The Clean Energy Trajectory

The thesis of this report is that the clean energy transition is more than an environmental initiative: it is a foundational component of economic convergence and long-term human resilience. Its significance comes from the interaction of seven transformations: decarbonising energy production; reducing air pollution and its disease burden; expanding affordable access; diffusing technology into developing economies; electrifying transport, industry and households; reducing dependence on volatile fossil-fuel supply chains; and building more resilient, diversified energy infrastructure.

Each transformation reduces a different driver of human vulnerability. Decarbonisation sets how hot the century becomes, and so how many people live with months of hazardous heat. Cleaner combustion cuts the fine-particle pollution that kills about a million people a year by McDuffie and colleagues' conservative attribution, and several times that by others. Access gives households light, refrigeration, pumping, cooling and connectivity. Lower fuel imports cut exposure to price shocks, which in 2022 pushed import bills up sharply across developing economies.

The benefits are conditional. The IPCC notes that the costs of solar, wind and batteries fell steeply between 2010 and 2019 and that deployment grew more than tenfold for solar, but also that adoption of low-emission technology lags in most developing countries because enabling conditions are weaker. Deployment pays off for development only when it comes with grids, affordability, finance, institutional capacity and environmental safeguards. Sections 4 and 12 return to these conditions.

Ember's data show 2025 as a turning point. World generation rose by 822 TWh, clean sources added 873 TWh and fossil generation fell by 51 TWh: the first fall since the pandemic year of 2020. Solar alone added 636 TWh, and wind and solar together supplied 17.3% of the world's electricity, against 4.5% in 2015. The clean share, 42.6%, is the highest in the series. Yet coal remains the largest single source, a third of the total, and fossil generation, at about 18,200 TWh, is only just below its 2024 peak.

The geography of the transition has shifted. Upper-middle-income countries, dominated by China, now get 15.7% of their electricity from wind and solar, almost the same as high-income countries (16.1%). Lower-middle-income countries reached 9.3%, with some of the fastest recent growth in Pakistan, Kenya and Sri Lanka. Low-income countries get 64% of their power from clean sources, mostly hydropower, but generate so little electricity that the share says more about scarcity than progress: wind and solar provide only 3.4%.

In 2015, 20 countries generated at least a tenth of their electricity from wind and solar; in the latest data, 72 do, including Brazil, Pakistan, Jordan, Kenya, Chile, Viet Nam and South Africa. This diffusion matters for the thesis of the report because the technologies now spreading fastest are modular and distributed. They can be built in months, close to demand and without fuel imports, which is what makes them development infrastructure rather than only climate policy.

The price of a solar module fell from $132 a watt in 1975 to $2.51 in 2010 and $0.12 in 2025 (in 2025 dollars). The average levelised cost of electricity from new solar plants fell from $0.41 to $0.044 per kWh between 2010 and 2025, an 89% decline, and from onshore wind from $0.11 to $0.033. The IPCC's Sixth Assessment found that unit costs fell 85% for solar, 55% for wind and 85% for lithium-ion batteries between 2010 and 2019 alone.

For development, the cost curve changes what is possible. A country without coal or gas reserves, with a young and growing population and weak finances, can now build power from a resource it already has, with equipment whose price falls each year, and without committing to decades of fuel imports. The cost of a solar project is mostly up front, however, so its price depends heavily on the cost of capital, which is where developing countries are disadvantaged.

These are global averages. Costs vary with resource quality, land, grid connection and above all financing. They also exclude the costs of integrating variable supply: storage, transmission and flexible capacity. Those costs rise with the share of wind and solar, but the same learning curves are now at work in batteries. The question for 2026–2100 is no longer whether clean power is affordable, but whether it can be financed and connected fast enough where demand is growing.

Across 175 countries, life expectancy and the logarithm of electricity use per person are correlated at 0.85. The median country using under 0.25 MWh a person a year has a life expectancy of 64 years; between 1 and 3 MWh, 73; above 3 MWh, 80. Electricity does not cause longevity on its own, since income, health systems and education move together with it. But almost nothing that a modern health system, school, farm or factory does is possible without reliable power, and no country has reached high human development at very low electricity use.

