Is the climate stabilizing or breaking?

So who is putting the carbon up there — and is the flow still rising?

Annual emissions hit a record 38.6 billion tonnes in 2024, and the map behind that total has been redrawn: Asia alone now emits more than the whole rest of the world combined.

Begin with the flow: the carbon we pour into the air each year, the engine behind every other chart on this page. It has never run faster. In 2024 we put about 38.6 billion tonnes of CO₂ into the air from fossil fuels and industry, more than in any year on record and roughly six and a half times the 5.9 billion of 1950. The line has pointed up for two centuries and has not yet bent over. Until this number falls, the concentration keeps climbing. It is the honest floor of the whole story.

But a single world total launders a map that has flipped. In 2000, North America and Europe each out-emitted the rest of their continents; by 2024 Asia alone released about 20 billion tonnes, over half the world’s output, while North America’s fell from 7.0 to 5.9 billion and Europe’s from 6.2 to 4.9. The rich world’s emissions are genuinely declining. Almost all the growth of the last quarter-century is Asia industrializing, with Africa still a rounding error at 1.5 billion tonnes for a fifth of humanity.

Then zoom to one person and the blame flips again. China emits more than any country in total, but about 8.7 tonnes per person, below the United States at 14.2, a fifth of Qatar’s 41.3. An Indian emits 2.2 tonnes a year, a sixth of an American; a Nigerian, 0.6; someone in DR Congo, a tenth of a tonne. Who emits the most and who emits the most per head are different questions with different villains. Almost every argument about fairness in climate is really a fight over which of those two charts to put on the table.

Whose carbon is actually sitting in the air — and who is paying for it?

The world has put about 1.85 trillion tonnes of CO₂ into the air since 1750, and a quarter of it came from one country: the United States.

The atmosphere does not keep a yearly ledger; it keeps a running total. CO₂ lingers for centuries, so what warms the planet is the whole stock poured in since the first coal furnaces, not last year’s flow. That stock now stands at about 1.85 trillion tonnes, up from almost nothing in 1750 — roughly nine million tonnes that year, a number the modern world emits in a couple of hours.

Split the stock by country and the bill stops being shared. The United States alone is responsible for about 435 billion tonnes, close to a quarter of everything ever emitted, more than China and Russia put together. China, despite leading the world in emissions today, has banked about 285 billion tonnes over its whole history; Russia 123. India, with four times America’s people, has emitted about 66 — a seventh of the US total. Ethiopia, with a hundred-plus million people, sits at 0.3.

This is the stock the atmosphere actually integrates, and it carries a charge the yearly flow does not. The warming the whole world now lives with was poured into the air, overwhelmingly, by a handful of countries that got rich burning coal and oil over two centuries. It is a debt with a strange shape: the countries that ran it up are, as the last chart on this page shows, the ones best shielded from what it does.

What about the greenhouse gases that aren’t carbon dioxide?

Methane and nitrous oxide add about 12 billion tonnes of CO₂-equivalent a year — roughly a third again on top of the CO₂ that gets all the attention.

Carbon dioxide is the gas everyone names, but it isn’t the only one warming the planet. Two others matter enough to track. Methane, mostly from livestock, leaking fossil fuels, and rotting waste, reached about 9.5 billion tonnes of CO₂-equivalent in 2024 — up from roughly 1.2 billion in 1850. Nitrous oxide, mostly off the nitrogen fertiliser spread on the world’s farms, came to about 2.9 billion. Both are measured in CO₂-equivalent tonnes so they can sit on the same axis as the carbon chart, weighing each gas by the warming it does over a century.

The two behave very differently. Methane is short-lived, broken down in the air within a decade or two, but while it lasts it traps heat far more fiercely than CO₂, molecule for molecule. That cuts both ways. It means a cow today is not still warming the world a century from now the way a tonne of CO₂ is. It also means cutting methane buys fast relief, one of the few levers that works on a human timescale rather than a geological one.

