Ten physical realities the energy transition must tackle

To transform the energy system would require the substitution of billions of physical assets. Under McKinsey’s 2023 Achieved Commitments scenario, to reach stated national climate commitments, about one billion EVs, more than 1.5 billion heat pumps, and about 35 terawatts of low-emissions power generation capacity would need to be deployed by 2050, for instance.¹This scenario defines progress made to date and the magnitude of the transformation needed, with detail across different economies and types of assets about the deployment levels that would be required for those economies to meet the climate commitments they have made (as of the end of 2022). This scenario assumes that countries that have committed to net zero (some by 2050, some later) meet those commitments and that warming reaches 1.6ºC relative to preindustrial levels by 2100. See Global energy perspective 2023, McKinsey, October 2023. This scenario (and, in some cases, other similar scenarios) is used here not as a projection, but to understand progress to date on the transition and the physical challenges to overcome. Other net-zero scenarios may contain slightly different combinations of technologies and rates of deployment, but the broad trends and themes described in this research would still apply. This report is based on analysis as of September 2023. Subsequent developments in the energy system may lead to different outcomes, which will be covered in forthcoming McKinsey research.

There has been momentum toward that deployment. For instance, almost 90 percent of all passenger BEV sales and 60 percent of all solar and wind power capacity additions are estimated to have happened in the past five years alone.²Renewable energy statistics 2024, International Renewable Energy Agency, July 2024; Lauren Holtmeier, “Global renewables capacity increased by new record in 2024, IRENA boss says,” S&P Global, January 12, 2025; Global EV outlook 2024, International Energy Agency, April 2024; and data from EV Volumes (2024). Note that all passenger BEV sales from 2010 to 2024 (including only cars) are accounted for in the calculation of how many passenger BEV sales occurred in the past five years. But thus far deployment of low-emissions technologies is only at about 10 percent of the levels required by 2050 to meet global commitments in most areas—and far less than that in others (Exhibit 2). For instance, less than 1 percent of the 90 million tonnes of hydrogen produced today comes from low-emissions production. Overall, therefore, the energy transition is in its early stages.³See also, “The energy transition: Where are we, really?,” McKinsey, August 2024.

The heating and cooling needs of buildings account for almost 85 percent of total CO₂ emitted from buildings, with space heating and water heating responsible for more than 75 percent.¹Total CO₂ emissions for buildings are allocated as follows: 54 percent for space heating, 22 percent for water heating, 9 percent for appliances, 8 percent for space cooling, 5 percent for lighting, and 2 percent for cooking. See Brodie Boland, Shailesh Lekhwani, Sebastian Reiter, and Erik Sjödin, Building value by decarbonizing the built environment, McKinsey, June 2023.

The need for heat is currently largely met by burning fossil fuels—for example, in gas boilers. Fossil fuels could be replaced by using electric options. Heat pumps are highly efficient heating technologies and the main option being explored in most markets.

But sweeping electrification of heating in buildings will only add another layer of demand to the power system. Demand for electricity would spike—sharply—during the coldest hours of the coldest days of the year when many people rush to turn the heating on at the same time. In the United States, for instance, peak demand would shift from the summer, when many buildings use air conditioning, to the winter as heat pumps spread.²Michael Waite and Vijay Modi, “Electricity load implications of space heating decarbonization pathways,” Joule, volume 4, issue 2, 2020.

Overall growth in demand implies that the entire power system would need more capacity. Let’s take the United States as an example of what this might look like. If and when heat pumps spread, peak demand (the largest amount of power that is ever required during the course of a year) may rise well beyond the peak capacity of today’s grid. In one scenario in which heating in buildings is fully electrified, external research has estimated that peak power demand could be 1.7 times today’s peak across the United States.³Michael Waite and Vijay Modi, “Electricity load implications of space heating decarbonization pathways,” Joule, volume 4, issue 2, 2020. This scenario assumes that every other source of demand is held constant and that no additional measures to balance supply and demand are put in place.