Low-income countries average 0.23 MWh a person a year, lower-middle-income countries 1.1 and high-income countries 9.2. 58 countries with 1.96 billion people use less than 1 MWh, and 64% of those people live in sub-Saharan Africa. Lifting every one of them to 1 MWh by 2050, while also supplying their growing populations at today's level, would take about 2,300 TWh a year of additional supply: less than three years of the world's 2025 clean additions. A floor of 3 MWh, roughly where life expectancy flattens, takes about 13,400 TWh, about the size of today's entire clean generation.

That is the development trajectory at stake. The convergence need is large relative to poor countries' current systems but small relative to the global build rate, so the constraint is not the world's capacity to manufacture clean power. It is getting finance, grids and institutions to the places where the floor is lowest. A trajectory in which these countries climb the energy ladder on clean electricity avoids locking a new generation of coal plants into the regions most exposed to heat and air pollution.

Cumulative exposure is the right unit for a century-long question because a person who lives through decades of hazardous heat, polluted air or energy poverty bears the sum of those years. On the stalled pathway, humanity accumulates about 290 billion person-years of hazardous heat (60 or more days a year with a heat index above 35 °C) and 161 billion of dangerous heat (30 or more days above 41 °C) between 2026 and 2100. It also loses about 87 million lives to fossil-fuel air pollution, and spends about 98 billion person-years without electricity and 240 billion without clean cooking.

Each faster pathway cuts every account. The 1.5 °C transition reduces hazardous-heat exposure by 37% (to 184 billion person-years) and dangerous heat by 62% (to 61 billion). It cuts fossil air-pollution deaths by 62%, to about 34 million, and all but eliminates energy poverty. Net importers' fuel bills to 2050 fall from about $67 trillion to $49 trillion at 2023 values. The 2 °C transition delivers roughly three quarters of these gains. The current trajectory delivers about a third of the heat and air gains but half to three quarters of the access gains, because access is already improving in many countries.

The channels move on different clocks. Air pollution and fuel bills respond within years of a cut in combustion; the climate channel responds over decades, because warming follows cumulative emissions. Even so, by 2100 the stalled pathway puts 5.2 billion people (on SSP2 population) in hazardous heat, against 2.4 billion on the 1.5 °C pathway. And because energy access is the first rung of development, the access channel is where the transition changes trajectories soonest.

The heat channel uses the World Bank Climate Change Knowledge Portal's CMIP6 multi-model median climatologies on a quarter-degree grid. Each pathway is a climate scenario: SSP3-7.0 (about 3.6 °C of warming by 2081–2100), SSP2-4.5 (2.7 °C), SSP1-2.6 (1.8 °C) and SSP1-1.9 (1.4 °C), the IPCC's best estimates. Population is the same in every pathway, the SSP2 projection, so differences come from climate alone.

The paths diverge after mid-century. In the 2040s the gap between them is a few hundred million people; by the 2090s, the stalled pathway has 5.2 billion people in hazardous heat and 3.6 billion in dangerous heat, against 2.4 billion and 1.0 billion on the 1.5 °C pathway. Of the 106 billion hazardous-heat person-years that the 1.5 °C pathway avoids, sub-Saharan Africa gains 36%, East Asia and the Pacific 27% and South Asia 18%. India, Nigeria and China gain the most; for dangerous heat, India alone avoids 35 billion person-years.

The heat channel is where the transition's benefit is most delayed and most durable. Warming follows cumulative carbon dioxide, so decarbonisation in the 2020s and 2030s determines heat exposure in the 2070s and 2090s, and the IPCC finds that warming stops only once net carbon dioxide emissions reach zero. One caveat: cleaning the air also removes sulphate aerosols that currently mask part of the warming (the IPCC estimates 0.0 to 0.8 °C of cooling from aerosols and other non-greenhouse drivers), so the climate benefit of cutting coal arrives with a short-term penalty.