Add the two together and they come to about 12 billion tonnes of CO₂-equivalent. Set against the 38.6 billion tonnes of CO₂ in the previous chart, that is close to a third again on top — a whole extra slab of warming that the carbon headline quietly leaves out. The CO₂ line is the master dial, but it is not the only one turning.

If we fixed the power plants, would the problem be solved?

Electricity and heat is the single largest source of greenhouse gas at about 17 billion tonnes a year, but it is well under half of the total — there is no one switch to flip.

Break the world’s emissions apart by what they actually come from and the biggest single piece is generating electricity and heat: about 17 billion tonnes of CO₂-equivalent in 2023, roughly a third of all greenhouse gas, and nearly double its 8.6 billion in 1990. That is the part most people picture when they picture the climate problem, and it is the part the cheap-solar story is built to attack. It is also far from the whole picture.

Line the other sectors up behind it and the trouble spreads out. Transport put out about 8.5 billion tonnes in 2023, agriculture 6.2, manufacturing another 6.2, the heating and cooling of buildings 3.2. None of those is solved by a wind turbine. A cow’s stomach, a cement kiln, a cargo ship, a gas furnace in a cold flat — each is its own stubborn engineering problem, with its own timeline and its own missing technology. Clean the grid completely and you have handled the largest slice and left most of the meal untouched.

Notice which line refuses to move. Every other sector grew over these three decades; buildings barely budged, from 3.23 down to 3.20 billion tonnes, because efficiency gains and a warming planet that needs less winter heat have roughly cancelled the extra floor space. The lesson of the chart is the absence of a lever. There is no master valve. There are five or six medium valves, and most of them are harder to turn than the one we already know how to.

Are we getting more economy out of every tonne of carbon?

The world emits about half as much carbon per dollar as it did in 1990 — and emissions still hit a record, because the economy grew faster than the intensity fell.

Carbon and money have come unstuck. For every dollar of output, the world released about 0.53 kilograms of CO₂ in 1990; by 2022 it released 0.29. The same dollar of bread, steel, or software now comes with a little over half the carbon it used to, the cleanest the world economy has run in over 150 years.

This is real progress, and it is everywhere. Europe cut its intensity from 0.61 to 0.20 over those years, North America from 0.54 to 0.26, Africa from 0.47 to 0.22. Even Asia, the continent driving the emissions record, squeezed its figure from 0.46 down to 0.33 as it shifted from the dirtiest coal toward gas, renewables, and lighter industry. Not one major region got more carbon-heavy.

Here is why every line in this chart can fall and the one that matters still rises. Intensity is carbon divided by GDP; emissions are intensity times GDP. The world cut the first by roughly half while the economy roughly trebled, so the product, the actual tonnage in the air, kept climbing to a record. Running cleaner per dollar does not pull carbon out of the sky. It only means the total grows slower than the wealth does — useful, and nowhere near enough.

Whose footprint is it really, once we count what gets imported?

Counted by what people actually buy, not by where the factory stands, the average person’s carbon comes to about 4.7 tonnes a year — and the rich world’s true share is bigger than its own smokestacks suggest.

The emissions chart counted carbon where the chimney stands. This one counts it where the thing gets bought. Move a steel mill from Ohio to Hebei and the production map says America cleaned up; the consumption map says the same American still ordered the steel, so the carbon is still hers. By this accounting the average person on Earth was responsible for about 4.7 tonnes of CO₂ in 2024, up from 4.3 in 1990 — a global figure that has barely moved while the map underneath it churned.

The churn is the whole story. North America’s footprint per person fell from about 14.0 tonnes in 1990 to 11.0 by 2023, and Europe’s from 10.7 to 7.5, genuine declines. Asia’s more than doubled, from 2.0 to 4.4, as its factories took on the world’s manufacturing. But notice what the rich-world drop hides: a good slice of those falling Western numbers did not vanish, it relocated, banked instead inside Asia’s rising line as the carbon cost of goods bound straight back to Western shelves.

