← Back to Knowledge

With the price cap moving from ¥15 to ¥10 and ¥7.21 named as the next step, the question becomes how far it goes after that. Some people say "down to thermal's ¥3." Follow the data and the floor is set not by the price of thermal but by what the battery itself could earn elsewhere. And that amount is not a single number: it differs by time of day.

This page deals only with price. The size of the room (procurement volume) is on 47-2, and the volume arriving is on 47-3.

1. The market moved before the cap came down

Apr May Jun Jul Aug Sep 0 2 4 6 8 10 12 14 16 JPY/ΔkW·30min (award-weighted average) Kyushu ¥4.9 (cap still ¥15) Figure 1 Primary offline price: before the cap fell, bids brought Kyushu down (2026) Price cap Tohoku Tokyo Japan Chubu Kyushu

Figure 1 is the actual clearing price of primary offline, the battery seat (the average weighted by awarded volume). From April to June a bid reaching the ¥15 cap was there in almost every slot, and the nationwide average sat around ¥12 without moving. Bidding was short, so a bid placed at the cap still cleared.

From July it changes. In Kyushu, with the cap still at ¥15, August fell as far as ¥4.9. Chubu and Kansai came into the ¥9 range. Meanwhile, ¥11–14 is where Tokyo and Tohoku stayed. As 47-3 showed, the battery seat was still half open even in August. Even so, area by area as bidding grew, the price began to be set by bids rather than by the cap. The way bids are placed changed without waiting for the seats to fill.

When the cap came down to ¥10 on 1 September, the nationwide average was ¥6.8. Tokyo at ¥8.6 and Tohoku at ¥9.4 sit just under the new cap, while Kyushu at ¥3.4, Chugoku at ¥4.6 and Kansai at ¥5.1 sit away from it. At the same time as the rules lowered the ceiling, the market itself has begun to set the price.

2. The floor is "what the battery could earn elsewhere"

← scroll horizontally →

0h 6h 12h 18h 24h 0 5 10 15 20 25 30 JPY/kWh This gap "what batteries earn elsewhere" Shape of a JEPX day (FY2026 Apr–Sep average, system price) 8 cheap slots: charge (buy) 8 top slots: discharge (sell) 0 2 4 6 8 10 JPY/ΔkW·30min Tohoku Hokkaido Kyushu Shikoku Tokyo Chubu Chugoku Hokuriku Kansai 8.5 1.4 8.2 1.4 6.9 1.2 6.8 1.1 6.6 1.1 6.6 1.1 6.4 1.1 5.9 1.0 5.9 1.0 ¥7.21 Converted to ΔkW (4-hour battery, by area, FY2026) 8 evening slots Over 48 slots Figure 2 The battery floor is what JEPX pays: ¥6–8 in the evening, ~¥1 over a day

In the hours it is not offered into the balancing market, a battery can earn on JEPX, the wholesale power market: charge in the cheap hours and discharge in the expensive ones, which is arbitrage. Selling ΔkW in the balancing market means giving up the arbitrage in those hours. So the lowest price a battery puts on ΔkW is what it could earn from arbitrage at that time. Economists abroad call this opportunity cost (what the battery could earn elsewhere).

The left of Figure 2 is the shape of a JEPX day in FY2026: cheap at midday, expensive in the evening. With a 4-hour battery charging in the 8 cheap slots and discharging in the 8 expensive ones, that difference is "what it could earn elsewhere."

The right converts that into ΔkW units. Looking only at the 8 evening slots, it is ¥6–8 depending on the area. Tohoku and Hokkaido are in the ¥8 range, above the next cap of ¥7.21. Averaged across 48 slots, on the other hand, it is in the ¥1 range. Late at night there is no arbitrage opportunity, so the floor in those hours approaches zero.

So the floor is not one line but the shape of a day: high in the evening, low late at night.

3. The areas where bidding grew are already in that shape

0h 6h 12h 18h 24h 0 2 4 6 8 10 12 14 16 JPY/ΔkW·30min (award-weighted average) Cap ¥15 (as of August) Figure 3 Full areas took the shape "high by day, cheap by night" (primary offline, August 2026) Tokyo Kansai Chugoku Kyushu

Figure 3 lines up the primary offline price for August 2026 by time of day. Tokyo is almost flat across all slots at ¥11–12. Kansai, Chugoku and Kyushu, where bidding grew, have taken the shape of high by day and cheap by night. Kansai is in the ¥11 range by day and around ¥6 late at night.

