Technical notes and working papers

Rosenow et al. (2026) compare today’s global energy system with a stylised, highly electrified system that delivers the same useful energy. They report that rejected energy falls by 64%, ambient heat rises from about 2 EJ to 52 EJ, and primary energy falls by 49%, or nearly 300 EJ a year. We accept the central point: primary energy overstates what an electrified system needs, and the gain from electric end uses is real. We reconstruct the undisclosed scenario totals from the paper’s own energy balance and split the saving. Electric end uses, the carrier change, account for 116 EJ named in the paper and a further 43 EJ implied by its balance. Replacing thermal generation with wind and solar, the source change, accounts for 115 EJ. A 26 EJ fall in fuel processing serves both. The authors say the primary-energy figure depends on convention; we size it. Under the substitution convention the Energy Institute used until its 2024 edition, the headline falls by at least 111.2 EJ, more than a third. The paper concludes that its result removes the objection that an electrified system’s primary-energy footprint ‘would be unmanageably large’. Two things that bear on that footprint are held outside the comparison. Manufacturing energy is fixed at today’s level, so the energy to build and rebuild the new system, which is most of the lifetime energy requirement of wind and solar, is not added. And demand for energy services is fixed, so rebound, which one of the authors has written the method neglects, is excluded. We have asked the authors for the figures needed to check the result.