Zoom to the single shopper and the gap is stark. An American carried about 15.8 tonnes in 2023, an Australian 14.1, a German 9.1 — and a Chinese consumer 7.6, already above the world average. An Indian sat at 1.8, an Ethiopian at 0.2, about an eightieth of the American. The poorest people on the planet buy almost nothing with carbon in it. The footprint that warms the world belongs, overwhelmingly, to the people who can afford to shop.

How thick is the heat-trapping blanket now — and is it still getting thicker?

Carbon dioxide in the air has gone from about 278 to 426 parts per million, and more than three-quarters of that rise has come since 1950, inside a single lifetime.

Three gases do most of the trapping, and the main one has run away from its natural band. Before we started burning coal and oil the air held about 278 parts of CO₂ per million; in 2025 it holds 426, the highest reading in at least several million years. For the eight-hundred-thousand-year stretch readable from ice cores the number never climbed above roughly 300, through ice ages and warm spells alike. Then the line left the band and kept going.

The shape of the climb is the part that should land. Of the whole increase above the pre-industrial level, more than three-quarters has happened since 1950. The air sat near 313 that year, so most of this was added not in deep geological time but within the life of someone now old enough to be a grandparent. Each year we pile on more, and the gas does not leave on any human timescale, so the curve keeps steepening.

This chart has no regional version, and the absence is the honest point. A molecule of CO₂ released over Shanghai or Ohio mixes through the entire atmosphere within a year or two, so every place on Earth breathes very nearly the same air. There is no national figure to stand behind. The 426 is the blanket all eight billion of us share, and it is the master dial under the heat, the sea, and the ice that the rest of this page traces.

Has the planet actually warmed — and do independent thermometers agree?

The world is about 1.3°C warmer than the mid-20th-century average, and two separate records agree to within a tenth of a degree.

The extra carbon traps heat, and the heat is now measured past serious dispute. Two records built in completely different ways say the same thing. NASA’s GISTEMP reads surface thermometers; the Copernicus ERA5 reanalysis blends satellites into a weather model. Lay one over the other and they track to within about a tenth of a degree. Averaged over the whole surface, 2024 ran about 1.3°C above the 1951–1980 average on GISTEMP and a shade higher on ERA5, the warmest year in the instrumental record, with 2025 just behind. On the scale that drives policy, 1.3°C here is roughly 1.5°C above the pre-industrial world. That is the line the Paris Agreement was written to stay well below, now brushed in single years.

The single global figure hides a split that everyone who lives on land actually feels. Break the warming into land and sea and the lines pull apart. By 2024 the land had warmed about 1.8°C against the mid-century baseline while the combined land-and-ocean planet sat near 1.3. That gap is the ocean doing us a brutal favour. Water covers seven-tenths of the surface and has soaked up more than ninety percent of the extra heat, warming slowly because it is so vast and so deep.

So the air over your city has warmed more than the headline number admits, and an enormous amount of heat is now banked in the sea, waiting. The wiggles in the line are El Niño and La Niña sloshing heat between water and air. Smooth them out and one trend is left, bending upward, steepest in the years nearest now.

Does the global average match what any one country actually feels?

Against the 1991–2020 baseline, the world warmed about 0.7°C by 2024, but Canada warmed 1.95°C and Russia 1.25 — the far north heats at roughly twice the global pace.

The planetary average is an accountant’s number that almost no one lives inside. Measured against the recent 1991–2020 baseline — a different, closer reference than the 1951–1980 one used earlier on this page, so every figure here looks smaller — the whole world sat about 0.72°C above that mark in 2024. Canada that same year ran 1.95°C above it, Russia 1.25. The high-latitude land is warming at something close to double the global rate.