In colder regions, this effect could be even more pronounced. In New England, for example, peak demand could be three times higher than today. Remember, too, that heat pumps are currently less efficient at colder temperatures. For example, when temperatures drop from 5ºC to minus 10ºC, the coefficient of performance of standard heat pumps almost halves.⁴Duncan Gibb et al., “Coming in from the cold: Heat pump efficiency at low temperatures,” Joule, volume 7, issue 9, September 2023. The efficiency of a heat pump is measured by the coefficient of performance (COP), which typically ranges from two to five for heat pumps, meaning that for every unit of electrical energy consumed, the heat pump is able to deliver two to five units of heat. For further context, see Understanding COP: Coefficient of performance of heat pumps, Learn Metrics, accessed May 2024.

Action to minimize peaks—and therefore how much power capacity is required in the system—could be implemented, through a combination of more efficient models of heat pumps, more use of heating technologies that combine electrification with other options (so called dual-fuel systems and other technologies) to limit use of electricity on the very coldest days, or even smoothing the demand for power by shifting demand for heating to different times of the day by combining heat pumps with thermal energy storage.⁵Demand-side solutions to winter peaks and constraints, ACEEE, April 15, 2021. Demand-side measures include, for instance, improvements in thermal envelope, intelligent thermal control, space-conditioning load shedding, and connected hot-water heaters.

The deployment of passenger BEVs is increasing, but the extent to which they save on CO₂ in comparison with vehicles powered by internal combustion engines (ICEs) varies. Although passenger BEVs can have lower running emissions per kilometer than ICEs, they have higher emissions when they are being manufactured. As a result, how much is saved depends on how clean the grid that powers them is (Exhibit 5).

Where grids are already relatively clean, as in the European Union, small and midsize BEVs already save 45 to 65 percent of emissions in comparison with ICEs over the lifetime of the vehicle, even if the grid did not decarbonize further.¹The R&D GREET model (Software v1.3.0.13991) developed by Argonne National Laboratory was used to determine the vehicle manufacturing CO₂-equivalent (CO₂-e) emissions and the well-to-pump CO₂-e emissions of ICEs. Data inputs for tailpipe CO₂-e emissions of vehicles were adjusted using European Environmental Agency and US Department of Energy data. BEVs have no direct tailpipe or well-to-pump emissions—their impact is measured by emissions from the electrical grid, known as grid intensity. The 2022 emissions intensity data come from Climate Transparency report 2022, Climate Transparency, October 2022. Future grid intensity projections come from Global energy perspective 2023, McKinsey, October 2023. The McKinsey Center for Future Mobility tracks manufacturing emissions associated with BEVs. In India, however, about 75 percent of all electricity is generated today using fossil fuels, notably coal.²Global energy perspective 2023, McKinsey, October 2023. Absent any change to the emissions intensity of its grid, passenger BEVs could end up emitting more rather than less than top-performing ICEs, and only somewhat less than even average ICEs.

Of course, grids are decarbonizing, and if India’s grid were to decarbonize in line with McKinsey’s 2023 Achieved Commitments scenario, a midsize BEV purchased today in India could achieve lifetime carbon savings of as much as 15 percent against even a top-performing ICE. Continuing to improve on grid emissions intensity is therefore a critical factor in reducing overall emissions from a transition to BEVs.

The four pillars of modern civilization—steel, cement, plastics, and ammonia—alone account for about two-thirds of industrial emissions.¹Vaclav Smil, “The modern world can’t exist without these four ingredients. They all require fossil fuels,” Time, May 12, 2022; and Global energy perspective 2023, McKinsey, October 2023.

These industries are particularly hard to fully decarbonize, because they rely heavily on fossil fuels as feedstocks (ingredients such as oil for plastics) and as the source of the very high-temperature heat their production requires.²Across all industries, industrial energy consumption is categorized by the temperature requirements for both thermal and mechanical energy. High-temperature heat supports processes like smelting and chemical reactions; medium-temperature heat is used in drying and other moderate-temperature processes and often for mechanical energy demands; and low-temperature heat is applied for pre-heating or maintaining specific conditions. Mechanical energy demands, such as compression work, are typically met by steam turbines or electric motors. The “big four” account for the vast majority of very high-temperature heat used in the energy industry (Exhibit 6).

