit are all our own observations from the field; there are no reliable published statistics for them.

For a 2 MW high-voltage-connected system, the site is as a rule around 1,000 sq m. For a 50 MW-class extra-high-voltage system, it is roughly 10,000 sq m or more. The order of magnitude changes with the scale, so always read an area figure together with the scale.

Size, though, is not the deciding factor. Of the items that determine a site's worth, the one that dominates is the distance to the point where you connect to the grid.

People often say you cannot put one in an urbanization control area, but that is not accurate. You can, if you clear two gates: the municipality's development permit and the fire department. However, since a 2 MW high-voltage-connected system does not count as a "power generation business operator" under the Electricity Business Act, it cannot ride on the exemption for generation facilities, and you have to seek a development permit head-on as a Class 1 designated structure. And in areas where the municipality has not yet drawn up review criteria, there are no criteria, so there is no way for a permit to be issued. Get an answer from the municipality in writing before you sign anything.

The fire department gate applies separately from the development permit. Battery installations with a capacity of more than 20 kWh require an installation notification, and for outdoor installation a separation of at least 3 m from buildings is one of the standards. There are relaxations and exemptions, and how the capacities of multiple containers are added together varies by local fire department, so consult them before you design.

The distance to homes is not a pass-fail line. Think of it as a dial that tightens the checks required as you get closer. It becomes a walk-away line only when nighttime noise standards are exceeded and countermeasures cannot fix it (Q19).

Hazard maps are the same: do not rule a site out simply because it is shaded. Those maps are made for evacuation, and they depict a single worst case that occurs with a probability of 0.1% or less per year. If you exclude every shaded area, your candidates drift toward hillsides and plateaus, and you simply trade for a different set of costs: landslide risk, earthworks, and distance to the connection point. That said, "areas at risk of house collapse from flooding," "special landslide disaster warning areas," and inundation too deep for raised ground to handle (as a guide, more than 3 m to 5 m or deeper under the maximum assumed scale) are walk-away lines that no countermeasure saves.

There are three things you can check yourself before consulting anyone. Measure the straight-line distance to the nearest transmission tower or substation on Google Earth. Look at the grid headroom maps published by the transmission and distribution utilities. Check the land category and area in the land register. But this is only a first screen. The headroom maps often do not reflect the two-way issue specific to batteries, and what settles the matter is the grid connection study response (an application costs about 200,000 yen per site, the response takes two to three months as a rule, and the response is valid for one year).

There is no uniform per-square-meter market rate for land. The price is set by the grid conditions, the connection capacity, the location and earthworks, the market environment, and how far the rights have progressed.

For a 2 MW/8 MWh high-voltage-connected system, separate two numbers. "What it costs to build" and "what it costs to buy" are different.

The only transaction amount that has been made public is a single case in which a completed unit of about 2 MW/about 8 MWh was acquired for about 700 million yen (a timely disclosure filing on July 30, 2026). The 500–600 million yen is the rough guide for initial investment reported by Nikkei Energy Next. The 150–200 million yen for the rights stage, the 800 million to 1 billion yen for units with a track record, and the upper end for completed units are figures we observe in our brokerage work; they are not published statistics.

You may hear that you can start with just the rights for 150 to 200 million yen. Be careful here. That 150 to 200 million yen is a down payment. You still have to put 350 to 450 million yen into equipment and construction before it becomes a completed unit, and both the risk of finishing the build and the grid risk are yours.

The 350–450 million yen is a back-calculation: the rights portion subtracted from the 500–600 million yen rough guide for initial investment. It is not a figure compiled from quotations.

Beyond the listed price, there are two items to build into your funding plan. The consumption tax on the equipment portion (a taxable business recovers it through the input tax credit, but you front it until the refund arrives), and the price difference that comes from payment timing. On the same asset, we see a gap of about 6% between immediate cash settlement and payment on handover. That difference is not a discount; it is the price of the seller taking on the buyer's settlement risk.

Also, putting a single unit of this size onto bank project finance is structurally unfavorable at present. The fixed costs of review and documentation are heavy, and on a project of several hundred million yen they cannot be fully passed into the interest rate. The large loans that have been made public were either extra-high-voltage projects or portfolio deals bundling 14 sites nationwide (Q11).

It is also risky to count on subsidies and discount them off the total. Only three cost items qualify — detailed design, equipment, and construction — while land, earthworks, the grid connection charge, substation equipment, basic design, and consumption tax do not. Our sense from the field is that 30 to 50% of the total cost falls into non-qualifying items (Q12).

Let us say up front: what follows is revenue, not take-home.

We tallied the full set of contract results on JEPX and EPRX (365 days, July 2025 to June 2026), and for a 2 MW/8 MWh high-voltage-connected system the annual revenue came to 270 to 290 million yen at the median across the nine power service areas. The largest figure in the tally is Hokkaido at about 286 million yen, but over these 12 months the balancing market itself was thin in Hokkaido, and it was not an area where that environment could be fully captured. Tohoku is the only area that satisfies all three of procurement volume, shortfall rate, and unit price, and we read Tohoku's roughly 285 million yen as the effective top figure. And about 80% of that comes not from buying and selling electricity but from the balancing market — payment for standing by.

Three cautions follow from here.

First. This is revenue. The fee to the operating agent (the aggregator) comes out of it. The only published rate is one company's "5% of market profit." Since that company describes its competitors as "around 10%," we take the range to be roughly 5 to 15% of market profit. Other secondary sources speak of levels of 10 to 30%, but we have not been able to verify those. If you applied 10 to 30% to the annual revenue from the balancing market alone (median, about 230 million yen), 23 to 68 million yen a year would go to the operator. Maintenance, insurance, wheeling charges and the like come out on top of that.

Second. The difference between areas is not several-fold. The totals for the nine areas sit in a narrow band, from about 286 million yen in Hokkaido at the top to about 271 million yen in Kansai at the bottom. Where the difference clearly shows is in the two layers of the capacity market and JEPX, about 15.02 million yen a year between top and bottom.

That said, the band looks narrow partly because of how it was tallied. For the balancing market, which accounts for 80% of revenue, no battery-specific unit prices by area are published, so we apply a single figure common to all areas. Whether an area can actually capture that environment differs by area, and we judge it on three points: procurement volume, shortfall rate, and local unit price.

