Accelerating the Global Clean Energy Transition

Ember's data show the clean share of world electricity at 33.5% in 2015, 36.3% in 2019, 40.9% in 2024 and 42.6% in 2025. The gain since 2019 is 1.05 percentage points a year (1.57 for wind and solar alone), and 0.91 a year over the decade from 2015. A straight line from 42.6% in 2025 to 100% reaches the target in 2050 at 2.29 points a year, in 2045 at 2.87, in 2040 at 3.82 and in 2035 at 5.74. These straight lines are a convenient yardstick, not a forecast: transitions follow S-curves, which accelerate in the middle, and the right comparison is therefore between the current pace and the pace needed on average over the period.

The yardstick is demanding because demand keeps growing. World generation grew 2.8% a year in 2019–25, and cheap clean power plus electrification of transport, heat and industry will push it higher. Meeting that growth and also displacing fossil generation needs, on average, about 1,540 TWh a year of new clean generation to the end of 2050 if demand grows 2% a year, 2,120 if 3% and 2,840 if 4%. The clean additions of 2025 were 873 TWh and the record in 2024 was 963 TWh. A fully clean grid by 2040 needs 1,950, 2,390 or 2,910 TWh a year on the same assumptions.

These numbers show why the headline pace understates how close the world could be. Annual additions are not constant: wind and solar additions have been rising 19% a year. A system whose additions double every four years needs only a few doublings to reach the required level. The question is not whether the world can build that much in a year; it is whether the growth in additions can be sustained for fifteen years, against grid, financing and supply-chain limits. Section 7 lets the reader test that.

Wind and solar added 839 TWh of generation in 2025, up from 287 TWh a year on average in 2018–20 and 684 TWh in 2023–25: growth of 19% a year over five years. Clean sources added 963 TWh in 2024 and 873 TWh in 2025 against demand growth of 1,271 and 822 TWh, so in 2025 clean power covered all of the growth in demand and fossil generation fell by 51 TWh, only the third fall in the decade since 2015 (the others were 2019 and 2020). In 2019–24 taken together, across 153 countries, clean generation rose 2,926 TWh against demand growth of 4,115 TWh, covering 71% of it; fossil generation rose by 1,189 TWh.

IRENA counts 5,149 GW of renewable capacity at the end of 2025, after a record 692 GW of additions (15.5% growth; solar 511 GW, wind 159 GW). The COP28 pledge to triple capacity from about 3,400 GW implies roughly 11,000 GW by 2030; IRENA's target figure is 11,174 GW. From the 2025 level, reaching it needs about 1,205 GW of additions a year on average, 1.7 times the 2025 record, and growth of about 16.8% a year compound, against 15.5% achieved. IRENA's tracking in October 2025 judged the pledge reachable but off track; a summary of that report gives about 1,122 GW a year from the end-2024 base.

Two features of this record deserve emphasis. First, the achievement is concentrated: Asia accounted for 74% of the world's new renewable capacity in 2025. Second, it is cost-driven: solar modules fell 95% in price since 2010, so each unit of capital buys more. The chart of the tripling pledge shows that the gap to the target is a gap of about 1.7× in annual additions, not 5×, which is why the debate is increasingly about grids and integration rather than module supply.

The power sector is where the transition is furthest along, but electricity is only about a fifth of the world's primary energy. The IPCC's Sixth Assessment finds that in pathways limiting warming to 1.5 °C with no or limited overshoot, the global use of coal, oil and gas in 2050 falls by medians of about 95%, 60% and 45% relative to 2019; for 2 °C the medians are 85%, 30% and 15%. A straight line from 2019 to those 2050 values implies falls by 2024 of about 15% for coal, 10% for oil and 7% for gas on the 1.5 °C path, and 14%, 5% and 2% on the 2 °C path.