McDuffie and colleagues attributed the world's ambient fine-particle pollution (PM2.5) to its sources for 2017 and calculated the deaths associated with each. Summed over countries, coal combustion accounts for about 610,000 deaths a year and oil and gas for about 530,000: 1.14 million in all, of 3.82 million deaths from ambient PM2.5. Upper-middle-income countries bear 56% of the fossil deaths and lower-middle-income countries 32%; China (467,000) and India (271,000) lead, followed by Indonesia, Bangladesh, Pakistan and Egypt.

The ledger scales each country's coal and oil-and-gas deaths by its population and by the pathway's fuel use. A stalled transition costs about 87 million lives between 2026 and 2100; the current trajectory 68 million; the 2 °C transition 48 million; the 1.5 °C transition 34 million. Coal dominates the early gains: the IPCC's pathways cut coal far faster than oil and gas, so by 2050 the 1.5 °C pathway reduces annual fossil deaths from about 1.2 million to 0.3 million.

This is a conservative estimate. Attribution methods differ: Lelieveld and colleagues estimate 5.1 million deaths a year from fossil-fuel air pollution using a different exposure–response model. The ledger also holds age-specific mortality fixed, though ageing populations are more vulnerable, and it excludes household air pollution indoors, which falls with clean cooking (section 8). Cleaner air arrives with the cut in combustion, which makes the health channel the fastest-paying part of the transition.

In the latest World Bank data for the countries covered here, about 640 million people lack electricity and about 2.0 billion lack clean cooking; the 2025 Tracking SDG7 report puts the electricity gap at 666 million in 2023, 85% of them in sub-Saharan Africa. If access rates froze, population growth would push the number without electricity to about 755 million by 2030 and past a billion by the late 2040s: about 98 billion person-years without power by 2100.

Continuing each country's 2015–2023 pace of progress cuts that to 26 billion person-years, because many countries are closing the gap quickly. But the countries that are not, many of them in the Sahel and Central Africa where populations are growing fastest, leave a long tail. Universal electricity by 2030, as in the SDG target, reduces the total to about 1 billion person-years. Distributed solar, home systems and mini-grids are now usually the cheapest way to connect dispersed rural households, which makes the target technically reachable even where grids are far away.

Clean cooking is the larger gap: about 240 billion person-years without it on the stalled pathway, 119 billion on the current trend, and 4 to 15 billion if universal access arrives by 2030 or 2040. It is also a health and gender issue. Smoke from wood, charcoal and dung affects women and children most, and the ledger counts only its outdoor share. In the McDuffie data, residential biofuel burning causes about 570,000 deaths a year through ambient pollution, and the ledger scales those deaths with the share of people still cooking with solid fuels.

121 of the 176 countries covered are net fuel importers, home to 71% of humanity. Developing-country importers spent about $1.2 trillion on fuel imports in 2023; for the median such country with more than 5 million people, that was 5.8% of GDP. Lebanon, Mauritania, Liberia, DR Congo, Burkina Faso, Senegal, Thailand and Mali spent more than a tenth. Those bills are paid in foreign exchange, swing with world prices, and in 2022 rose sharply across almost every importer.

At 2023 import values, net importers would spend about $67 trillion on fuel imports between 2026 and 2050 if oil and gas use stayed at today's level. On the 1.5 °C pathway they would spend about $49 trillion: $18 trillion less, of which developing importers keep $7.9 trillion. Held constant, these numbers ignore growth in demand and prices, and the transition itself would lower world fuel prices, so the saving for importers is a lower bound on the volume effect and silent on the price effect.

An earlier H Heuristics analysis found that the 2022 jump in import bills was unrelated to how clean a country's electricity was, because most imported oil and gas is burned outside the power sector, in transport, industry and buildings. That is why electrification is one of the seven transformations: the fuel-security dividend of clean power arrives only as vehicles, cooking, heating and industrial heat move onto domestic electricity.

