The Framework

China earns 46% of Britain's income per capita and consumes 23% more energy per person.¹ Energy is the prime driver of civilizational development so it behooves us to pay more attention to the latter number.

I · The analytical void

Most contemporary macro analysis is unmoored from physical reality. It runs on financial-flow accounting and headline-driven development assessment. GDP is treated as the metric of national progress. Capital markets are treated as the mechanism of wealth creation rather than its allocator. Services-economy frames are imported wholesale into commentary on industrializing countries. Development progress is judged event-by-event, country-by-country, in the rhythm of news cycles.

This financialized lens misses the physical substrate under the flows: the energy, materials, and infrastructure that actually generate the wealth finance then distributes. It misses the aggregate trajectory of human development, which is more predictable than country-by-country news framing suggests. And it misses the time-scale on which civilizational transitions operate, measured in decades and generations rather than election cycles or quarterly reports.

Two well-known writers have done the foundational work to correct these gaps. Vaclav Smil began a fifty-year, fifty-book body of work in 1976 constructing a detailed history of energy, materials, and civilizations. Hans Rosling's work on societal development has been well-read and his TED talks have been watched tens of millions of times. But in my view their work is rarely operationalized as a unified framework for projecting macroeconomic and civilizational outcomes. People read Smil, agree that civilization is made of stuff, and then think about development in GDP anyway. People watch Rosling, agree the numbers are better than the news, and keep using headlines as a heuristic anyway. Few use either as the operating system for actual analysis.

II · The demographic momentum

Hans Rosling's central claim is that human development, in aggregate, progresses on a more reliable trajectory than developed-market commentary suggests. The headline indicators of child mortality, life expectancy, female education, fertility, electricity access, and vaccination rates improve at predictable rates as countries move up the development curve. His claim is not that every country succeeds, that progress is linear, nor that setbacks don't happen. It is that the aggregate trajectory is reliable even when individual countries disappoint. The typical Western frame, which treats global south progress as fragile, exception-laden, and perpetually reversible, is empirically wrong about the aggregate, even when it's right about specific countries in specific years.

To make the curve legible, Rosling divided humanity into four income tiers, at roughly $2, $8 and $32 a day, each describing a recognizably different lived reality.² He used dollars to group the tiers because dollars are what household surveys collect.

What differentiates Rosling's work is not its optimism but rather its discipline of looking at aggregate data over multi-decade windows instead of country-by-country news flow. Most commentary on the global south runs on news flow: this election was contested, that currency collapsed, this conflict erupted, that program failed. The lens produces a permanent perception of fragility because those are the stories with salience. The aggregate trajectory of billions of people getting grid electricity goes unreported. Operationalizing Rosling means treating the demographic tide as the signal and the political news flow as the noise, which is the reverse of what analysts do.

III · The energy constraint

Vaclav Smil's central claim is that energy throughput is the substrate of civilization. Every civilizational capacity (food production, healthcare, education, transportation, communication, manufacturing, defense) is downstream of how much energy a society can produce, deploy productively, and convert into useful work. GDP is a lagging accounting artifact of energy throughput. Financial flows allocate energy-derived wealth, they do not generate it. The wealth comes from the joules and atoms.

This sounds reductive until you trace the implications. Infrastructure stock turnover is brutally slow: a dominant fuel takes, by Smil's estimates, 50–60 years to work through the world's energy mix. After crude oil reached 5% of global supply, it needed another 40 years to reach 25%.³ Renewables will not displace fossil fuels quickly either: not for want of political will but because of the sheer mass and capital of infrastructure assets, which carry 30–50 year lifespans. You cannot software-update a gigawatt power plant.

Material throughput, like energy, is not optional. Smil identifies the four pillars of modern civilization (cement, steel, ammonia and plastics), none of which can be substituted at the volumes civilization requires.⁴

Per-capita energy is the binding constraint on development. A country at 20 gigajoules (GJ) per person per year cannot have OECD healthcare regardless of policy choices. A country at 200 GJ can afford it; whether it delivers and for how many is a question of governance.

