Norway sells electric vehicles to 96% of new car buyers. Its EV fleet now represents roughly 35% of all cars on the road. And yet, because internal combustion vehicles waste so much of the fuel they consume, the electricity used by those EVs accounts for only 14% of the Final Energy used by Norwegian road transport. If the country that has done more than any other to electrify its passenger car fleet cannot get halfway to the COP31 target using the proposed metric, the metric itself deserves serious scrutiny.
That is the core problem with the “35 by 35” announcement made by Australia’s Minister of Climate Change and Energy, Chris Bowen, co-President of COP31, at the Bonn preparatory meeting on June 9. The target asks for electricity to represent 35% of Final Energy by 2035. The direction of travel is right. The measurement framework is not.
Why Final Energy Distorts the Picture
Final Energy is the energy content of whatever is sold to end users, including all the energy they waste. It is not a measure of what actually powers the economy. When a petrol-powered car consumes 100 units of Final Energy, only around 20 units of that go into moving the vehicle. The other 80 are lost as heat. An electric vehicle covering the same distance uses roughly 22 units of electricity, nearly all of which drives the wheels.
This efficiency differential creates a mathematical trap in the Final Energy metric. When you replace a fossil technology with an electric one, you reduce both the numerator and the denominator of the ratio. Because each unit of electrification can reduce Final Energy by three or more units, the dominant effect at low penetration levels is the shrinkage of the denominator. The ratio barely moves. Then, as electric penetration rises further, the ratio accelerates sharply. The result is not a linear relationship between physical electrification and the metric used to measure it, but a curve that misleads at both ends.
The numbers from the road transport sector illustrate the problem precisely. Reaching 35% EV penetration of the vehicle fleet electrifies only around 11% of that sector’s Final Energy, depending on the relative efficiencies of the two drivetrains. For heat pumps with a Coefficient of Performance of three, the analogous calculation is slightly less stark but still damaging: 35% penetration of space heating by heat pumps corresponds to roughly 14% electrification as a share of Final Energy. These are not edge cases. They describe the mainstream deployment scenarios around which climate policy is built.
The Arithmetic of a 2035 Deadline
The temporal dimension compounds the difficulty. To reach 35% of Final Energy from electricity in road transport by 2035 would require approximately 70% of vehicles on the road to be electric by that date. Given that just under half of the cars that will be registered in 2035 have already been manufactured and sold, this is arithmetically out of reach for virtually every country on earth. A policy target that cannot be translated into physically achievable sectoral and national goals does not constitute a target in any useful sense.
This is not a peripheral technical concern. It goes to the core of whether the COP process can set goals that actually guide investment, industrial policy, and infrastructure planning. China has electrified 10% of its economy per decade for the past two decades and now leads both the United States and Europe on electricity’s share of Final Energy, having started from significantly behind. Europe and the US, which electrified around 4% of their economies per decade between 1950 and 2010, have since stalled. A metric that obscures how much physical electrification is actually occurring, and that delivers misleading signals about progress, is a liability at precisely the moment when the pace of deployment needs to accelerate.
What Should Be Measured Instead
The alternative that most directly addresses the distortion is penetration of clean electric technologies by sector. This metric tracks what is actually happening in the physical economy: what share of vehicles are electric, what share of heating systems are heat pumps, and what share of industrial process heat below certain temperature thresholds has been electrified. Penetration rates follow well-understood S-curve dynamics, are comparable across countries, and can be mapped directly onto investment and manufacturing timelines.
A second alternative is electrification as a share of Useful Energy rather than Final Energy. Useful Energy is the portion of purchased energy that actually performs economic work: propelling vehicles, heating spaces to the required temperature, and driving industrial processes. It is the quantity that actually powers the economy, and it is the denominator that neither Primary Energy nor Final Energy correctly represents. At 35% of Useful Energy, the target would be genuinely ambitious and would reflect real progress rather than a statistical artefact.
The obstacle is that no authoritative, up-to-date, internationally comparable Useful Energy dataset currently exists. The IIASA Primary, Final, and Useful Energy Database terminates in 2014. A country-level database produced by a team at the University of Leeds covers 1970 to 2020 but lacks the institutional authority needed to anchor a UN-level target. The IEA maintains an Energy End-Uses and Efficiency Indicators database that is updated annually and is structurally close to what would be needed, but it has not been used to produce a comprehensive Useful Energy dataset.
A Data Gap With Policy Consequences
The absence of authoritative Useful Energy data is not a minor methodological shortcoming. It means that energy ministers, investors, and multilateral institutions are making large-scale decisions about the energy transition without a reliable measure of what the transition is actually achieving. The question of what percentage of the heat warming homes in any given country comes from clean electricity versus fossil fuels versus ambient heat harvested by heat pumps cannot be answered from publicly available data without constructing bespoke calculations that carry no institutional authority.
Defining Useful Energy is also non-trivial, and the definitional choices matter. Whether the Useful Energy of an AI data centre should be measured as the minimum thermodynamic requirement of the computation, the total power drawn by GPUs, or some figure that includes cooling load is not a question with a single correct answer. Useful Energy also shifts over time as technologies improve. These complexities are tractable, but they require methodological decisions that need to be transparent, peer-reviewed, and consistently applied across jurisdictions over time.
The practical implication is that the IEA, with its annual update cycle and established relationships with member country statistical agencies, is the natural institution to produce this data. The case for member countries formally requesting that the IEA incorporate a comprehensive Useful Energy dataset into its World Energy Outlook is straightforward. The WEO already provides the authoritative global reference for energy supply data. Extending its scope to demand-side Useful Energy would give policymakers the measurement infrastructure needed to set targets that are specific, achievable, and verifiable by sector and country.
Space heating, land transport, and low- to mid-temperature industrial heat together account for over 40% of global emissions, and electric alternatives in each of these sectors are at or approaching cost parity with incumbent fossil technologies. The policy and investment case for electrification in these areas does not depend on the 35 by 35 framing. But if the COP process is going to anchor global momentum around an electrification target, that target needs to reflect physical reality rather than a ratio whose mathematical properties actively obscure the pace of the transition.