Decarbonizing a range of other industries, such as food production and paper manufacturing, is not as difficult, because 90 percent of the heat they need is low or medium temperature.

Across industries (including the big four), where low- or medium-temperatures are needed, efficient and widespread heating technologies are available. They include, for instance, electrification through high-efficiency industrial heat pumps, waste-heat recovery (for example, for parts of the ammonia production process), and use of nuclear or geothermal-generated heat. And for areas where heat energy is being used not to raise temperatures, but to produce steam that is in turn used to perform mechanical tasks, electric drives could be used instead.

To address areas that need high heat for thermal applications, some electric-based low-emissions technologies can play a role. Some progress is being made. In steelmaking, for instance, electric arc furnaces deliver very high temperatures and are a mature technology. In cement and plastics, electrification projects have started, including the use of new rotodynamic heaters to provide sufficiently high temperatures in the calcination of cement, or electro-cracking for plastics.

But deployment of these approaches remains fairly limited, and many are still nascent. Scaling them would also require massive asset reconfigurations. This is because the form of heat transfer often needs to change. Other heat sources, such as alternative fuels like biofuels, could produce high-temperature heat and often would require less retrofitting, but there may be difficulties securing reliable inputs.

While these challenges are hard to overcome, this transformation could also bring about new opportunities. Electrification can often be cost-effective. Electrifying industrial processes can also open up new forms of flexible demand, for example, when used with thermal energy storage.

The global steel industry produces about 1.8 billion tonnes of steel every year, enough to build the equivalent of 24,000 Golden Gate Bridges.¹McKinsey Basic Materials Institute; “Golden Gate Bridge fast facts,” CNN, July 19, 2023. Currently, more than 1,400 high-emitting blast furnaces are in operation around the world.²Global blast furnace tracker, Global Energy Monitor, accessed May 2024.

Steelmakers seeking to decarbonize face some difficult decisions and fine judgments about timing. Most steel today is produced using the blast furnace-basic oxygen furnace process. This process uses coking coal to split the oxygen from iron oxide in iron ore to convert it into pig iron, before converting that pig iron into steel. Between now and 2030, about 60 percent of the world’s blast furnaces are due to be relined—their interior linings replaced or refurbished (Exhibit 7). Such relining typically takes place once every ten to 20 years.³Valentin Vogl, Olle Olsson, and Björn Nykvist, “Phasing out the blast furnace to meet global climate targets,” Joule, volume 5, issue 10, 2021. This is a significantly capital-intensive process, and the choices that steelmakers make are critical for the transition. During this window, they could go ahead with relining or replace blast furnaces with low-emissions alternatives.

The challenge is that some of the key alternative approaches being explored, such as hydrogen-based direct reduction coupled with electric arc furnaces, are not commercially scaled and not cost-competitive. Moreover, implementing these alternatives requires up-front capital investments and the right capabilities to install them. Unless those alternatives become available and viable, and make sense cost-wise, there is a danger that the relining window will close and high-emissions assets are locked in for more years.

One option to reduce the emissions of high-emissions assets that still have long, useful lives is, rather than prematurely retiring them, to retrofit them to include carbon capture, utilization, and storage (CCUS) technologies.

CCUS has been around for decades but is largely used where CO₂ streams are very concentrated—and the CO₂ is therefore easier to capture—as is the case in natural gas processing or ethanol production.

However, most emissions currently come with CO₂ in relatively low concentrations, for instance in cement production and natural gas power plants. Capturing the CO₂ is more difficult and less efficient in these cases. For CCUS to play to its full potential, it would need to reach these new, harder processes. Using CCUS in such processes could be three to four times more costly than it is in current use cases (Exhibit 8). This is because more energy and equipment would be required, and new technologies would be needed to capture CO₂ effectively at low concentrations. Improving the performance and reducing the cost associated with capture technologies is critical to reaching these harder, low-concentration use cases.¹“The world needs to capture, use, and store gigatons of CO₂: Where and how?,” McKinsey, April 2023; and “Scaling the CCUS industry to achieve net-zero emissions,” McKinsey, October 2022. And, of course, once captured, the CO₂ would need to be transported and used or stored. Expanded storage capacity would be required. Improving the commercial feasibility of new use cases for captured CO₂ would also help, for instance, the production of synthetic fuels.

