Building before demand: who carries the risk in France, South Africa and Chile

Serguey Asael Shinder13 October 2026, 06:542 min read

Four recent announcements from three countries look like separate stories: a French capacity auction for 2030-31, French grid works in an industrial zone, a South African utility looking for partners, and a Chilean battery on a coal site. Each is an answer to one question — who carries the risk of building capacity before it is certain to be needed.

France, capacity: the consumer, carefully. CRE's proposal for the first long-term capacity auction targets 82 GW for winter 2030-31, with only 3 GW eligible for multi-year contracts, and keeps both price caps where they are. Its reasoning is explicit: RTE sees a moderate supply risk and a need for 0 to 2.5 GW of new capacity, so the regulator chooses a prudent approach to limit the cost borne by consumers. Here the risk of building too much is judged larger than the risk of building too little.

France, connections: the network operator, ahead of time. At Fos-Lavéra the same regulator lets RTE build 2,500 MW of hosting capacity before every connection is firm, with works starting at once because requests are numerous. Connecting industrial users pay a set share later. The risk of stranded assets moves from the first applicant to the network operator, under regulatory control — the opposite choice from the capacity auction, because here the cost of waiting is industry that cannot electrify.

South Africa: shared with private capital. Eskom's RFQ puts at least 2 GW of solar and 1 GW / 4 GWh of batteries on its own land, with grid access in progress, and asks partners to co-develop and co-finance them in ring-fenced project companies, mostly on a limited- or non-recourse basis. The utility contributes sites and grid position; partners contribute capital and delivery risk.

Chile: the incumbent, on its own balance sheet. ENGIE's 142 MW / 802 MWh BESS Lile reuses the substation and land of a thermal complex whose coal units were shut at the end of last year. The cost — 174 million dollars — and the bet that the system will pay for five-hour storage sit with the company.

What the comparison shows. None of these is the "right" model; each fits a different uncertainty. Where demand is doubtful (French capacity), regulators protect consumers by buying little and capping prices. Where demand is visible but the grid is late (Fos-Lavéra), they let the network build first. Where the utility has land and connections but not money (Eskom), it shares the project. Where a company already holds the site (Mejillones), it carries the risk itself. For anyone tracking energy transition timetables, the question to ask of any announcement is not only how many gigawatts, but whose balance sheet pays if they arrive early.

Sources: CRE, 2030-2031 capacity auction — https://www.cre.fr/actualites/toute-lactualite/la-commission-de-regulation-de-lenergie-publie-ses-deliberations-relatives-au-parametrage-de-lenchere-de-long-terme-2030-2031-du-mecanisme-de-capacite.html ; CRE, Fos-Lavéra — https://www.cre.fr/actualites/toute-lactualite/la-cre-valide-lanticipation-et-la-mutualisation-des-travaux-de-raccordement-des-consommateurs-de-la-zone-de-fos-lavera.html ; Eskom — https://www.eskom.co.za/eskom-holdings-launches-strategic-partners-selection-process-for-up-to-6gw-utility-scale-renewable-energy-projects/ ; ENGIE Chile — https://www.engie.cl/concluimos-la-primera-etapa-de-la-reconversion-del-complejo-energetico-mejillones/