Energy sets the ceiling. Set that beside a fifty-to-sixty-year clock on fuel transitions and the implication is bleak: the poorest countries wait two generations for the fuel mix to turn over. The two claims only compose that way if the transition in question is a fuel substitution.

The electron exception

Smil's own skepticism on the pace of energy transitions is well founded, and it is a claim about fuels. Electrification is a different process. Once a machine is electrified, it no longer cares about how the grid generates electricity. The electron is not a fuel: supply and demand transitions decouple, and can run at the same time. A factory can burn coal in its own boiler or run the same process on electricity; a household can pump gasoline or charge the car.

Whatever the transition, it runs ahead in the country that builds the equipment. Britain ran on coal while the world still burned wood, and America motorized decades ahead of everyone because it built the cars and pumped the oil. The machines of the electrostate^§ – solar panels, batteries, electric cars, heat pumps – are built in China. Britain was 2% of humanity when it led coal; postwar America was 6% when it led oil; China is 17%, so when the manufacturer converts its own market, the world total moves with it.⁶

Plain Sight Research measures the electron transition as the share of final energy arriving as electricity, source agnostic. A fossil joule of work is accompanied by two fossil joules of waste heat. Replacing an engine with a motor means replacing three joules of fuel with a single joule of electricity. Thus, joule-counting statistics make this shift look smaller than it is: a 50% electrified economy does 75% of its work through electrons.⁷

IV · The Smil-Rosling framework

Smil and Rosling never engage each other's work: Smil is an energy and material historian, Rosling was a global-health statistician. Rosling observed the trajectory of human development; Smil explains its physical cause. You cannot get the Rosling outcomes without the Smil throughput. Rising life expectancy requires hospitals; hospitals require power, water, climate control, equipment, supply chains. Rising female education requires lighting, transport, and the labor-saving machines that free women from subsistence work. Defeating child mortality requires cold-chain pharmaceuticals, sanitation, electrified clinics.

So the synthesis is a change of unit: keep Rosling's tiers, restate them in Smil's joules. Income and energy track each other closely, so most countries barely move. The exceptions use far more energy than their incomes predict: petrostates, cold countries, heavy industry, and China.

V · The Smil-Rosling tiers

Four tiers, at 30 GJ, 70 GJ and 150 GJ of primary energy per person per year. Only the floor is published. 70 GJ is set here, roughly where a country stops rationing energy and starts building with it, and 150 GJ roughly where it stops building and starts maintaining.¹⁰

Across 140 countries, measured outcomes cross their thresholds early: electricity access at 12 GJ per person per year, basic sanitation at 15 GJ, infant survival at 24 GJ, life expectancy at 30 GJ.¹¹ These are what the best-performing countries reach at each energy level, and the relationship is not deterministic: a typical country arriving there does not get the same result.

China walked the whole curve: life expectancy near 63 years in 1978, when just over half of rural households had electricity; past 70 years by 1995, at 30 GJ per person. The 95 GJ it added since then bought seven more years of life, and everything else.¹²

"Everything else" is the part with no threshold. Soap saturates; steel does not. Female education, urbanization, mechanized transport, air conditioning, and the four pillars themselves keep scaling with throughput far past the survival floor. Past survival, energy is no longer just keeping people alive; it builds where they live, determines what they drive in, and what the texture of their every day life looks like. Survival is bought crossing tier I. The climb through tier III is a claim on materials.

VI · The current state of humanity

Three and a half billion people, 43% of humanity, live in countries averaging under 30 GJ, and between them they consume less than a tenth of world energy.¹³ The richest tenth of humanity consumes 35%. The poorest half consumes just 12%.