It is high by day because that is when the battery is busy charging. It is cheap at night because there is nothing else to do. When bidding grows, the price does not collapse uniformly; it takes the shape of the battery's own day. What happened overseas is happening the same way in Japan's 30-minute slots.

The median of the lowest awarded price (the cheapest bid to clear in that hour) is ¥2.4–3.0 in Kyushu, ¥1.4–2.5 in Kansai and ¥2.6–3.5 in Tokyo. That sits exactly between Figure 2's "averaged across 48 slots" (¥1.0–1.4) and its "evening" figure (¥6–8).

4. The ladder of ceiling and floor

0 2 4 6 8 10 12 14 16 JPY/ΔkW·30min Price cap (from 9/1) Next step (conditional) Battery's evening earnings elsewhere (¥6–8 by area) Kyushu actual (August, cap still ¥15) Median lowest award (¥1.4–3.5) Battery's daily-average earnings elsewhere (¥1.0–1.4) Pumped bilateral (¥0.2–0.7, per 30min) Thermal (start-up settled ex post; can bid near zero) Figure 4 Price ladder: ceiling by rules, floor by the battery itself Ceiling set by rules Battery floor (set by itself) Floor outside the market (TSO's options)

Figure 4 puts the ceiling and the floor on one ladder.

The ceiling is set by the rules: a cap of ¥10, with ¥7.21 as the next step. The indicators and the timing of the decision are as written on 47-2.

The battery's floor is set by the battery itself: ¥6–8 in the evening, in the ¥1 range averaged over a day. Kyushu's ¥4.9 in August and the median lowest award of ¥1.4–3.5 both fall in between.

There is a floor outside the market as well. The bilateral contract under which a transmission and distribution operator (TSO) contracts pumped storage directly ran at ¥0.47–1.34 per kW per hour on FY2025 results, or ¥0.2–0.7 on a 30-minute basis, and in FY2026 5 companies locked up 266 × 10,000 kW in total. Since April 2025 thermal has been able to settle start-up costs ex post, so it no longer has to load start-up costs onto the ΔkW price and can bid lower. In practice 70% of primary online awards are thermal, and the nationwide average in August was ¥4.4. Governor-free response only leaves a little headroom on a plant that is already running, so the ΔkW bid can be cheap. You see it said that thermal cannot bid below ¥10, but that does not match the record.

There is no single floor at "thermal's ¥3." Thermal can go down to a price with start-up costs taken out, and a battery in the evening stops higher than that.

5. How it landed overseas

← scroll horizontally →

2022 2023 0 25 50 75 100 125 150 175 Frequency response revenue (£k/MW-year) 150 50 Floor was not provision cost (£0.5–1.5) stopped at opportunity cost £5/MW/h UK Jul 2025 Jul 2026 0 20 40 60 80 100 aFRR share of revenue (%) 91 52 Below thermal, above its own opportunity cost clearing level (still pre-saturation) Germany 2023 2025 0 20 40 60 80 100 Ancillary services revenue share (%) 84 48 Ancillary service prices fell to the level of arbitrage earnings US ERCOT 2022 2023 2024 0 20 40 60 80 100 120 Battery revenue ($/kW-year) 103 78 53 In 2024, 82% of revenue came from energy (arbitrage) US CAISO Figure 5 4 overseas markets: after saturation ΔkW revenue fell, shifting to arbitrage and capacity (batteries stayed in use)

There are 4 markets overseas where batteries have saturated the balancing capacity market. What they have in common is that the price floor was set not by "the cost of the cheapest provider" but by the batteries' own opportunity cost.