People sometimes say that the same 2 MW can be worth 290 million yen a year or 60 million yen a year, but that 60 million yen is not about area. It is a difference in how you pick the unit price for the same operation: whether you apply the prevailing winning prices for batteries, or the average price across the whole market including thermal and pumped storage. That roughly fourfold gap is the distance between "the regulator's statistics" and "what operators feel."

Third. The regime that produced these numbers is already changing. Of the 365 days we measured, 256 fell under the old regime with a 19.51 yen cap, and only the last 109 days were at 15 yen. In other words, the 270 to 290 million yen rests on a price level that has already ended. And from delivery on September 1, 2026, the cap falls to 10 yen (it is 15 yen through delivery on August 31). Since about 80% of the revenue is the balancing market, one step down in the cap will shrink the revenue from that balancing market almost proportionally.

So we separately estimate a cruising level three to five years out, after the cuts have run their course and battery prevailing prices have moved closer to the market-wide average. This does not mean the move to 10 yen on September 1 will drop things to that level all at once.

These too are rough annual revenue (top-line) figures. Bull = the 15 yen cap holds, the doubling of the capacity market reference price is reflected, and the simultaneous market is delayed; base = one step down to 10 yen, the doubling reflected, competition increasing gradually; bear = 7.21 yen is reached, competition surges, and the simultaneous market is partly applied within the period. These are estimates with stated assumptions, not fixed values, and they vary widely with each project's connection costs, grid connection charge, and operating skill.

So do not take a one-line "the yield is X%" at face value. What you should ask is: as of when, revenue or take-home, and what fee percentage was assumed.

For a 2 MW / 8 MWh high-voltage-connected system, our columns give exactly one payback figure. On a rough model assuming construction costs of 640 million yen and annual operating profit of 40 to 80 million yen (annual revenue of 60 million to 110 million yen), payback comes in 8 to 16 years. That figure is as of April 2026, and it assumes a balancing market price cap of 15 yen.

What we want you to notice is that this annual revenue of 60 million to 110 million yen differs from the actual figures in Q8 (270 million to 290 million yen a year) by more than a factor of three. It is the same 2 MW / 8 MWh; only the assumed unit prices differ, and the result moves that much. Whenever you ask about payback, always ask alongside it what annual revenue the calculation assumed.

There are also things this number leaves out: taxes, loan repayments, and the battery replacement cost expected in years 10 to 15. Treat it as nothing more than annual operating profit divided by construction cost.

You will also hear that payback can come much faster. One scenario presented as another company's estimate assumes every bid submitted is accepted (full clearing) and arrives at roughly 450 million yen a year and payback in about 1.5 years. But that is an estimate rather than a track record, and the assumptions are extreme. It changes a great deal once the price cap comes down.

Here is a calculation you must not do: take a purchase price of 700 to 800 million yen, divide it by the 270 to 290 million yen of annual revenue in Q8, and call it three-year payback. The 270 to 290 million yen is revenue, not profit, and in any case these are figures from columns with different assumptions and different dates, so the division does not hold.

The variables that move payback are these.

• The cost of the work needed to connect to the grid. Even with the same annual earnings, a change in total investment moves the return (in a 50 MW-class example, a total investment of 10 billion yen gives an IRR of 10%, and 11.8 billion yen gives 7.8%)
• Regulatory changes on the revenue side. The balancing market price cap has come down from 19.51 yen to 15 yen, and will become 10 yen from the delivery period beginning September 1, 2026. A level of 7.21 yen has been indicated beyond that, but it is conditional and has not been decided yet
• The gap between when you start receiving power and when you start earning in the markets. This normally takes one to three months
• The battery replacement cost in years 10 to 15

Some things you cannot count on. A 20-year project life is not something you can simply assume. Without a 20-year fixed revenue contract, how many years you run is decided by your contracts and your cash flow (the 17 years used for tax purposes is a separate matter; see Q23). Nor can you count on recovering money by selling the equipment at the end, because a market for disposing of used batteries has not developed yet.

One more inconvenient fact. Where revenue is not fixed by contract, a sensitivity analysis submitted by Mitsubishi Research Institute to a Ministry of Economy, Trade and Industry study group put the base-case IRR at minus 1.5% even with capacity market revenue included (assuming construction costs of 60,000 yen/kWh). Turning a profit depends on bringing construction costs down to 50,000 yen/kWh or below, or on wholesale price spreads coming in wider than expected.

A rate of return (IRR) and how many years until you get your money back are two different things. In an estimate for a 47 MW extra-high-voltage project, an IRR of 13.3% gives payback in 7 years, 6.6% gives 11 years, and 3.5% gives 14 years, which is not a simple inverse relationship.

Once operation begins, the money going out falls into six broad categories.

• Maintenance (O&M) - monitoring, inspections, callouts when something fails, and keeping licensed staff available
• Fees to the operating company (the aggregator)
• Insurance premiums
• Grid usage fees, the generator-side grid charge, and the like
• Loan repayments, if you have borrowed
• Battery replacement or expansion expected in years 10 to 15

Rough figures can be given for each item, but let us say this first: do not add them up into a single total of so many yen per year. The assumptions behind each one differ (scale, region, how the contract is structured, whether discounts apply), so adding them produces a number with a precision that does not exist.

The item where we most need to be honest is the cost of maintenance. There is still no officially established standard unit cost in this field. What you can refer to is the labor unit cost assumed in the model used to calculate the price cap in the national auction (5,000 yen per kW per year), and NREL's figure in the United States of about 2.5% of construction cost. The latter includes the cost of adding battery capacity and assumes 15 years of operation, so it does not translate directly into a maintenance figure for Japan. In the field, some put it at around 5 million yen a year including insurance, and others at 2.5 to 3 million yen a year excluding insurance, but both are figures presented by operators and neither is a level that has been officially verified.

So do not compare quotes by their totals. Only when they are broken out line by line - licensed staff, monitoring, callouts, communications, insurance - can you have a discussion about whether a quote is high or low.

On the operator's fee, always confirm what the percentage applies to. The same 5% means nearly double the payment depending on whether it is 5% of gross revenue or 5% of profit after charging costs and the like have been deducted. On a gross basis, the industry range is said to be 3 to 10%. There are four things to check before signing: whether capacity market revenue is part of the calculation; whether it is before or after charging costs are deducted; who bears the penalties; and whether exchange fees are separate.