Tier III is the one that has grown. Two and a half billion people, 30% of humanity, now live in countries between 70 GJ and 150 GJ, and more than half of them are Chinese.¹⁴ On income, China is a middle-income country in Rosling's third tier. On energy, it sits with Germany and Japan.

National averages are already the softer of the two available readings. On national averages, only 8% of people consume less than 10 GJ a year, but measured on individuals, 38% of people consume less than 10 GJ, meaning the national averages understate energy poverty five-fold.¹⁵ Nigeria has people in every tier. So does India. Countries do not climb the curve; people do, at different times inside the same borders. I mostly use national averages because that is what is published for every country on a consistent annual basis, so every count and every cost is already optimistic.

VII · The math of the next 25 years

How much incremental energy will the world need every year if we lift every country up to each of the tiers? This exercise gives us an idea of the magnitude, and is not a forecast.¹⁶

Lifting every country over the 30 GJ floor costs 46 exajoules (EJ) a year, 7.5% on top of the 620 EJ the measured world already uses (1 EJ = 1 billion GJ). Lifting everyone to 70 GJ a year – out of rationing, into building – costs 211 EJ. Lifting everyone to 150 GJ a year, a rich-world standard, costs 666 EJ, more than doubling world total energy, and is not going to happen this quarter-century. Population growth stacks on top: the UN expects 1.4 billion more people by 2050, which is another 42 EJ at the floor and 98 EJ at 70 GJ per person per year.¹⁷

7.5% incremental world energy, in the right places, ends energy poverty as the survival threshold defines it. That has been affordable for decades, and it has not happened for reasons that are political and financial, not thermodynamic. The climb is what is expensive: roughly 34% incremental world energy for the existing population, roughly 50% once population growth is counted. The survival tier is cheap and the rest of the climb is not.

Counting an electric buildout in fossil units

The deficit is computed in primary energy, which counts one joule of solar electricity as 2.44 joules: the fuel a thermal plant would have burned to deliver it.¹⁸ So the engine-to-motor swap, three joules of fuel down to one of electricity, books as three down to 2.44, and a 67% saving in delivered energy shows up as 19%. The unit absorbs about seven-tenths of the electrification dividend.

The 211 EJ therefore overstates the climb in proportion to how electric it turns out to be. Fully electric it books as 171 EJ, and the 40 EJ between them is most of what ending energy poverty costs.

VIII · Who supplies the substrate

Smil-Rosling supplies the framework and the energy accounting, but who actually builds the substrate? Concrete is poured by entities with cement plants. Solar arrays are installed by crews with panels somebody manufactured. Transmission lines are laid by firms with crane fleets and engineering staff. Whoever does this work, at the scale and pace required, becomes the dominant economic and political actor of the era they supply.

In the decade and a half following WWII, the United States was the undisputed supplier of physical reality. It held the manufacturing depth, the engineering capacity, and the patient capital to supply the rebuilding of Europe and the industrialization of Japan. It emerged from the war already the wealthiest nation on earth; supplying the postwar buildout is what converted that lead into three decades of compounding dominance rather than a peak. To analyze the US in 1955 through a purely financial lens was to entirely miss where the next 50 years of global wealth were being formed.

Today that stack of capabilities sits in one place. Whether producing a thing is the same as supplying the world with it varies by category.¹⁹

A Nigerian building a house buys Nigerian cement, and Vietnam and Indonesia are net cement exporters themselves. What China supplies is not the substrate entire but the part of it that moves: the machines that make electricity, the machines that store it, and increasingly the machines that use it. That stack, assembled, is the subject of Building the Electrostate.

Electricity is on the order of a fifth of world final energy today,²⁰ so a buildout that reproduces the current mix is roughly a fifth Chinese-supplied. A buildout that runs on the cheapest new generation, solar with storage, is mostly Chinese-supplied. The share and the size move on the same lever: every point of electrification makes the buildout smaller in primary energy and larger in Chinese content.