The simulator runs the same ledger as section 5 in your browser, country by country and year by year. The climate scenario sets the heat channel; the fuel sliders set the air and fuel-bill channels; the access path sets the electricity and clean-cooking channels. Each output is shown against the stalled pathway, so the numbers in green are what your pathway avoids.

Two experiments are instructive. First, keep the climate at SSP2-4.5 and cut coal by 95%: most of the air-pollution benefit of the 1.5 °C pathway arrives, because coal causes about half of fossil air-pollution deaths and is the cheapest fuel to displace in power. Second, keep fuels unchanged and set access to universal by 2030: energy poverty all but disappears, at a small fraction of the cost of the full transition, because it is a matter of distribution rather than scale.

The sliders combine choices that, in reality, are linked: fuel cuts of the 1.5 °C size are not consistent with SSP3-7.0 warming, and fast access is easier when clean power is cheap and well financed. The presets keep the combinations consistent; mixing them is a way of seeing which channel each lever moves, not a forecast.

Finance. Clean energy is capital-intensive and fuel-free, so the price of money sets the price of power. The IEA finds that the cost of capital for utility-scale solar in emerging and developing economies is more than twice that in advanced economies, and that narrowing the gap by one percentage point would cut their clean energy financing costs by about $150 billion a year. Guarantees, currency-risk instruments, blended and concessional finance, green bonds and deeper local capital markets lower that cost, and multilateral development banks can take the risks that private lenders price highest.

Grids and systems. Cheap generation is of little use without the wires and flexibility to deliver it. Transmission, distribution, storage and system operation need to grow with generation, and utilities need to be solvent enough to pay for power. Distributed solar and mini-grids can reach rural households ahead of the grid, while interconnectors and regional power pools let countries share resources such as hydropower, geothermal and the best solar and wind sites.

Institutions, industry and fairness. Clear planning, transparent procurement and stable regulation lower risk as much as any subsidy. Clean cooking needs its own programmes for fuels, stoves and finance. Countries with minerals or good solar and wind resources can capture more value by processing and manufacturing locally, and workers and regions that depend on coal, oil and gas need support through the change. Electrifying transport, industry and buildings is what turns cheap clean power into lower fuel imports.

Electricity: Ember yearly electricity data (generation by source, demand per person, carbon intensity, capacity), world to 2025 and countries to their latest year (2024 or 2025). Income groups from World Bank classifications. Costs: IRENA solar module prices and levelised costs via Our World in Data (constant 2025 dollars). Access, trade, GDP and life expectancy: World Bank WDI. Population: UN World Population Prospects 2024 (medium) for national channels; SSP2 gridded population for heat. Fuel bill = fuel share of merchandise imports × merchandise imports ÷ GDP, 2023; net importer if imports exceed exports.

Pathways: climate scenarios SSP3-7.0, SSP2-4.5, SSP1-2.6 and SSP1-1.9; fuel-use changes by 2050 from IPCC AR6 WGIII SPM C.3.2 medians (current = half the 2 °C medians; oil-and-gas pollution uses the mean of the oil and gas changes); fuel use linear from 2025 to 2050 and held after. Heat: CCKP CMIP6 0.25° multi-model median days with heat index above 35 °C and 41 °C, interpolated from 1995–2014, 2040–2059 and 2080–2099 and extrapolated to 2100, with SSP2 population at decadal steps. Air: McDuffie et al. 2021 country PM2.5 source shares and GBD2019 deaths (2017), scaled by population and fuel use.

Caveats: the ledger isolates channels and does not model feedbacks between them, such as cooling demand rising with heat. Linear scaling of attributable deaths ignores the non-linear exposure–response, changing baseline mortality and ageing. Access trends are extrapolations, not forecasts. Fuel bills hold 2023 prices and exclude growth in demand. Heat thresholds summarise heat stress, not mortality. National values hide sub-national variation, and data years differ by country. All cumulative figures cover 2026–2100.