There is another cost that moves with design. How you contract for the grid capacity you use can make a difference of 10 million to somewhere in the 19 million yen range per year at high voltage (an average of about 15.36 million yen across ten utilities). This is not the total cost, but the range that good or poor design moves.

Battery replacement is not an annual cost, but it may arrive in years 10 to 15. When you look at a 20-year plan, check whether this cost is included.

The 2 MW / 8 MWh model in our column assumes 10 million to 20 million yen a year for O&M, insurance and similar items, and several million yen a year for grid usage fees and the like (a different model puts total outgoings at 15 million to 25 million yen a year). But the operator's fee, battery replacement, and loan repayments are not in that model, so do not read it as the whole of your annual costs.

You can borrow. But once it comes to a single 2 MW high-voltage-connected system on its own, the picture changes.

First, let us correct an explanation you often hear: that financing is unavailable because the asset cannot be taken as security. The cause and effect are wrong. Taking security is perfectly possible. There is a way to pledge the equipment itself (and it can be registered), and if the land is yours, there is also a way to treat land and equipment together as a single security. What is lacking is not the ability to create security, but the prospect of selling that equipment to recover the loan, because a market for disposing of used batteries has not developed yet.

So what decides whether a lender lends is not security but how readable the cash flow is under contract. The published examples make this clear.

The ratio of 6.2 billion yen to 6.5 billion yen at Nishigo cannot be applied to other projects as it stands, because it is not disclosed whether repayment could be pursued up to the parent company, whether there was a parent guarantee, or how much equity was put in.

The difficulty starts here. Every example above is either a large extra-high-voltage project or a bundle of multiple sites. A single 2 MW high-voltage-connected system is, for now, at a structural disadvantage when it comes to project finance. The fixed costs of due diligence, contract drafting and agency work are heavy, and on a project of a few hundred million yen they cannot be absorbed into the interest rate.

Cash settlement is also the norm for a reason other than financing being unavailable. Sellers want to close in about a month, while a buyer using debt needs 1.5 to 2 months for internal approval and bank review, and half a year to a year for project finance. It is not that the financing cannot be obtained; it cannot be obtained in time.

There are three realistic routes.

• Secure the asset with cash first, then refinance once it is running and has a track record. Operating performance is the most effective credit material there is
• Lock in a long-term fixed revenue contract first, then borrow
• Use a lease or an installment purchase

In the smaller size range, a loan-based crowdfunding arrangement of 495 million yen against a collateral valuation of 602 million yen (a loan-to-value of roughly 80%) has also been published. The interest rate, however, is higher than bank project finance.

Two cautions at the end. If the land is leased, the ways of taking security are constrained (you cannot place a mortgage on a leasehold right). And on projects that used a subsidy, pledging the equipment as security also counts as disposal and requires prior approval, which reduces your freedom to bring in financing later.

There are no reliable published statistics on how long sellers and buyers take for their reviews; this is our own observation in the field.

The programs exist. But what we want to write about here is not how to get one, but how to decide whether to use one.

A 2 MW high-voltage-connected system falls in the bracket of up to 1/3, capped at 1 billion yen. But do not multiply your total construction cost by 1/3. Only three things are eligible: detailed design costs, equipment costs, and construction costs. Land, site preparation, receiving and transforming equipment, the grid connection charge, basic design costs, and consumption tax are not eligible. Our sense from the field is that 30 to 50% of total costs falls into that ineligible territory.

With that said, here is what comes attached if you use one.

• For 17 years, you need prior approval to sell, lease, scrap, or pledge the equipment as security
• Three years of effect reporting, and five years of document retention
• Conditions on safety standards and certifications apply, which in practice pushes you toward domestically made packages

The figure of 17 years is not written in the application guidelines themselves. What the guidelines say is the statutory useful life period, and 17 years is what results from reading the ministerial ordinance categories.

That 17-year restriction stays with the asset as a discount whenever you come to sell. On top of that, the guidelines do not spell out how the repayment amount is calculated if you sell during the restriction on disposal period; it depends on administrative practice. Whether transferring the shares of an SPC amounts to a disposal in substance is also something you end up asking about case by case.

On price, too, a subsidy is not automatically the better deal. Government materials note that equipment plus installation work under subsidized projects runs at 68,000 yen per kWh, while projects that skip the subsidy and choose overseas-made equipment can be seen at levels of 20,000 to 40,000 yen. Take 1/3 off 68,000 yen and you are still at about 45,000 yen, so buying overseas-made equipment with your own money can come out cheaper.

These per-kWh figures are the price of the storage system (and installation work); they do not include land, site preparation, receiving and transforming equipment, or the grid connection charge. Whether the 20,000 to 40,000 yen for overseas-made equipment includes construction work is not clear from the government materials. The 500 to 600 million yen in Q7 is a total that includes those items, so the two cannot be compared directly.

The situations where skipping the subsidy is the rational choice have also been set out. When rising prices or a weak yen make you want overseas-made equipment. When the six to twelve months of waiting for selection would be fatal. When co-locating with FIT or FIP facilities. And when your exit is an M&A within a few years.

Two final points. The most recent round (fiscal 2025) has already closed. The programs are revised every year, so check the guidelines in force when the next round opens. And the JC-STAR 1-star cybersecurity certification is both a subsidy requirement and, from April 2027 for high voltage, scheduled to become a requirement for connecting to the grid regardless of whether you take a subsidy.

Let us start with the idea that you can write the whole thing off in the first year (immediate expensing).

That scheme (the bold investment promotion tax measure) is being created, but it has not started yet. The amendment to the Industrial Competitiveness Enhancement Act on which it rests has to be enacted and brought into force, and the cabinet order setting the effective date, the Ministry of Economy, Trade and Industry ordinance, and the application forms all have to be in place before you can apply. Enacted, in force, and open for applications are three separate events.

And here is the more important point. The words grid-scale battery storage and storage facility do not appear even once in the text of the tax reform outline. What the outline lists is only categories of equipment: machinery and equipment, tools, fixtures and fittings, buildings, and so on. Whether these assets qualify will be decided in two stages: by the criteria in a ministerial ordinance yet to be written, and by the Minister of Economy, Trade and Industry's confirmation of each investment plan. As of today, there is no primary source that lets anyone state flatly that grid-scale storage facilities can be immediately expensed.

Even if it does become usable, there are four things to watch.