What happened was the reverse. Between 2019 and 2024, world coal use rose 5.1% (43,628 to 45,851 TWh), oil 2.4% and gas 5.7%. Fossil fuels' share of primary energy fell from 84.1% to 81.3%, because total energy use grew and low-carbon sources grew faster. At the 2015–24 trend of −0.49 points a year, that share would take 63 years to reach 50%. The power sector's clean share is gaining about twice as fast, which says that most of the transition's remaining work lies in the fossil uses outside the grid.

That work is, in large part, electrification: moving transport, building heat and industrial processes onto clean electricity, and where that is hard, onto other low-carbon carriers. Electricity's share of primary energy rose from 17.3% to 19.8% over 2015–24, slowly. The IPCC notes that net-zero energy systems entail widespread electrification of end uses along with a substantial reduction in overall fossil fuel use. The pace of the power transition therefore matters twice: it cleans the existing grid, and it determines how clean the electricity is that other sectors switch to.

The global gap is very concentrated. Of the world's fossil electricity, China generates 35% (6,237 TWh in 2024), the United States 14%, India 9%, Russia 4% and Japan 4%: five countries, 66%. The top ten generate 76% and the top twenty 86%. Of the 1,968 TWh by which fossil generation rose in countries where it rose between 2019 and 2024, China accounts for 1,097 TWh, India 305 and Iran 90; against this, fossil generation fell by 779 TWh in countries where it fell, led by Japan (107), Germany (79), the United Kingdom (52), Spain (46) and Italy (41).

Classified by trajectory, 74 countries (40% of world generation, 32% of fossil generation) saw fossil generation fall between 2019 and 2024, among them the United States, Japan, Korea, Mexico, South Africa and Germany. 27 countries (38% of generation, 40% of fossil generation) are "catching up", with fossil generation still rising but clean additions covering at least half of demand growth; they include China, Türkiye, Vietnam and Malaysia. 52 countries (22% of generation, 28% of fossil generation) are lagging, with clean additions covering less than half of demand growth: India, Russia, Saudi Arabia, Iran, Indonesia, Egypt, Thailand and Iraq lead the list.

The pace needed differs sharply from the pace achieved. China gained 1.33 points a year against 2.38 needed for a clean grid by 2050 (1.8 times faster needed); the United States 0.82 against 2.24 (2.7 times); India 0.41 against 2.98 (7 times); Indonesia 0.37 against 3.15; Saudi Arabia 0.42 against 3.76; Egypt 0.29 against 3.42. South Korea (1.88 against 2.31) and Japan (1.64 against 2.64) are closest. Russia's and Iran's clean shares fell. The largest fossil systems are therefore not slow because they are poor; China's pace is above the world's. They are slow relative to what a fully clean grid needs.

Across 111 countries with at least 5 TWh of generation and a clean share under 95%, a regression of the 2019–24 pace on five variables explains 24% of the variation, and two things stand out. Demand growth has a large negative coefficient (−0.73 points a year per standard deviation, standard error 0.16): the faster demand grows, the slower the clean share rises, because new clean generation is absorbed by new demand. Starting clean share and coal share have coefficients of about zero. Institutions are positive (0.41, standard error 0.29) and income small and not significant. The pattern also shows in the raw data: countries whose demand grew by less than 3% a year had a median pace of 0.9 points a year; at 3–5% it was 0.4; above 5% it was 0.01.

By income group, the median high-income country gained 1.52 points a year, upper-middle-income 0.71, lower-middle-income 0.33 and low-income 0.02. But the spread within each group is large: the 90th-percentile pace is 3.7 in high-income, 2.3 in upper-middle-, 2.5 in lower-middle- and 2.7 in low-income countries. In every group the fastest tenth of countries are already moving at or near the 2.3 points a year the world needs for 2050. That is what makes peer pace a useful yardstick: it is a rate that countries with similar incomes have already achieved.

Among large systems, the fastest gains in wind and solar share came from the Netherlands (6.2 points a year, to 45%), Chile (4.2, to 35%), Greece, Finland, Spain, Poland (3.1, to 25%), Australia (3.05, to 29%), Germany, Belgium and Brazil (2.8, to 24%). These are countries with different climates and grids, mostly high-income but including upper-middle-income Brazil. This says that high pace is possible at scale under a variety of policy and market designs.