This is Arctic amplification, the white-ice feedback working overtime. As snow and sea ice retreat, dark land and water take the place of a mirror, soaking up heat that used to bounce back to space. The warming feeds on itself fastest where the ice is, which is the top of the world. So the people who live nearest the poles are handed the steepest curve, whether or not they did much to cause it.

Now read the tropics on the same axis. Brazil came in at 0.79°C above the 1991–2020 average in 2024, Indonesia at 0.88 — near the global figure, far below the northern lands. A warm, wet equator does not amplify the same way. The uneven heat is the quiet injustice under the whole subject: a single global line that hides a Canadian winter loosening twice as fast as a number meant to describe everyone at once.

If the air has warmed a degree or so, where did the rest of the trapped heat end up?

More than nine-tenths of the planet’s extra heat is in the ocean, and the heat stored in the top two kilometres has roughly tripled in twenty years.

The thermometer in the sky only measures a thin skin of the problem. The air holds almost none of the heat the carbon traps; the water holds nearly all of it. More than ninety percent of the planet’s accumulated extra energy has gone into the ocean, and the cleanest way to see it is to count the joules directly. Measured down to two thousand metres, the world ocean’s stored heat climbed from about 10 units of 10²² joules in 2005 to roughly 32.5 in 2025 — a little more than a tripling in two decades.

That number is hard to picture, so anchor it. A unit here is ten thousand billion billion joules; the rise of about 22 of them is the energy equivalent of detonating several Hiroshima-scale bombs in the sea every second, year after year, without pause. Unlike the air temperature, which jumps and dips with each El Niño, this line barely wavers. It climbs almost in a straight diagonal, which is why oceanographers treat it as the single most reliable gauge of how fast the planet is heating.

The record is short, and it is honest to say so: reliable global coverage only begins around 2005, when a fleet of drifting floats finally spanned the deep ocean. Twenty years is a sliver. But the deep point does not need a long record. The heat is not gone; it is banked. Water gives its warmth back slowly, over centuries, which means the sea will keep nudging the sea level higher and the storms wetter long after the air above it stops warming.

Is the warming actually rearranging the physical map yet?

The sea sits about 26 centimetres higher than in 1880, and the Arctic’s summer ice has shrunk by roughly a third.

The heat has stopped being a number on a chart and started moving water. Global sea level has risen about 26 centimetres since 1880, a touch over a quarter of a metre, and the climb is speeding up: a slow tide-gauge creep through most of the 20th century has become a satellite-measured surge. The rise has two sources working at once. Ice runs off Greenland, Antarctica, and the world’s glaciers, and the ocean itself physically swells as it warms. A quarter-metre sounds like nothing until it shows up on a storm surge, at high tide, against a city whose walls were built for the old shoreline.

The clearest single fingerprint is at the top of the world. Each September the Arctic Ocean’s ice shrinks to its yearly minimum, and that minimum has fallen from about 7.0 million square kilometres in 1979 to 4.7 million in 2025. That is a loss of roughly a third, an expanse of vanished summer ice larger than Greenland. And it does not just record the warming, it adds to it. White ice bounces sunlight back to space; the dark open water that takes its place soaks that heat up instead, which melts more ice, a feedback the whole planet helps pay for.

Like the temperature and the carbon, these are global by nature: one sea, one polar cap, with no honest regional version to add, only the planetary lines themselves. They are the slow, heavy proof that everything in the charts above has crossed from forecast into measurement.

Where is the water in the rising sea coming from?

Greenland has shed about 4,900 billion tonnes of ice since 2002 and Antarctica about 2,700 — and the world’s reference glaciers are losing mass faster every decade.

Two satellites flew in formation for two decades and weighed the ice from orbit, and the answer is bleak. Since 2002, Greenland has lost roughly 4,900 billion tonnes of land ice and Antarctica about 2,700, on lines that barely pause. Antarctica’s record is the jumpier of the two, because its snowfall and melt fight each other year to year, but the trend underneath both points the same way: down, and not slowing.