First, your total tax bill does not fall by a single yen. Under tax law, the battery itself and the PCS are expensed over 17 years, and immediate expensing only pulls that forward into the first year. Because there is less expense in later years, profit actually rises. What you gain is only the time value of deferring the payment; on an estimate of 3 billion yen at an effective tax rate of 30% and a discount rate of 5%, that is around 200 million yen net.

Second, there is a floor on investment size. You need to meet 3.5 billion yen or more, or 500 million yen or more for small and medium-sized enterprises, at the level of the investment plan. It is not a matter of one site costing 500 to 600 million yen and therefore reaching the 500 million yen route. An SPC in which a large company holds a majority stake becomes a deemed large enterprise, so even with capital of 100 million yen or less it cannot use the 500 million yen route and faces the 3.5 billion yen threshold.

Third, the form of your contract can rule it out. The outline excludes assets held for the purpose of leasing to others. Running the asset yourself points toward being eligible; a tolling structure that fixes your revenue points toward being ineligible. That is the awkward part: the contracts that make financing easier to obtain are the ones that move you further from immediate expensing.

Fourth, combining it with a subsidy does not give you a double benefit. The portion covered by a subsidy shrinks the acquisition cost for tax purposes, so it is expected to fall outside immediate expensing in substance (this view follows how similar tax measures have been treated up to now; the details await the coming orders, ordinances and circulars).

Look at the exit as well. If you sell equipment whose book value has been driven down to near zero, the gain on sale is taxed all at once. There is no special rule to soften that. If you intend to sell within a few years, there are situations where choosing the 7% tax credit rather than immediate expensing is the more rational option (the deduction is capped at 20% of that year's corporate tax).

There is one more tax, separate from corporate tax, that bites. Under the Local Tax Act, a grid-scale storage facility counts as an electricity supply business, so enterprise tax is levied on revenue rather than profit. In other words, as long as you have electricity sales revenue, you are taxed even in a loss year (before you begin selling electricity, during construction, this tax does not apply). Adding the special corporate enterprise tax to the standard rate of 0.75% gives an effective burden of roughly 1.05% of electricity sales revenue. Whether the balancing market's payment for standing by falls within this base, however, has not yet been answered by the government. In practice, you make an advance inquiry to the prefectural tax office.

Always confirm tax matters with a tax accountant.

You can. Day-to-day operation and buying and selling in the markets can be outsourced.

First, what you do not have to do at this size. A 2 MW high-voltage-connected system does not require notification as a power generation business, a construction plan notification, a pre-use self-inspection, or a safety management review. The heavy procedures begin at an output of 10,000 kW or a capacity of 80,000 kWh.

Under the Electricity Business Act, there are three things you must do (separately from these, there are other filings such as notifying the fire department of the installation).

• Keep meeting the technical standards
• Appoint a licensed chief electrical engineer
• Draw up safety regulations and file them

You do not have to employ the chief engineer yourself. There is a route where you outsource to an electrical safety corporation or similar and obtain government approval (available where output is under 5,000 kW and voltage is 7,000 V or below). Bidding in the markets, submitting plans every 30 minutes, and settling the deviations are all handled in practice by the aggregator.

So what is left for the owner? Four things.

First, deciding in the contracts who is responsible for what. A storage facility runs on a bundle of contracts: three or four are in force at any given moment, and across a single day eight to ten kinds are at work. In particular, who moves first when something goes wrong becomes nobody's job unless it is written down (Q16).

Second, keeping records in your own hands. Record your monthly market participation in a set format, and keep BMS and EMS logs and operating histories in the cloud for three years or more. Not as a legal obligation, but as the foundation for actually collecting on your insurance when an accident happens (Q20).

Third, checking the tax position. As long as you are selling electricity, enterprise tax applies to your revenue even in a loss year (the revenue-based portion of enterprise tax, an effective burden of roughly 1.05% of electricity sales revenue). And since there is no official answer on whether the balancing market's payment for standing by falls within that base, making an advance inquiry to the prefectural tax office where your head office is located, and settling the point, becomes the owner's job.

Fourth, the preparation before you start. In practice, the operating arrangements, the chief engineer, and communications are put together starting about six months before you connect to the grid. It takes 2.5 to 3 weeks just to get a quote from an operating company. A pre-configured backup router can take 2.5 to 3 months to procure (a rough figure based on the general lengthening of lead times for network equipment; we have not been able to confirm lead times for storage-facility routers specifically).

One last caution. Even if you order the construction on a turnkey basis, post-commissioning maintenance, warranties, and communications drawings are not necessarily included. Confirm in the contract whether they are. And even once the facility is complete, if a single procedure needed to enter the markets is outstanding, your revenue does not begin.

The party that builds the charge and discharge schedule and earns money in the markets is an outsourced operator called an aggregator. For a 2 MW high-voltage-connected system this is not a matter of whether to choose one; without one, there is no revenue at all. The practical work of bidding, submitting plans and handling communications cannot be run for a single unit on its own.

Note that maintaining the equipment is not part of the aggregator's role. The party that goes to the site when it stops is a different one (Q16).

Start with fees. Only one company publishes its rate, and that company states 5% of market profit (with a contract term of either one year or five years). Others do not disclose theirs, but that one company's comparison table on its own website describes competitors as around 10%. If that is accurate, the range can be estimated at roughly 5 to 15% of market profit. It is not a published industry statistic. And what is deducted before arriving at market profit is not consistent from company to company.

The same 5% can mean nearly double the actual amount depending on what it applies to. Before signing, confirm these four points in writing.

• Is capacity market revenue included in the calculation?
• Is it before or after charging costs (the cost of power purchased on JEPX) are deducted?
• Who bears the penalty (imbalance) when actual output deviates from the plan, including cases caused by a system failure or a dispatch error on the aggregator's side?
• Are exchange fees separate, or included?

That said, something matters more than the fee rate: the difference in revenue that comes from how well the asset is operated is far larger. In the United Kingdom, batteries operating in the same market under the same rules show annual revenue differences of more than 30% (over 100,000 pounds per MW at the top, against an overall average of 69,000 pounds; these figures exclude capacity market revenue). And one of the top performers earned more while cycling 15% less than the lowest-ranked battery. Running it harder is not what makes money.

Japan does not yet have comparable data of this kind. So the most useful thing when choosing is whether a company can produce its operating record for the past 12 months: annual revenue per MW, bid acceptance rate, and availability. Fee rates can be moved through negotiation; operating skill does not change easily.