The scenarios take each of the 153 countries' pace from 2019 to 2024 and ask what happens if every country at least matches a benchmark taken from the countries in its own income group. In "own pace", each country continues as it has: the world's clean electricity reaches 65% by 2050 and 80% only in the 2070s. If every country at least matches its income group's median pace, 75 countries are lifted and the world reaches 73% by 2050. At the 75th percentile, 113 countries are lifted and the world reaches 81% by 2050, passing 80% in 2050 itself; at the 90th, 136 are lifted and the world reaches 95% by 2050, passing 80% around 2040.

The difference is large in carbon terms. Assuming demand grows 2.8% a year and fossil generation keeps its 2024 intensity, matching the peer median avoids about 40 billion tonnes of CO₂ from the power sector between 2025 and 2050 against own pace; matching the 75th percentile avoids 93 billion; the 90th percentile 191 billion. For comparison, the whole power sector emitted about 14 billion tonnes in 2024 at current intensity. The scenarios are not predictions, but they show that the transition's future depends less on any one country's breakthrough than on whether the laggards close the distance to what their peers already do.

There are three reasons for caution. First, percentile benchmarks are computed over countries of every size, and small systems can move quickly; it is harder to match when a system is large. Second, pace in a five-year window is noisy, especially for hydro-dependent countries whose share swings with rainfall, and 2019–24 includes the pandemic and the energy-price shock of 2022. Third, the scenarios assume linear gains to 100%; the last third of a clean grid is more expensive to integrate than the first.

The simulator is deliberately simple. It starts from the 2025 world: 31,734 TWh of demand, of which 13,530 TWh clean, with wind and solar additions of 839 TWh. Each year demand grows at the chosen rate, wind and solar additions grow at the chosen rate up to the ceiling, other clean sources add a fixed amount, and fossil generation is whatever clean power does not cover. It models generation, not capacity, and it does not model storage, transmission, curtailment or the cost of integrating variable power at high shares; the ceiling stands in for them.

With the defaults (additions growing 12% a year to a ceiling of 2,500 TWh, demand 2.8%), the world's clean share is 54% in 2030, 73% in 2035, 90% in 2040 and 100% by 2044, and power-sector CO₂ from 2026 to 2050 totals about 157 billion tonnes. With flat additions, it is 56% in 2050 and 412 billion tonnes. Growth of 6% gives 94% in 2050 and 251 billion tonnes; 19% growth gives 81% in 2035 and 97% in 2040. The results are most sensitive to the ceiling and to demand: with a ceiling of 1,500 TWh the world reaches only 80% in 2050, and with 19% growth and demand growing 4% a year (fast electrification) the share is 85% in 2050 and clean power is not complete in the horizon.

The ceiling therefore captures the central question. Power demand will grow faster with electrification, and each year's required additions grow with it. Whether the world can sustain additions of 2,500 TWh a year, three times today's, depends on factories, finance, grids, permitting and skilled labour, which the simulator does not describe but which the next section examines. The point of the model is arithmetic, not forecasting: it shows how much of the transition's timing hangs on whether the growth in additions continues.

Wires. Generation is rarely the binding constraint. The IEA has reported that about 1,650 to 1,700 GW of renewable capacity in advanced development is waiting for a grid connection, with hundreds of gigawatts of batteries also in queues, and has argued that grid investment must roughly double by 2030 and that the queues themselves need reform. Faster permitting for transmission, connection reform that prioritises ready projects, and more flexibility (storage, demand response, interconnection) are the levers.

Capital. Wind and solar are mostly capital cost, so the price of money sets the price of power. The IEA reports that the cost of capital for utility-scale solar in emerging and developing economies is more than twice as high as in advanced ones, which is one reason the pace in low- and lower-middle-income countries is so much slower than in rich ones: median 0.02 and 0.33 points a year against 1.52. Guarantees, local-currency finance, concessional capital and clear offtake contracts lower the price of the first project and the thousandth.