From 2000 to 2024, US electricity generation grew 13%. China's grew roughly 650%. China today generates more than twice the electricity of the United States, and more than the United States, the European Union and India combined.²¹

This is not a political argument. A planetary buildout cannot be supplied through declarations of intent or through reshoring plans operating on 20-year timelines. It is supplied by entities that already possess the factories, the engineering crews, the balance sheets, and the institutional patience. Capability is upstream of preference. Taking China's industrial capacity seriously is not an aesthetic choice but an arithmetic one.

IX · The incumbent vs. the manufacturer

The postwar analogy holds with one asymmetry: the United States today is not structurally the United States of 1955. Its energy consumption has not moved this century (95.6 EJ in 2000, 95.5 EJ in 2024) while its population grew 23%, so throughput per head fell 19% and sits below its 1979 level.²² It maintains manufacturing depth in isolated, high-margin categories – aerospace, advanced semiconductors, software – but has hollowed out the base layers: cement, steel, transformers, basic chemicals, heavy machinery.

It has also lost the institutional memory of how to build at scale. Where like-for-like data exists, US urban rail costs run multiples above their Chinese equivalents.²³ And its financial architecture has been optimized for short-horizon, extractive returns and asset-light software – the exact opposite of the 30- to 50-year capital horizons that civilizational infrastructure requires.

China, conversely, is the rising manufacturer: current-generation engineering capacity, throughput momentum, and patient financing. This asymmetry between a financialized incumbent and an industrialized manufacturer is why the next 25 years will not look like a continuation of the last 25. It is the analytical core of this publication, and it is why we treat China's Five-Year Plans as deployment orders with proven execution capacity and track record rather than as merely political propaganda. The last 25 years validate the reading. The next 25 will be determined by if they can stay the course. So far, their energy and materials build out says they will.

X · Operating parameters

Physical capacity is the upstream variable of human history. Aggregate demographic trajectories are more reliable than headline narratives. Stated long-term plans are deployment orders worth taking at face value when made by states with execution capacity. The next 25 years of global wealth creation will follow the buildout, not the financial commentary about the buildout.

Watch the physical stack and the supply chains. Weight aggregate data over event-by-event commentary. Be willing to land on conclusions that sit outside developed-market consensus when the framework points there.

Three and a half billion people sit below a floor that costs 46 EJ a year to clear. The climb above it costs four and a half times that, and the machines that make the climb electric are built in one country. China's integrated energy infrastructure stack is becoming the operating system of Global South development. Building it is one question and converting to it another; selling it is a third.

Sources for every load-bearing figure. Where a number is this publication's own construction it says so, and the derivations below give the working, including the full country list, so that the counts can be checked and not merely the totals.

1. Income: World Bank via Our World in Data, GDP per capita, PPP (constant international $), 2025: China 25,066.52, United Kingdom 53,993.12.

   Energy: Our World in Data Energy use per person, 2024: China 34,514.457 kWh, United Kingdom 28,015.771 kWh, which at 0.0036 GJ per kWh is 124.3 and 100.9 GJ.

   Ratios: 46.4% and 123.2%. Note the two-year offset between the vintages; it moves neither ratio materially.

   Basis: territorial, deliberately; see derivation B2, which also states what happens to this comparison on a consumption basis, because it reverses.

   ↩

2. Hans Rosling, Factfulness (2018), four levels at roughly $2, $8 and $32 a day in 2011 purchasing power, and populations of about 1 / 3 / 2 / 1 billion. Those are the book's figures.

   Gapminder has since moved the underlying data to 2017 and then 2021 PPP while holding the boundaries, so the current tier populations differ from the book's.

   ↩

3. Smil gives 50 to 60 years for a dominant-fuel transition and 50 to 75 for "two to three generations".

   He's considering is a new source's climb through shares of world primary energy (after crude oil reached 5% of global supply it took another 40 years to reach 25%) rather than a completed swap.