Two more contract points. Assuming there will be moments over 20 years when you want to change operators, look at the contract term and the clauses on early termination and substitution of resources. And you can keep the contracts in your own name while entrusting only market participation and plan submission to someone else.

As of March 2026, 142 companies had completed the national registration required to act as an aggregator. Only some of them, however, have a track record operating grid-scale battery storage. Overseas there are also floor-price, tolling, hybrid and SaaS contract structures, but in Japan almost all are revenue-share. Tolling, where you hand over operating rights in exchange for fixed revenue, is a form used mainly on large extra-high-voltage projects, and domestic pricing levels for a 2 MW high-voltage-connected system have not been published.

This is the most misunderstood point in this business.

People often say that as long as you have an aggregator, a licensed chief electrical engineer and the manufacturer, you are covered. Here is how it actually works.

What matters here is that the gap is not in responsibility but in who actually does the repair. In law, the duty to supervise safety across the whole installation, including the battery side, is imposed on the chief engineer. But what is imposed is supervision, not repairing the PCS or the battery itself.

On top of that, if someone not approved by the manufacturer works on the equipment, the warranty is customarily voided even if that person is licensed. It is not that the battery side cannot be touched; it is that touching it costs you something else.

The result is projects where the person who first isolates the fault and starts the repair when something goes wrong is not written into any contract. That is why deciding on an O&M company that looks after the whole installation and acts as a single point of contact becomes the owner's job.

The thing most often missed in the field is communications. When a communications failure is displayed, the party you need depends on whether the problem is the router or the line. And unless the as-built documentation showing which router was configured how has been handed over, nobody can fix it at the moment it fails. This is not because people are cutting corners on paperwork; it is because the safety regulations do not require it to be written down. If no spare router is kept on site, the outage can run for several weeks before it is fixed.

Look at repair times as well. Procuring new major equipment takes 6 to 9 months for a domestically made PCS, 4 to 6 months for a battery container, and 6 to 12 months for a cubicle or transformer. Recovery periods by severity of damage are given as rough guides of 30 to 90 days for minor partial damage, 120 to 180 days for moderate damage, and 240 to 360 days for a total loss.

There is insurance (machinery insurance), but it covers damage from internal causes such as a burnt-out PCS or a BMS failure. Wear from age, battery degradation (falling SOH), manufacturing defects, inadequate inspection, and deviation from operating conditions are excluded. And the money does not move until the cause can be established.

One benchmark when reading contracts. In mature overseas markets, a company that takes on the whole installation guarantees around 97% availability a year by contract, excluding external factors. In Japan, no party able to take that on has emerged yet. If you are told it cannot be guaranteed, listen for whether that is because of how broad the exclusions are, or because there is nobody to stand behind the guarantee.

Before you buy, also check whether the as-built documentation exists (completion drawings, single-line diagrams, test reports and so on). Equipment without drawings makes you pay the cost of having no drawings every time it is repaired. And a 20-year warranty is a piece of paper if the company behind it does not exist in ten years.

It hits you through three channels. It does not end with the revenue you failed to earn.

First, the revenue that never comes in. For a 2 MW/8 MWh example, the revenue lost per day of downtime is put at 66,000 yen on a conservative view, 151,000 yen as a standard case, and 288,000 yen on the high side. But these are not actual results. They come from assumptions used for insurance design, which set annual revenue at 24 million yen (conservative), 55 million yen (standard) and 105 million yen (high side). The premises differ from the measured figures in Q8 (270-290 million yen a year). Because they simply add up three markets, they skew high, and the column itself cautions that actual revenue will come in below that. For your own project, divide the annual revenue in your own business plan by 365 and substitute that.

Second, penalties in the balancing market. If you cannot respond to a dispatch instruction, you are charged an amount equal to the contracted amount (1.0 times). If the same facility fails to respond three or more times in one month for the same product, new trading in that product is suspended. There is a relief framework that leaves out of the count events arising on the grid side, such as output curtailment, that could not have been foreseen at the time of bidding.

Third, penalties in the capacity market. This applies where you hold a capacity contract, and it is the most structural of the three. A planned outage you have notified counts as one time block each, but an outage from a sudden failure is counted at five times. From that total, 8,640 time blocks a year (equivalent to about 180 days) are deducted, and what remains is what the penalty applies to.

The five-times figure is a multiplier for counting time blocks, not a fivefold increase in the amount you pay. Normal nights and holidays are excluded.

In other words, the rules are not saying "never stop." They are saying "turn unplanned outages into planned ones." If you can catch the warning signs and convert an outage into a planned one, your exposure to capacity market penalties shrinks to as little as about one fifth. That is the economic reason to spend money on maintenance.

There are limits to what insurance fills in, too. Revenue lost while the plant is down is normally covered by business interruption (BI) insurance, but that covers losses caused by an accident in which something physically breaks. An outage with no physical damage behind it, such as communications going down or being shut out by market rules, sits outside how this insurance is built. There is also a deductible period, and you bear the first stretch after the incident yourself (negotiable from around 14 days). Assuming 180 days of cover, a 14-day deductible and 151,000 yen a day, the total you need to prepare for comes to about 25.07 million yen.

One more thing. In European policies, a clause capping the payout at one twelfth of annual revenue unless monthly revenue records are kept is spreading, and it is starting to be raised as an issue in Japan as well. Battery storage revenue is concentrated in summer and winter, so if that clause is triggered the gap from the actual loss widens considerably.

And if establishing the cause drags on, payment is effectively frozen. In a fire in Kagoshima Prefecture in March 2024, it was reported that more than a year after the incident the insurance money remained unpaid, even though the operator carried fire insurance. In the same case, in which a 1 MW / 6,400 kWh facility burned down completely, the opportunity loss from the outage is estimated at more than 40 million yen.

No public data exists showing how much downtime battery storage plants in Japan actually experience.

It is not zero. It has actually happened in Japan.

In March 2024, a fire broke out at a battery storage plant in Isa City, Kagoshima Prefecture. It took about 20 hours to put out, four firefighters were injured, and the storage equipment burned down completely. It was a 1 MW-class facility using NMC (ternary lithium) cells.

Those injured were the firefighters tackling the blaze. Reported capacity ranges from 6,400 kWh to 7,000 kWh depending on the source.