Like CCUS, hydrogen has many distinctive features and potentially has a large role to play across the energy system as a complement to electrification. Hydrogen’s physical properties could make it a flexible tool in decarbonization efforts. It can be used as a feedstock in many industrial processes, including the manufacture of steel and chemicals, or as a source of high-temperature heat required by some processes. Hydrogen can also be an effective energy carrier. It has high (gravimetric) energy density (per unit of weight), which could be important for long-range transportation or long-duration energy storage.

But a major difficulty would need to be addressed: Using hydrogen typically involves many conversion steps, and energy is lost at every stage of the journey that the hydrogen molecule travels. Overall, 40 to 75 percent of energy can be lost when hydrogen is used in power, industrial heat, or mobility applications.¹This is based on analysis of end-to-end efficiencies starting from the point of final energy. The efficiency of different technologies is based on Hydrogen insights, McKinsey and the Hydrogen Council, 2023; Net-zero heat: Long-duration energy storage to accelerate energy system decarbonization, McKinsey and the LDES Council, 2022; and Net-zero power: Long-duration energy storage for a renewable grid, McKinsey and the LDES Council, 2022. For instance, if hydrogen is used in a fuel-cell EV (FCEV), only about 25 to 35 percent of the energy originally available in the form of electricity to make the hydrogen molecule is converted into actually running the vehicle. By comparison, a BEV may have 80 to 90 percent energy efficiency, or up to quadruple that of a hydrogen-powered FCEV (Exhibit 9).

To make broader use of hydrogen a reality, energy losses would need to be minimized. Options would include innovation of new electrolyzer models and new configurations of production and transportation. For instance, it may be more effective to transport intermediate products made from hydrogen such as hot-briquetted iron rather than the hydrogen itself. And, more broadly, it is important to consider the strategic use of hydrogen. Hydrogen could be considered, in particular, in cases where its beneficial properties are most evident and where other low-emissions alternatives are less feasible.

Many low-emissions technologies rely on critical minerals, from lithium for batteries to rare earths for wind turbines and electric vehicles.¹The net-zero materials transition: Implications for global supply chains, McKinsey, July 2023. For the energy transition to advance at pace, and more clean technologies to be deployed, demand for and supply of critical minerals would need to increase substantially, particularly in the period to 2030 under McKinsey’s 2023 Achieved Commitments scenario. Demand for nickel could double, for dysprosium and terbium, quadruple, and for lithium, increase sevenfold (Exhibit 10).

There are sufficient reserves to meet expected demand, but additional supply often takes many years—sometimes more than a decade—to come on line.²Material and resource requirements for the energy transition, Energy Transitions Commission, July 2023. Current projections of supply based on announced projects would not be sufficient to meet the demand needs of the transition, particularly in the period to 2030. Meeting increasing demand is even more complex when the source and processing of a mineral are geographically concentrated in a limited number of economies. Many of the critical minerals required for the energy transition, including cobalt, lithium, natural graphite, nickel, and rare-earth elements, rely on the three largest supplying economies for more than 50 percent of their extraction—and over 80 percent in some cases.³Global critical minerals outlook 2024, International Energy Agency, May 2024; Global Materials Perspective 2024, McKinsey, September 2024. Refining is even more concentrated.

Some approaches could enable faster supply ramp-up. New extraction technologies, surveying approaches, and modular construction are accelerating lead times. Of course, there is another option: reducing demand for these much-sought-after minerals. There could be more recycling to extract and reuse them. And they could be avoided altogether. For instance, rare-earth-free motors could scale from less than 10 percent of total supply to making up the majority of new supply by 2030.⁴The net-zero materials transition: Implications for global supply chains, McKinsey, July 2023.