The number is hard to feel, so try this. A billion tonnes is a cube of ice a kilometre on each side. Greenland has dropped close to five thousand of those into the ocean in twenty years, Antarctica nearly three thousand more. That water did not vanish; it is the sea in the chart before this one, a quarter-metre higher and climbing, arriving on coasts built for the old shoreline.

The ice sheets are the giants, but the smaller glaciers tell the cleanest story, because they have been watched the longest. The world’s reference glaciers held a small surplus in 1950, about five metres of water-equivalent above their later baseline. By 2025 they sat 28 metres below it, and the curve steepens decade by decade. These are the glaciers that feed rivers a billion people drink from. When they go, the meltwater spikes, then dwindles for good.

Is all this warming actually killing more people?

Recorded disasters have multiplied since 1900, but most of that is sharper record-keeping, and the death toll has not climbed with the count.

Logged climate-related disasters have gone from about five a year around 1900 to several hundred now, even after earthquakes are taken out. The obvious reading is a planet turning violent. The honest reading is quieter. For most of the 20th century there was no satellite watch and no global disaster database at all, so a flood that drowned a valley in 1920 often went uncounted while one today is logged within hours. The count even flattens after about 2000, once the recording caught up. Some of the recent rise in floods and heat is real and physically expected as the air warms. Much of the long climb is the world simply learning to see.

The number that should rise if the planet were turning lethal is the one that has not. Over the stretch measured well, 2000 to 2021, deaths from disasters of every kind start and end near nine thousand a year with no upward trend. The huge outliers that survive in the record are mostly geophysical, not weather: the tallest, roughly 246,000 in 2004, was the Indian Ocean earthquake and tsunami, and the next two, 2008 and 2010, were a cyclone-and-quake year and Haiti’s earthquake. As forecasting and evacuation have spread, a smaller share of those caught in a disaster die, and that gain has come fastest in poor places that once had no warning at all.

There is a hard edge under the relief. Fewer people die, yet more are exposed and more is destroyed each year, because we keep building on every coast and floodplain. And the bill, in lives and in money, falls heaviest on the countries near the bottom of the emissions chart. The people who burned the least carbon live where the heat arrives first. A falling death rate is a genuine win of preparation. It is not the same as being safe.

Has the cost of clean power actually fallen — and by how much?

A solar module cost about 128 dollars per watt in 1975 and about 26 cents in 2024, a fall of roughly five hundred times — one of the steepest sustained cost declines in the history of energy.

Almost everything else in this article climbs. This one drops, and it has dropped for half a century without turning back up. A watt of solar module ran about 128 dollars in 1975, when panels powered satellites and little else. By 2000 it was down to about six dollars, and by 2024 to roughly 26 cents — close to a five-hundred-fold collapse inside a single working life.

The reason is almost boring, which is why it is so reliable. A solar panel is a manufactured product, churned out on a line, not poured on-site like a dam or a reactor. Manufactured things ride a learning curve: each time the world’s cumulative output doubles, the price drops by a roughly fixed fraction, because factories get better at making the same thing again. Panels have doubled and redoubled so many times that the curve has run for fifty years without bending up. The clean choice quietly stopped being the expensive one.

A price chart is not a finished transition, and the rest of the hope act is about that gap. But this is the hinge everything else swings on. When the cheapest way to make a unit of power also happens to be the clean way, decarbonization stops depending on conscience and starts running on the same dull arithmetic that builds anything else. That arithmetic, finally, points the right direction.

Is the electricity we make getting cleaner — and is everyone moving at the same speed?

Low-carbon sources made about 35% of the world’s electricity in 2000 and roughly 43% in 2025. A real shift, but a slow one, and split four ways underneath.

The share of electricity that comes from sources other than burning fossil fuel — wind, solar, hydro, nuclear — has climbed, and it has climbed for the right reason: clean power got cheap. Across the world it went from about 35% in 2000 to roughly 43% in 2025. That is genuine motion in the direction everyone wants.