What we want you to take from this is less the fire itself than the length of time your money stops. In this incident it was reported that, although the operator carried fire insurance, more than a year after the event the insurance money had still not been paid. Pinning down the cause of a thermal runaway fire can take over a year, and during that time the insurance money is effectively immobile. At a plant that combines equipment from several manufacturers, it can also become impossible to attribute responsibility to any one company. In South Korea there were 23 ESS fires between 2017 and 2019, but no conclusive cause was ever proven.

The type of battery makes a difference too. Outdoor container-type LFP (lithium iron phosphate), which is now the standard the insurance market will underwrite, progresses more gently than NMC (ternary lithium) once thermal runaway starts, and underwriting terms are correspondingly easier. That does not mean it cannot burn; it means the degree of risk is different. Conversely, using NMC raises the difficulty of everything: consultation with the fire department, insurance, and explaining the project to residents.

The regulatory framework looks like this.

• From January 2024, batteries above 20 kWh must be notified to the fire department. Outdoor installations must in principle be at least 3 m from buildings, with relief available for certified cubicles and the like
• Electrolyte is treated as a hazardous material (Class 4) under the Fire Service Act, and depending on the quantity you may face regulation as a hazardous materials facility and have to appoint a safety supervisor
• From November 2025, batteries above 20 kWh became subject to accident reporting to the national government as major electrical facilities

One caution is needed. A 2 MW high-voltage-connected system sits below the size threshold for the government's statutory checks: construction plan notification, pre-use self-inspection and safety management review. You cannot say it is safe because it has passed government inspection. The buyer needs to confirm that the safety regulations notification, the appointment of a licensed chief electrical engineer, and the pre-use self-confirmation paperwork are actually in place. An existing project running without the required notifications is one to pass on at that point.

Underwriting terms are decided by choices made at the design stage, not by effort after you buy. The battery chemistry, fire suppression and fire spread prevention measures, distance from housing, 24-hour remote monitoring and automatic isolation: these translate directly into the terms of your insurance. And once operation starts, keep at least three years of BMS and EMS logs and operating records. They are the footing on which you receive a payout after an incident.

Water can start a fire too. At a PG&E storage plant in the United States, faulty panel installation let rainwater into the container, and that was identified as the cause of ignition.

What makes the noise is not the batteries themselves. It is the equipment that cools them, and the fans on the machine that converts the electricity (the PCS). On top of that, a grid-scale battery storage plant is designed to run 24 hours a day, and shutting down at night is effectively not an option for the business. So what bites is the night-time limit.

These are values 1 m from a single unit, not what you would hear at the site boundary or at a house. At a large site with dozens of containers lined up, the combined sound is substantially higher than any individual unit.

Regulatory limits vary by land-use zone and time of day.

In residential zones, that means it has to be quieter than an air conditioner outdoor unit. The actual figure is set by each prefectural governor within that range, so you need to check locally. The hours that count as night-time also differ between municipalities.

Note that the legal obligation attaches to the regulatory standards above. Separate from these there are environmental quality standards (45 dB at night in general areas, for example), which carry no penalties but can become grounds for resident litigation or administrative guidance. In practice, some take the stricter environmental standards as their management target.

The Noise Regulation Act has no category for battery storage plants. However, cooling equipment is likely to fall under blowers (rated output of 7.5 kW or more), in which case the site is regulated as a designated factory. Municipalities also have their own ordinances, so consult the environmental department at the planning stage.

Do not judge by distance alone. In theory, doubling the distance cuts the level by about 6 dB, but at a large site where containers stand in a row, the drop can be only about 3 dB at close range. At night, a temperature inversion overhead can refract sound downward, so attenuation is sometimes smaller than predicted. That is why, in practice, noise at the site boundary is predicted in advance with a specialist simulation (from several hundred thousand yen).

Here is a rough guide to countermeasures and their effect.

• Choose low-noise equipment: 5-15 dB (absorbed in the difference in equipment price)
• Lay out the site carefully: 5-10 dB (design cost only)
• Build a noise barrier: 10-20 dB (from several million yen)
• Throttle night-time operation: a last resort, since it hits revenue directly

A noise barrier is not a cure-all. Simply blocking the line of sight gives about 5 dB, raising it gives 10-15 dB, and an optimal design gives up to 20 dB, but against low-frequency sound such as transformer hum (the 100-200 Hz band) it is worth only about 5-8 dB. And what is a matter of several million yen if built into the design jumps to several million to over ten million yen if retrofitted after complaints start once the plant is running.

When buying, the test cannot be how many meters it is from the nearest house. Distance is not a pass/fail line; it is a dial that raises the standard of evidence you should demand. If it is close, ask for three things: a prediction calculation against the night-time limit, confirmation that noise countermeasures are built into the design and the budget, and a record of the explanation given to residents together with the consent of adjacent landowners. There is only one line at which you give up on a project: when even with countermeasures the night-time limit cannot be met.

One last point. The biggest cause of trouble is not the noise itself but poor explanation. When a seller says the residents have already been briefed, it is safest to treat that as no briefing at all unless there is a record. For the 20 years after operation starts, the one who goes on being that community's neighbor is the buyer, not the seller.

Yes, you need it. And the same design as for solar will not work. The reason is clear: the comprehensive movables insurance that is standard for solar excludes, in its policy wording, generating facilities of 500 kW and above. The main equipment of a storage plant will not be underwritten at all.

You need six types.

• Property insurance - for damage from an external event such as fire, wind or flood
• Machinery insurance - for internal breakdowns that stop the plant (PCS burnout, BMS failure, transformer insulation breakdown and so on)
• Business interruption (BI) insurance - replaces revenue while the plant is down
• Liability insurance (PL) - third-party liability for fire spread, electric shock, toxic gas and the like
• Cyber insurance
• Theft

If any one of them is missing, the chain of losses does not stop. With property but no machinery cover, a PCS burnout comes out of your own pocket; without BI, you run out of cash while waiting for repairs; without PL, a fire that spreads ends the business.

As a rough guide, premiums run 0.5-1.0% of replacement cost a year. For an illustrative 2 MW/8 MWh project with a total cost of 600 million yen, one calculation puts a 0.75% rate at 4.5 million yen a year, but that is an estimate worked back from overseas benchmarks, and no published data on actual Japanese rates exists (even at the same 2 MW/8 MWh size, construction costs range from 500 million to 800 million yen).