Integration and displacement. The last third of a clean grid is harder than the first, and the fossil uses outside the grid (vehicles, heating, industry) will not switch themselves. Retiring coal where it is still being built, pricing carbon and methane, and electrifying end uses are what turn clean additions into fossil displacement. The data show it plainly: in 2019–24, clean power covered 71% of demand growth, which is why fossil generation still rose by 1,189 TWh even as clean generation grew by 2,926 TWh.

Move the biggest systems first. Five countries generate two thirds of fossil power and China alone 35%. The most valuable acceleration is in China's and India's coal fleets, the United States' gas and coal fleet and Russia's and Japan's, through faster additions, retirement schedules, capacity and carbon pricing, and market designs that reward flexibility. Where large systems have accelerated (Korea, Japan, Brazil, Australia, Poland), the gains have come from sustained policy as much as technology.

Lift the laggards to peer pace. The scenarios show that matching the pace already achieved by income peers moves the world from 65% to 81% (75th percentile) or 95% (90th percentile) clean power by 2050. That is a development agenda: lower the cost of capital, extend transmission, reform utilities, support fast-growing demand centres in India, Indonesia, Egypt and Vietnam, and finance the first projects in lower-income countries, where the median pace is 0.02 to 0.33 points a year.

Unclog the grid and the fossil uses. Connection reform, grid-enhancing technologies, storage and interconnection release projects already built or ready; electrification of transport, heat and industry and the retirement of fossil capacity turn clean additions into displacement. The data show a world that now adds clean power faster than demand grows and still sees its fossil fuel use rise; the shift from "adding" to "replacing" is the heart of acceleration.

Electricity: Ember yearly electricity data (generation by source, TWh, capacity, power-sector emissions), world to 2025 and 153 countries with at least 1 TWh in 2024 and data for 2015, 2019 and 2024. Clean = wind, solar, hydro, nuclear, bioenergy and other renewables. Pace = (clean share 2024 − clean share 2019) ÷ 5. Required pace to reach 100% by year T = (100 − share) ÷ (T − base year). Required average additions = (demand in T − clean generation in 2025) ÷ (T − 2025) at the stated demand growth. Fuel use beyond power: Our World in Data (Energy Institute Statistical Review) primary energy in TWh. IPCC paths: linear from 2019 to the SPM C.3.2 median 2050 declines (an assumption: the IPCC gives only the 2050 values). Capacity: IRENA Renewable Capacity Statistics 2026 and the IRENA tracking of the COP28 pledge; the 1,205 GW and 16.8% figures are H Heuristics arithmetic on 5,149 GW and 11,174 GW.

Trajectory classes: falling = fossil generation in 2024 below 2019; catching up = fossil higher but clean additions at least half of demand growth; lagging = otherwise. Regression: OLS of pace on starting clean share, coal share, ln GDP per person (PPP), WGI government effectiveness and annual demand growth, all standardised, for 111 countries with at least 5 TWh and clean share under 95%. Peer scenarios: country share = clamp(share₂₀₂₄ + pace × (year − 2024), 0, 100); pace floors at the median, 75th or 90th percentile of countries in the same World Bank income group; world share weights by 2024 generation; fossil generation grows with the stated demand growth; CO₂ uses the 2024 world fossil-generation intensity (0.776 t/MWh, Ember). Simulator: demand D(t) = D₂₀₂₅ (1 + g)^(t−2025); wind and solar additions min(cap, 839 (1 + r)^(t−2025)) plus other clean additions; clean generation capped at demand.

Caveats: scenarios are arithmetic, not forecasts; the world simulator excludes storage, transmission, curtailment and cost; five-year paces are noisy and include the pandemic and the 2022 energy crisis; clean shares of hydro-heavy countries move with rainfall; country weights are held fixed in the peer scenarios; the IPCC linear paths and the percentile benchmarks are choices; and grid-queue and investment figures come from IEA analyses reported by secondary sources and should be checked against the reports.