Now look at how little ground twenty-five years bought. The world line moved only about seven points across a quarter of a century, while electricity demand itself grew, so most of the new clean power went to feed growth rather than retire coal. The world average also hides four very different countries. The United States drifted from 29 to 43%, a steady creep. China is the leap: from about 18% in 2000 to nearly 42% in 2025, more than doubling its clean share while building more power than anyone. India sits lower and rises slower, from about 17 to 27%, the curve of a country still adding coal as fast as it adds solar.

Then there is Africa, and the trap of reading it on this chart at all. Its clean share went from about 21 to 27%, but the live question across much of the continent is not how clean the grid is. It is whether there is a grid. Where hundreds of millions still have no reliable electricity, a high clean percentage of a tiny amount is not the same victory it is elsewhere.

The catch sits over the whole figure. A clean grid is not a clean economy. Electricity is roughly a fifth of all the energy we burn; the trucks, the furnaces, the cement, the ships are barely on this chart. This is the most hopeful gauge in the climate system, and it has moved the world line seven points in a generation.

If clean power’s share is rising, why is fossil generation rising too?

World electricity demand more than doubled since 2000, so fossil generation rose with it and coal hit a record in 2024 — even as the clean share climbed.

Clean power kept winning a bigger slice of a pie that kept getting bigger, and the bigger pie is the catch. World electricity demand went from about 15,300 terawatt-hours in 2000 to roughly 31,700 in 2025, more than double in a quarter-century, as billions of people plugged in air conditioners, factories, phones, and now data centres. Renewables took a rising share of that growth. They did not take all of it.

So the fossil line rose anyway. Generation from coal, gas, and oil climbed from about 9,900 terawatt-hours in 2000 to a peak near 18,300 in 2024, and coal alone set an outright record of roughly 10,500 terawatt-hours that same year — the dirtiest fuel at its highest output ever, in the very decade solar got cheap. The clean machine was mostly bolted on top of demand, not subtracted from the coal underneath it.

This is the keystone of the whole transition argument, and it is why the share chart and the tonnage chart can both be true and feel like opposites. By share, clean energy is unmistakably winning. By absolute output, fossil generation is near its all-time high. 2025 is the first hint of a turn — fossil and coal both ticked very slightly down — but one soft year is not a peak. Until the fossil line bends clearly and stays bent, the carbon keeps coming, whatever the share says.

Has everyone been reached by the grid yet — and is the gap still closing?

The share of the world with electricity climbed from 73 to 92 percent in a generation, then stalled there, leaving roughly one person in twelve still without any power at all.

Most of the planet got plugged in fast. In 1998 about 73 percent of people had electricity — enough for a light, a charged phone, a radio for a few hours; by 2023 that reached 91.6 percent. Hundreds of millions of homes went from dark to lit inside twenty-five years, a quieter victory than cheaper solar but a real one, and the precondition for almost everything else: refrigeration, schoolwork after dusk, a clinic that can store a vaccine.

Then the line flattened against a ceiling. Access sat at 91.3 percent in 2021, 91.3 again in 2022, 91.6 in 2023 — barely moving. The early gains were the easy ones, towns and dense villages a wire could reach. What is left is the hard remainder: about 8 percent of humanity, roughly one in twelve people, in places a grid struggles to follow. Remote, scattered, often poor, often at war.

Who they are matters more than the number. The unconnected are concentrated in the poorest parts of Sub-Saharan Africa and in conflict zones, the same regions barely visible on the emissions and clean-power charts. For most of this article the question has been clean power versus dirty. For these households it is power versus none — a lamp at all, before anyone argues about how it is made. The transition that the rest of the page debates has not yet reached the people who would gain the most from any of it.

Carbon goes up. Can the land pull any of it back down?

The world has lost about 200 million hectares of forest since 1990, but the map splits in two: the tropics are falling while China, India, and Europe are greening.