   He has not conceded the framing for solar or for China; his 2024 work still leads on the fossil share falling only from 86% to 82%.

   ↩
4. The four-pillars framing is Vaclav Smil's, How the World Really Works (2022). The tonnages here are Plain Sight Research independent verifications.

   Smil gives about 4.5 Gt of cement, 1.8 Gt of steel, 400 Mt of plastics and 180 Mt of ammonia. Current figures: USGS puts world cement at 3.80 Gt in 2025, worldsteel puts crude steel near 1.85 Gt, ammonia is near 190 Mt.

   Table comparison: China poured 6.73 Gt of cement in 2011–2013, against roughly 4.5 Gt in the United States across the entire 20th century.

   ↩

5. Smil's claim, consistently across How the World Really Works ("impossible to feed at least 40% and up to 50% of today's nearly 8 billion people"), Enriching the Earth ("about two-fifths of the world's population would not be around") and his 1997 Scientific American piece.

   Erisman et al. (2008), Nature Geoscience, reach 48% independently.

   A related claim that half the nitrogen in a person's body came from a Haber-Bosch plant circulates widely; it originates in popularization, carries a "living in a developed country" qualifier that is usually dropped, and is diet-dependent. It is not used here.

   ↩

6. China is 17.0% to 17.6% of world population and falling, from a peak near 22.8% in 1974.

   Britain's roughly 2% at the height of coal and America's roughly 6% at the height of oil are both for the home populations; the British imperial system of 1900 was 412 million people, about 23.5% of humanity, which is a different claim and not the one made here.

   ↩

7. The wedge follows from the thermal efficiency of combustion conversion: roughly a third of fuel energy reaching useful work in an internal-combustion engine or a condensing power plant, against electric drive at high single-digit losses. It is a claim about final energy.

   Derivation B7 shows what happens to it inside the primary-energy accounting this essay's own numbers use, which is most of the reason section VIII carries the caveat it does.

   ↩

8. Author construction across 30 large economies, from the two series in note 1.

   Gigajoules per $1,000 of PPP GDP: Ethiopia 0.80, Nigeria 0.98, Bangladesh 1.26, United Kingdom 1.87, Italy 1.92, Germany 2.17, Pakistan 2.27, India 2.80, Japan 2.89, Vietnam 3.40, United States 3.59, Korea 4.52, China 4.96, Russia 5.31, Saudi Arabia 5.53, Canada 6.07, Iran 8.38. Full table in derivation B5.

   ↩

9. Energy per person, GJ, 2024: China 124.3, Spain 122.0, Italy 102.9, United Kingdom 100.9, Portugal 93.3, against PPP incomes of $25,067, $49,318, $53,606, $53,993 and $42,577.

   Pakistan 12.7 against Nigeria 8.0 is 159% on the grapher series, but Nigeria's figure is EIA gap-fill and Pakistan's is Energy Institute, a basis mix inside one comparison, which is why the body says "roughly 60% more" rather than a decimal. On a consistent single-source 2023 basis the ratio is nearer 156%.

   Bangladesh's income is 6.5% above Nigeria's, which the body calls "about what Nigeria earns".

   ↩
10. The 70 GJ and 150 GJ boundaries are this publication's construction.

    Gapminder, Smil, the saturation literature and the decent-living-energy literature map no development tier to gigajoules per capita, and the saturation literature could not supply these: its highest threshold of any kind is 125 GJ, and its finding is that development outcomes saturate early.

    Derivation B3 gives the construction and its sensitivity.

11. R. B. Jackson et al. (2022), Ecosphere 13(4): e3978. 140 countries, nine metrics.

    The thresholds are defined in the paper as "the energy consumption at which the 99th percentile curve reaches 95% of the maximum observed metric score", fitted by nonlinear quantile regression to the best performers at each energy level, not to the average.

