← Back to the entry point (all 19 questions)

It does. But at midday it is squeezed, a thermal unit that has stopped cannot move, and the fleet is shrinking. What is left uncovered comes round to batteries.

Thermal generators carry a function that automatically raises output slightly when frequency falls and lowers it when frequency rises (governor-free). Because it acts without waiting for instruction, thermal remains the lead provider of fast balancing power. In the online bracket of the fast product, roughly 70% of awarded volume is thermal (47-4).

So can we simply ask thermal to do more? Three reasons say no.

1. At midday, thermal is already squeezed

The order in which resources are curtailed when electricity is in surplus is set by the state. Thermal goes first. On a sunny holiday midday, solar exceeds demand, so thermal is squeezed to minimum output and stopped altogether if surplus remains. A thermal unit that has stopped cannot restore frequency. Under its end-2023 package the state asks existing thermal units to lower minimum output voluntarily from 50% to 30% — but the further they are squeezed, the smaller the band left available for balancing.

← scroll sideways →

The surplus = curtail solarConsumption (demand)Nuclear: no daytime throttlingThermal: squeezed to minimumSolar00:0006:0012:0018:0024:00One sunny holiday (conceptual. Kyushu's solar curtailment rate was 4.8% in FY2024 and is forecast at 6.1% in FY2025)This is how the midday surplus forms
Fig. 1 — Midday on a sunny holiday. Thermal is squeezed to minimum, and whatever surplus remains is met by curtailing solar. A thermal unit squeezed that far has almost no headroom left for balancing (conceptual).

2. The fleet is shrinking

Thermal capacity fell by roughly 16 GW over the seven years from FY2016 to FY2023 — the net of oil-fired down about 22 GW, LNG down about 3 GW and coal up about 9 GW. The supply plans show 6.02 GW of thermal suspended or retired in FY2025 and 3.71 GW in FY2026, with suspensions and retirements expected to keep outpacing new and expanded capacity from FY2025 onward.

← scroll sideways →

Thermal suspensions and retirements (volume stopping in each fiscal year, GW)16 GWdecline, FY2016 → FY20236.02 GWFY20253.71 GWFY2026Source: Agency for Natural Resources and Energy, Basic Policy Subcommittee, Material 1 (July 2024)OCCTO, Compilation of Supply Plans (March 2025 and March 2026)
Fig. 2 — The thermal fleet keeps shrinking. Down roughly 16 GW from FY2016 to FY2023, with 6.02 GW suspended or retired in FY2025 and 3.71 GW in FY2026.

3. Standby costs money

Keeping a thermal unit ready to run at any time for balancing purposes costs money — staff, fuel and maintenance — even when it is not running. In Japan the capacity market pays for that; abroad, states and national governments have started paying directly. In May 2024 New South Wales, Australia, agreed to postpone the closure of Eraring, the state's largest coal plant (2.88 GW), from August 2025 to August 2027, under a contract in which the state absorbs 80% of any losses (up to A$225 million a year). In January 2026 the operator, Origin Energy, announced a further extension to April 2029. In 2026 Germany agreed with the EU on a plan to secure 12 GW of new capacity, mainly gas-fired, through auction (Reuters, 15 January 2026).

You can have thermal do it. But at midday it is squeezed, the fleet is shrinking, and keeping it costs money. Those three together are the pressure moving the balancing job towards batteries.

How an investor should read this
While thermal leads the balancing market, battery unit prices are held down by thermal's bids (roughly 70% of primary online awards go to thermal). But close to 6 GW of thermal is being suspended or retired across FY2025 and FY2026, and we have entered an era in which keeping it requires payment from a national or state government. Australia has a state covering losses; Japan supports it through the capacity market. On a ten-year view, the balancing seats are moving from thermal to batteries.
The questions in this series
I. Why the grid ran without batteries in the past
  1. 48-2The grid used to run fine without batteries, didn't it?
  2. 48-3What disappears from a grid with fewer spinning machines?
  3. 48-4Why "ten seconds"?
  4. 48-5Does more renewables mean more balancing power is needed?
II. Why thermal, nuclear and pumped storage are not enough
  1. 48-6Can't thermal just do the balancing? (this article)
  2. 48-7If nuclear grows, do we stop needing batteries?
  3. 48-8Wouldn't more pumped storage be enough?
  4. 48-9Where does the balancing power that the market failed to buy come from?
III. Where demand and the generation mix are heading
  1. 48-10Why do data centres point to batteries?
  2. 48-11What happens to solar from here?
  3. 48-12Thermal volume, nuclear operation, demand — what next?
  4. 48-13How much battery capacity is actually coming?
IV. The money, and what the state really wants
  1. 48-14Does the government actually want more batteries?
  2. 48-15What does battery storage resemble as an infrastructure investment?
  3. 48-16They say the earnings will thin out. Is that true?
  4. 48-17Can you copy the overseas playbook and make money in Japan?
  5. 48-18How much battery storage will Japan ultimately need?
  6. 48-19Is a battery the same wherever you put it?
The one question beyond the 1848-20 What do you pay, and what comes back?The answer differs by project, so we have built a calculator whose dials are made only from published figures. You can move unit price, clearing rate, seat fee, arbitrage spread and construction cost yourself.

Sources

Supervised by
Shinya Nakashima(Representative Director, Science X Inc.; Ph.D. in Engineering)

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