Why cheap equipment can still require expensive electricity contracts

Analysis

The seven connected reports describe different points along a cost chain. Hitachi Energy’s September factory plan concerns future transformer production. ENGIE’s storage total combines operating and construction assets. The Indonesian battery-finance study examines investable contracts. Japan’s October review describes regional network constraints. Fatehgarh concerns turbine supply, the Korean agreement supplier cooperation, and Supernode a storage integration and connection milestone. Their dates and quantities cannot be assembled into a single commissioned-capacity series. An order, factory investment, technical acceptance and electricity sale each have different evidence requirements.

This analysis uses sources published from August through 9 October 2026 and considers decisions during 2026–2029. Historical observations and issuer schedules remain separate from IEA scenarios. Every monetary or operating input in the calculations below is hypothetical, including interest-like discount rates. None is an observed October supplier price, tariff or project return. USD amounts are nominal; inflation, currency translation, tax and a probability model are excluded. The twenty-year recovery horizon is a modelling assumption, not a battery-life warranty or a forecast of future replacement needs.

Begin with an assumed 100 MW battery project, 400 MWh of usable discharge energy and USD 100 million of installed capital. Assume annual fixed operating expense of USD 2 million and twenty equal annual recovery payments. The capital-recovery factor is r / [1 − (1 + r) raised to −20], where r is the assumed annual discount rate. This is a simplified level-payment model for capital and operating expense, not an actual debt structure. Replacement, augmentation, residual value and construction-period financing must be added separately if relevant.

At assumed rates of 7%, 9% and 11%, annual capital recovery is about USD 9.4393 million, USD 10.9546 million and USD 12.5576 million. Add the same USD 2 million operating expense to each. If only 95% of the 100,000 kW rating qualifies for payment, the required annual charge per eligible kilowatt becomes USD 120.41, USD 136.36 and USD 153.24 respectively. These are calculated cost-recovery requirements, not market availability payments, investor profits or Indonesian tariff estimates. The discount rate is not a loan coupon; it stands for the assumed valuation requirement of the complete capital.

The IEA’s Indonesia study reports a battery weighted average cost of capital assessment of 9–11%. Our scenarios deliberately show sensitivity around a range of financing conditions without calibrating to any specific project. Contract currency, counterparties, termination provisions and degradation obligations can change perceived risk even if the equipment quote is unchanged. A lower nominal rate also cannot automatically be compared across currencies when inflation, hedging and settlement terms differ. The important comparison holds asset scope and performance constant before varying financing.

Procurement shocks can then be isolated. Within the assumed USD 100 million budget, define non-overlapping exposures of USD 10 million to a metal-price component, USD 4 million to freight, and USD 15 million to an imported package’s currency conversion. Suppose the respective changes are +25%, +50% and +10%. They add USD 2.5 million, USD 2 million and USD 1.5 million, making capital USD 106 million. With the same 9% recovery, USD 2 million operating expense and 95,000 eligible kW, the charge rises to USD 143.28/kW/year. No actual copper, steel, freight or exchange-rate move is claimed. A fixed-price contract or hedge can redirect exposure rather than remove the underlying cost.

Availability must not be confused with energy throughput. Assume separately 250 full discharge cycles a year and a 95% throughput-availability multiplier. At 400 MWh usable per cycle, delivery is 95,000 MWh/year. If retained usable energy falls hypothetically to 90% while the other inputs stay fixed, it becomes 85,500 MWh/year. An illustrative USD 10 million annual charge therefore spreads over energy at USD 105.26/MWh initially and USD 116.96/MWh after that retention change. This separate energy denominator is not the availability-payment model above, and the two charges must not be added as if they were independent costs.

At a separately assumed 85% round-trip efficiency measured at the same electrical boundary, supplying 95,000 MWh requires approximately 111,764.71 MWh of charging energy. The purchase bill depends on when that charging occurs, not just an annual average. A battery committed to system support may need reserves that reduce trading opportunities. Supernode’s grid-forming and performance-standards disclosure illustrates why the contract must identify these duties. Its published MW and MWh do not provide the guaranteed usable energy, price spread or lifetime throughput needed to calculate the project’s profit.

Transmission has its own boundary. For an entirely hypothetical 1,000 MW link, assume 3,000 scheduled transfer hours, 95% availability and 3% end-to-end losses. Energy delivered is 1,000 × 3,000 × 0.95 × 0.97 = 2,764,500 MWh. An assumed value difference of USD 20/MWh across the link yields USD 55.29 million before transmission charges, operating expense and capital recovery. This is a gross value illustration, not congestion rent or a guaranteed arbitrage income. Neither Korea’s corporate agreement nor Japan’s frequency split supplies those assumed inputs.

Fuel-based alternatives require the same discipline. Assume a thermal plant consumes 7 GJ of gas per delivered MWh at an assumed USD 10/GJ. Fuel alone costs USD 70/MWh; at 6.5 GJ/MWh it costs USD 65/MWh. This illustrates heat-rate sensitivity, not a reported Japanese gas price or plant efficiency. Startup fuel, part-load operation, maintenance, network charges and emissions costs remain outside the calculation. Comparing that fuel component directly with a renewable or storage capital component would mix accounting scopes; compare complete incremental services over the same dispatch period instead.

Delays can outweigh an invoice saving. Suppose average already-drawn capital is USD 60 million during a six-month delay and simple carrying cost is 9% annually. Financing exposure adds USD 2.7 million before lost receipts or penalties. It is not calculated on the entire project budget unless that budget is actually drawn. Hitachi’s planned 2029 production and Fatehgarh’s expected starting deliveries demonstrate why supplier and customer milestones need alignment. Factory completion, turbine delivery and network acceptance are separate critical-path events; one cannot establish the others.

Alternatives should be tested against the same required service. A battery, network reinforcement, demand response or redispatch can resolve different combinations of duration, location and stability. Their benefits can overlap, so avoided curtailment and service revenue should not both receive full credit for the same constrained energy. Useful evidence includes dated delivered-equipment offers, payment milestones, connection limits, loss boundaries, usable-energy guarantees, dispatch rights and matched revenue contracts. The seven disclosures establish dated opportunities and constraints; an auditable investment comparison connects each cost to accepted equipment, eligible service and cash actually collectible.