    Electricity access 12 GJ, basic sanitation 15 GJ, infant mortality (under one year, not child mortality under five) 24 GJ, life expectancy 30 GJ; the highest threshold for any of the nine metrics is 125 GJ for air quality, and eight of nine plateau between 10 GJ and 75 GJ.

    The 1971–2018 drift from about 22 GJ to about 30 GJ was run for life expectancy only and is not a property of the other metrics.

    The authors state that "the fitted curves do not imply direct causality with energy use as the primary driver of these metrics".

    ↩
12. Life expectancy from the UN World Population Prospects via Our World in Data: 63.1 years in 1978, 70.4 years in 1995, 78.0 years in 2023.

    Energy: 29.4 GJ per person in 1995 and 124.3 GJ in 2024 on the substitution series, an addition of 95 GJ.

    Basis: this essay runs on substitution throughout. The physical-content basis gives 80 GJ across the same span; the two diverge for China as its non-fossil share grows.

    Rural household electrification in 1978 was 53.3%, against 61.05% of villages and 86.86% of townships: China Yearbook of Electricity Industry (1980), reported in Pan Jiahua et al., Rural Electrification in China 1950–2004, Stanford Program on Energy and Sustainable Development, Working Paper 60, December 2006.

    ↩

13. Author construction; derivations B1 and B4, which include the full country list. 120 countries, 8.0 billion people, 620 EJ.

    Under 30 GJ: 48 countries, 3.5 billion people, consuming 59 EJ or 9.6% of covered energy. Against a world population near 8.2 billion that is 43.1% of humanity; the body says 43%, not the 44.2% that the covered set alone would give.

    Richest tenth 35% of energy, poorest half 12%.

    ↩
14. Same construction. The 70–150 GJ tier holds 2.48 billion people across 32 countries; China's 1.4 billion is 57.3% of it. Against a world population near 8.16 billion the tier is 30.4% of humanity; the body says 30%. The exhibit's 31% is the tier's share of the covered set. Next largest are Japan at 5.0%, Iran 3.7%, Turkey 3.5% and Germany 3.4%. ↩

15. Oswald, Owen & Steinberger (2020), Nature Energy 5, 231–239, across 86 countries and 374 country-by-income-group segments: "38% consume less than 10GJ yr⁻¹ capita⁻¹… Based on national averages we would measure that only 8% of the population consume less than 10GJ."

    Basis warning: these are consumption-based household final-energy footprints, roughly half the size of the territorial primary energy used elsewhere here, and they exclude the 30% of footprints that are government and capital formation, which is the construction this essay is about.

    A 10 GJ household footprint is on the order of 20 GJ of primary energy, which lands above Jackson's electricity and sanitation thresholds rather than below them, so the two literatures must not be placed side by side without converting.

    The finding is used here only for the direction and order of magnitude of the country-average bias.

    ↩
16. Author construction; derivation B6. Deficit to 30 GJ 46.4 EJ (7.5% of 619.9), to 70 GJ 210.7 EJ (34.0%), to 150 GJ 666.4 EJ (107.5%), each holding every country already above the threshold constant. Computed on national averages, so it understates what reaching every person would cost. ↩

17. United Nations, World Population Prospects 2024 revision, medium variant: the increment from 2025 to 2050 is 1.4328 billion, of which Africa, Asia and Oceania together are 1.3729 billion, or 95.8%. Africa alone is 64.0% and India 15.1%.

    That is a net figure concealing a large East Asian contraction: China falls by roughly 156 million over the same window.

    A commonly quoted 94% is a different calculation on a gross-growth denominator and is not the same number.

    ↩
18. The substitution method converts non-fossil electricity to a primary-energy equivalent by dividing by an assumed thermal generation efficiency of about 0.41, so one joule of non-fossil electricity is counted as about 2.44 joules. Fossil fuels are counted at their physical energy content. Derivation B7. ↩
19. Category selection: the tradable machine categories of new generation (solar, wind, storage, nuclear; hydro is site-built, coal and gas are the incumbent stack), and the four material pillars (cement, steel, ammonia, plastics).