In the UK, frequency response revenue fell from £150k/MW-year in 2022 to about £50k in 2023, but the price did not fall to the cost of provision (£0.5–1.5/MW/h); it stopped at about £5/MW/h, the opportunity cost (Modo Energy estimate). In CAISO in the US, battery revenue fell from $103/kW-year in 2022 to $53 in 2024, and 82% of it came from energy (arbitrage) (CAISO Department of Market Monitoring). In ERCOT, the ancillary services share of revenue went from 84% to 48% (Enverus estimate). In every case batteries kept being used, and the center of revenue moved to arbitrage and capacity. System operators in each country are adding new products (Quick Reserve and Balancing Reserve in the UK, ECRS in ERCOT).

The closest to Japan is Germany. Batteries there clear below thermal but above their own opportunity cost, still short of saturation — the state the analyst Modo Energy describes overlaps with Japan today, clearing near the cap while beginning to feel for the floor.

6. Where Japan differs from overseas

There are three.

The ceiling bites first. Overseas the fall was free, with no cap; in Japan the rules are lowering the ceiling in steps. With less room between ceiling and floor, price movement is gentler.

30-minute slots. The UK and Germany use 4-hour blocks; Japan uses 30 minutes. The market catches up faster with the floor for each part of the day. That is why the shape in Figure 3 appeared within 1 to 2 months of saturation.

There is a floor outside the market. The bilateral contract for pumped storage and the spare-capacity utilization contract are mechanisms with no overseas equivalent, and the higher the market, the more procurement goes outside it. Conversely, when bidding thickens, some areas move back to market procurement, as Hokkaido has.

One more thing, a little further out, is the co-optimized market. The direction under review is a shift in the first half of the 2030s to a mechanism that clears kWh and ΔkW together. There the concepts of a price cap and a procurement volume are themselves replaced by optimized clearing based on lost profit. It is a world in which "what the battery could earn elsewhere" becomes the market's pricing as it stands.

7. What an owner can do

If the price is not one line but the shape of a day, the way you offer does not have to be one line either.

Summary

Back to the overview (COLUMN 47).

Reference tables

Reference table 1 Primary offline clearing price (average weighted by awarded volume, JPY/ΔkW·30min, 2026)

AreaAprilJuneAugustSeptember (1–7)
Hokkaido3.95.34.01.5
Tohoku13.513.313.69.4
Tokyo11.512.111.58.6
Chubu14.613.39.06.3
Hokuriku14.512.78.85.4
Kansai13.312.59.35.1
Chugoku12.210.16.64.6
Shikoku3.73.24.04.0
Kyushu12.17.54.93.4
Nationwide12.311.79.66.8

Source: our calculation from EPRX trading results (preliminary). The price cap was ¥15 through August and ¥10 from 1 September. The aggregation window differs from the August figure in COLUMN 41 (nationwide ¥9.80, deliveries through 8/17)

Reference table 2 JEPX arbitrage converted into ΔkW (4-hour battery, top 8 slots and bottom 8 slots, efficiency 0.85, April–September FY2026)

AreaEvening 8 slots (JPY/ΔkW·30min)48-slot average
Tohoku8.51.4
Hokkaido8.21.4
Kyushu6.91.2
Shikoku6.81.1
Tokyo6.61.1
Chubu6.61.1
Chugoku6.41.1
Hokuriku5.91.0
Kansai5.91.0

Source: our calculation from JEPX spot market area prices. Wheeling charges, degradation costs and fees are not included. Deducting the volumetric wheeling charge on the charging side (extra-high voltage, excluding tax; on the discharging side the energy charge is exempted under special measures in each company's terms) brings the evening 8 slots down to ¥5.5–8.0 and the 48-slot average to ¥0.9–1.3

Notes: September in Figures 1 and 3 and in Reference table 1 is preliminary for 1–7 September. The JEPX shape on the left of Figure 2 is the April–September FY2026 average of the system price. The pumped storage bilateral contract unit prices in Figure 4 are FY2025 results converted to a 30-minute basis. The overseas figures are from published materials of each market's operator and monitor and from Modo Energy's analysis (details in Sources).

Sources

Supervised by
Shinya Nakashima, Ph.D. (Eng.) — Representative Director, Science X Inc.

Works on the development, sale and technical due diligence of grid-scale battery storage plants in Japan. This column is written and supervised on the basis of that hands-on transaction and due-diligence practice.

Energy Business Consultation

Project-specific information not covered in our articles
will be disclosed after NDA execution following your inquiry.

Contact Us →