Here are four points you cannot skip.

First, set the sum insured at replacement cost. Not what you paid, and not market value, but the amount it would take to buy the same thing again at the time of the incident. Market value deducts depreciation, which leaves you underpaid when something happens.

Second, do not let that sum fall short. If the sum insured is below the required amount, even a partial loss pays out only in that proportion (proportional reduction). Insure equipment with a replacement cost of 600 million yen for only the 400 million yen of hardware, and a 200 million yen loss pays 133 million yen, leaving the 67 million yen difference for you to bear.

The grid connection charge paid to the utility becomes the utility's own equipment, so it falls outside property insurance. Deduct it when calculating the sum insured. That said, check on the liability side for the risk of being pursued for damage your own incident causes to the utility's equipment. Do not leave the replacement cost itself untouched either. Equipment costs fall about 10% a year and construction costs rise 2-5% a year, so review it once a year. Over-insuring is wasted premium too.

Third, know the gaps in BI. There is a deductible period, and the first 14 days or so are on you. There is also a cap on the number of days covered. On top of that, unless you keep monthly revenue records, a clause capping the payout at one twelfth of annual revenue (a volatility clause) is becoming standard in Europe. Battery storage revenue is concentrated in summer and winter, so if that clause is triggered the gap from the actual loss widens considerably. And until the cause is identified, no money moves.

Set the number of days covered from the repair period for each level of incident. Thirty to 90 days for a light partial loss, 120-180 days for a moderate one, 240-360 days for a total loss. So in practice you take quotes at three points, 90, 180 and 365 days, and compare.

Fourth, earthquake, volcanic eruption and tsunami are excluded as standard. Tsunami is classified with earthquake rather than as flood, and liquefaction and ground subsidence fall on the same excluded side if they are earthquake-induced. Filling the gap takes an earthquake risk endorsement, but its underwriting terms are not published, so you have to ask each insurer individually. Even with flood cover you do not get everything back (there are reduced payment ratios, deductibles and payment limits).

Premiums can be brought down by design. Choosing LFP, presenting UL 9540A test data, fire suppression and fire spread prevention measures, keeping 100 m or more from housing, and a 24-hour remote monitoring setup: these are the items underwriters assess.

Judge insurance not by whether you have it, but by when, how much, and on what conditions it pays.

You may picture a fire or an accident, but what we cite as reasons to walk away comes much earlier than that. It is defects in the paperwork and the contracts.

Most of the storage plants brought to us as projects for sale are not yet operating as power plants. What is inside is land, grid-related paperwork, and a bundle of business plans. At the end of December 2025, about 640,000 kW was actually connected. About 172 million kW was stacked up at the application desk. That gap is what is circulating in the market.

Losses mainly take six forms.

First, mistaking the stage. It is not unusual to be told that consent to connect has been obtained, and then find that all that exists is a grid connection study response. The connection ladder has three rungs: (1) study response received, (2) connection contract applied for, (3) connection consent granted. Get the rung wrong and your whole view of the price goes wrong with it.

Second, defective contracts. A large deposit, with no mechanism to protect that money and no promise to return it if the project collapses. No exit written in figures, in amounts and timeframes, for what happens if construction costs rise or the schedule slips. In that case the buyer carries all of the variability. Paying before completion also means taking on the risk that the company building it goes under (bankruptcies and closures among power generation businesses hit a record high in fiscal 2024).

Third, it cannot be built where it is. In an urbanization control area where the municipality has not yet written its review criteria, there is no way for a permit to be issued, and as things stand it cannot be built. Even with perfect rights and perfect paperwork, land you cannot bring large equipment onto will not become a storage plant.

Fourth, an out-of-date business plan. The revenue assumptions moved in three directions at once in 2026: unit price, volume and cost. The cap price in the balancing market went from 19.51 yen to 15 yen, and will fall further to 10 yen from the delivery period starting September 1, 2026 (the three falling are primary reserve, secondary reserve 1 and the combined product; secondary reserve 2 and tertiary reserve 1 have already been at 7.21 yen since April 2024). The procurement volume, meaning the amount the market buys, has also been cut back sharply, and exchange fees have doubled. Plans drawn at 19.51 yen, and plans drawn at 15 yen too, do not depict what things look like from September 1.

Fifth, going offline. A sudden failure is counted at five times the time blocks of a planned outage under capacity market penalties (Q17).

Sixth, a warranty that means nothing. With a battery capacity warranty, the question is not whether one exists. Read the provisos. A guaranteed state of health at the end of the term that falls below 70% is taken as a sign that the back half of the business will thin out. And if the warranty conditions, such as annual cycles, charge and discharge rate, SOC range and temperature, conflict with how you actually plan to operate, the warranty becomes a dead letter.

Turn that around and here is how you prevent it. However plausible it sounds, what the seller says is a claim; it becomes a fact only once it is converted into a document issued by a third party. Anything that cannot be converted should be treated as unverified. And the absence of red flags is only a starting point. Only when the points that still lack documentary evidence are converted into conditions precedent, price, or representations and warranties does the checking actually protect the buyer.

A 2 MW high-voltage-connected system cannot take part in the auction that pays fixed revenue for 20 years (30,000 kW minimum). All of the revenue depends on market prices, so a single move in the rules makes the yield a different animal.

Yes, you can sell it. But to be honest about it, the market has only just started up.

The earliest case we have been able to confirm of a listed company disclosing, in explicit terms, the acquisition from a third party of a storage plant that is operating and participating in the market is dated July 31, 2026. The population is still thin and checking every case comprehensively is difficult, so we do not assert that it was the first in Japan. Japan's storage plant market has until now been a market for building. A market in which owners change hands is only now beginning to take shape.

There are four exits. The form of the transaction is itself the exit.

The going levels we observe in our own practice in the first half of 2026 are 150-200 million yen for rights only, 700-800 million yen for a completed plant, and 800 million to 1 billion yen with a track record (all excluding tax). These are not published statistics. Only one transaction has had its price made public, a completed plant at about 700 million yen.

What deserves attention is the 100-200 million yen between a completed plant and one with a track record. Not one yen of that is added equipment. The containers, the PCS, the receiving and transforming equipment and the land are all the same. The only things that move are the time it takes to reach market participation, and whether the business plan can be verified.

So if you plan to build a track record and then sell, there is a caution. A track record splits into two layers.