    Production shares. Cement 45% (1.70 Bt of a USGS world total of 3.80 Bt, 2025). Crude steel 52% (961 Mt, 2025). Solar PV above 80% at every stage (conservative, since wafers are nearer 97%). Lithium-ion batteries above 80% of world manufacturing capacity, with grid storage running almost exclusively on Chinese LFP cells. Wind: the top four turbine suppliers of 2024 are Chinese, led by Goldwind above 20 GW (GWEC supply-side data, 2024); the two-thirds figure is derived: roughly 80 GW of Chinese domestic installations plus 5.5 GW abroad, on 127 GW installed worldwide. Nuclear: 29 reactors under construction in China of roughly 60 worldwide (IAEA PRIS); the IEA counts 78 GW under construction across 15 countries, half of it Chinese. Ammonia: 31% on contained nitrogen (USGS, 2024: China 47 Mt N of a world 150 Mt N, about 190 Mt as ammonia). Plastics: 33.3% of 413.8 Mt (Plastics Europe, 2023).

    Tradability: world cement and clinker trade is between 3% and 6% of output, depending on the year and whether clinker is counted. That denominator includes China, which produces 45% of world cement and exports almost none; on non-Chinese production the traded share is nearer 11%.

    Vietnam exports about 30 Mt, Indonesia about 12 Mt, Nigeria about 1 Mt. India's net position is under 1% of its own output and it does not appear in world top-exporter rankings, so it is not listed.

    Ammonia trade is 18 to 20 Mt a year, under a tenth of output. Wind: Goldwind and Envision were 98% of the 5.5 GW installed outside China in 2024. Nuclear: Russia is building abroad in Egypt, Bangladesh, Turkey, Hungary, India and China; Korea's four UAE units are in commercial operation; China's exported reactors, from the early CNP-300s to the Hualong One, are all in Pakistan.

    The cement trade figures carry no primary source here.

    ↩
20. Electricity is on the order of a fifth of world total final energy consumption. No primary source is given here; the figure is approximate, and load-bearing for the size of the supplier claim.

    What it multiplies against, China's share of solar and battery manufacturing, is in note 19.

    ↩

21. United States: EIA net generation, all sectors: 3,802.1 TWh in 2000 against 4,308.6 in 2024, or +13.3%.

    On the Energy Institute's gross basis the 2024 figure is 4,634.8 TWh and growth is nearer 15%; the series is named because the two differ.

    China grew roughly 650% over the same window to 10,087 TWh in 2024, against India's 2,030. The comparison holds on every published basis.

    On the same EIA series 2025 was a record year, +2.8%, taking 2000–2025 growth to +16.5%.

    ↩
22. United States primary energy, same OWID series: 95.57 EJ in 2000 against 95.50 EJ in 2024, a fall of 0.1%, while population grew from 281.5 million to 345.4 million, or 22.7%.

    Per person that is 339.51 GJ falling to 276.48, a fall of 18.6%, and 19.0% below the 1979 peak of 341.4 GJ.

    Sourcing: the 2024 endpoint is taken directly from the grapher; the historical points come from a second-hand rebuild of the same series. Per capita times population reproduces the aggregate to within 0.1% at both ends.

    The stagnation claim attaches to the aggregate, not to the per-capita figure.

    ↩
23. Transit Costs Project: United States urban rail at about $1,601m per mile against a non-US global average of $478m, roughly 3.3 times. Chinese subway construction is put at around $250m per kilometre in purchasing-power terms, making the US–China multiple nearer four.

    Two caveats: the project notes that the five countries more expensive than the United States are building far more tunnel, and a competing study finds a premium nearer 50%.

    The project publishes no timeline comparison, only the qualitative finding that China and India build cheaply and fast.

    ↩