• The record as numbers, meaning contracted prices, award rates, availability and the fact of commercial operation, passes to the buyer in the sale
• Qualification to participate in the markets, and contractual standing, do not automatically come with the equipment

The dividing line is whether the contracting entity changes. Sell the equity in the company and the qualifications and contracts stay in place. If you sell only the equipment, the record is in principle preserved where the aggregator running the plant is taken over as is, but replace the operating company too and the screening and testing have to be redone by the buyer. The only market where succession of a track record is written into the rules is, for now, the capacity market, and even there it is conditional on prior consent from the national body. A balancing market record is tied to the name of the aggregator that runs the plant, not to the storage plant itself.

Allow for how long the procedures take as well. Simply changing the name on the rights takes roughly three to six months of review by the transmission and distribution utility. In practice the land sale contract and the transfer of rights are signed separately. Separating them lets you value the land and the rights independently.

What you must not do is hold the rights and do nothing. The study response has an expiry date, and once it passes the rights lapse. Neither the money nor the time you put in comes back.

For projects that used a subsidy, selling, leasing or pledging them requires prior approval for 17 years, and repayment can arise. Whether transferring the equity in the SPC amounts to a disposal in substance also needs to be checked case by case. A market for selling off used battery equipment has not developed yet either. How to assess state of health, whether units can be relocated, and whether warranties transfer are all unsettled, so you cannot plan on recovering value by selling the equipment off piecemeal.

Nothing is fixed here. It is a matter of what you settle in the contract and the design. This is an area where little can be said with certainty, so we will be honest about it.

First, decommissioning cost. Solar under the FIT has a system for setting aside disposal costs externally in advance, but it does not apply to storage plants. Handling lithium-ion batteries takes stricter controls than ordinary industrial waste. And at present there is next to no reliable published market rate for what removal costs.

So we cannot give you a figure. What we can give you is how to check. The last item on the pre-purchase checklist, number 43, is whether the decommissioning cost exists: whether there is a removal quotation for that specific project, and whether the contract secures that cost through a reserve or a guarantee. The absence of a quotation is itself a reason to stop.

The classic dangerous structure is a project where the company that owns the plant has capital below the estimated removal cost, with no reserve and no guarantee. In that shape, there is structurally nowhere for the removal money to come from 20 years later.

Next, the question of years. Do not mix up three numbers.

Seventeen years is the number used for calculating tax. It is separate both from how long the equipment lasts and from how long you run the business. And that gap can do real damage. If you wind the business up at around year 10, even under the declining balance method just under 30% of the acquisition cost is still on the books at year 10. On 700-800 million yen of equipment, that is on the order of 200 million yen. If the sale price falls below that remaining balance, the difference is a loss.

Third, how to hold capacity up. The exit from 20 years of operation involves two jobs: adding capacity to make up for degradation, and disposal and recycling once operation ends. Who pays and who is responsible for making up the capacity lost over 20 years, whether through additions or a full replacement, is a design item to write into the contract. A long-term contract model based on a 47 MW extra-high-voltage project assumes degradation of roughly 1% a year, and accepts that a discharge duration of four hours at commissioning falls to the equivalent of three hours by the end of the contract. That said, this is the assumption in one model; another column puts the general industry range at roughly 0.5-3.5% a year.

Fourth, the land. If it is leased, check before you buy that the rent revision clause and the contract term cover the 20-year business period. Also on the list: perfection requirements for the leasehold or superficies, whether there is a mortgage, and the landowner's consent to transfer or sublease.

A 20-year capacity warranty is also just paper if the company giving it does not exist in ten years. For an overseas manufacturer, check the substance and capital of its Japanese entity; for an EPC contractor, check the backing behind its completion guarantee and defect liability. Note also that it is not settled that everything must be removed after 20 years. In an actual case with a 20-year fixed contract, rebuilding after 20 years is positioned as an exit strategy (this is one extra-high-voltage-class case).

Looking for a project is not the first step. There is an order.

First, decide at which stage you will buy. Rights only, a finished plant, or one with a track record. These three are not the same product at different prices. They are different products. Compare prices without deciding the stage first and you end up measuring 200 million yen of rights and an 800 million yen operating plant with the same ruler.

Second, decide the vehicle you buy through. Do you buy the equipment itself, or the equity in the company (SPC) that owns the plant? This is what decides whether market participation qualifications and contractual standing survive or disappear (Q22).

To settle those two, there are four questions to answer.

• Can you take construction risk (do you have the in-house capability to manage an EPC contractor)
• Would you rather pay more money or more time
• Will you change the contracting entity
• If you are after an operating plant, will you change the operating setup

Third, request the documents. For a first-time buyer, the first job is not negotiating the price. It is working out what stage the bundle of documents you have been handed represents, and how solid the rights are. Requests fall into six areas: grid and power contracts, land, laws and permits, equipment and warranties, the site, and contracts and creditworthiness. Concrete items: the original grid connection study response, the register and cadastral map, the development permit, the original capacity warranty, dated site photographs, the removal quotation.

In practice, verification is split into two stages. Before paying a deposit, use the documents alone to establish quickly whether anything is fatal. Immediately before acquisition, go deeper, including the site, physical measurements and creditworthiness. If even one signal to stop is visible, do not pay a deposit.

Separate too what you can do yourself from what you cannot.

You do not have to do all of it yourself. But one distinction is yours to keep hold of as the buyer: what has been confirmed in writing, and what is still only spoken.

Know the rough timescales as well. Developing from bare land yourself: one to three months to find a suitable site, three to 12 months to secure the land and complete resident briefings and administrative checks, six to 18 months for the grid connection study and contract, three to six months for design and procurement, and 12 to 48 months for construction and connection. Two years at the very fastest across the whole process, and normally three to five years. Acquire a project that already has a study response and secured land and you can start at the design and procurement stage, cutting one to three years off the development period. Even when you buy a finished storage plant, it normally takes one to three months from energization to market participation.

If you own land, your entry point is different. You can check for yourself the distance to transmission towers and substations, how much room there is on the nearby grid, and the registered land category and area (Q6). Send us the address or coordinates and the area, and we will come back with an indicative valuation range and the assumptions behind it. There is no charge, and no obligation to sell.

We work on the buyer's side, helping with exactly this kind of assessment. Project-specific information is released after you contact us and a confidentiality agreement is in place. If your question is not on this page, please send it to us anyway.