There are two ways to cut the carbon in the air. Stop adding it, which is every chart before this one. Or pull some back out, and the cheapest machine for that is a growing tree. A forest breathes CO₂ in; a cleared or burned one breathes it straight back. On the world ledger the lever is sliding the wrong way, though slowly. The planet held about 4.34 billion hectares of forest in 1990 and about 4.14 billion in 2025, a net loss near 200 million hectares. That is roughly the whole land area of Western Europe, gone in one human generation.

The near-flat global line is not a calm middle. It is two opposite worlds cancelling out on the page. In the tropics the forest is falling hard: across Latin America the forested share of land slid from about 53 percent in 1990 to 46 by 2022, and across Sub-Saharan Africa from 34 to 26, the Amazon and the Congo basin cleared for cattle, soy, and farmland. At the same time the temperate and Asian world is quietly regrowing. South Asia rose from about 23 to 26 percent on the back of India’s planting drives, East Asia and the Pacific edged from 26 to 27 with China’s reforestation, Europe and Central Asia held near 38 as abandoned farmland filled back in with trees.

Here is the catch under the hopeful lines. A hectare regained is not a hectare lost undone. The forest disappearing near the equator is old and dense, full of species found nowhere else; much of what replaces it elsewhere is young, often a planted single-species stand. They do not store the same carbon or shelter the same life. The greening is real and worth having. It is no fair trade for what is being cleared.

How far is the carbon line from the one that would hold 1.5°C?

To stay near 1.5°C the world needed to cut emissions about 45 percent below 2010 by 2030; instead they rose to a record 38.6 billion tonnes.

There is a line emissions were supposed to be following, and it runs the opposite way from the one they took. The IPCC’s 1.5°C report set out the arithmetic plainly: to keep warming near 1.5°C, global CO₂ had to fall roughly 45 percent below its 2010 level by 2030, then reach net zero around mid-century. Start from the 33.3 billion tonnes the world emitted in 2010 and that path drops to about 18 billion by 2030 and to nothing by 2050.

Now lay the real line beside it. Far from falling, emissions kept climbing, to a record 38.6 billion tonnes in 2024. The needed line bends steeply down; the actual line still points up. The space between them, widening every year, is the whole argument of this article rendered as a single picture. One curve is what physics asked for. The other is what we did.

The dashed pathway is not a forecast, and it is not the only route; it is one illustrative line from the IPCC’s own scenarios, drawn here to give the record number something honest to be measured against. Read it that way. The gap is not destiny, because the cheap clean machine that could close it now exists. But on the only number that finally decides the climate, the distance left to fall is the entire job, and as of 2024 it has not started.

And who actually pays for all of it?

The countries most exposed to a warming world are almost exactly the ones that did least to warm it.

Lay the two maps side by side and they are near-mirror images. The ND-GAIN index scores how exposed each country is to climate change and how poorly equipped it is to cope, across food, water, health, ecosystems, homes, and infrastructure. At the top of the danger list sit Chad, Niger, Somalia, Bangladesh, and the Democratic Republic of Congo. At the safe end sit Norway, Germany, and the United States, scoring less than half as vulnerable.

Now read the carbon those same people emit. Someone in DR Congo or Ethiopia puts out about a tenth of a tonne of CO₂ a year, a Nigerian six tenths, an Indian a little over two, against an American’s 14 and a German’s nearly 7. The places first in line for the floods, the failed harvests, and the unlivable heat are the places that burned almost nothing to bring it on. The warming was made in the rich, temperate world and is collected, with interest, in the poor and tropical one.

This is the inequality sitting underneath every other chart on the page. The atmosphere mixes one shared blanket of carbon for all eight billion of us, but the harm does not land evenly, the heat itself does not even rise evenly (the land and the high latitudes run hottest, Fig. 14), and the means to defend against it track the very wealth that came from burning the carbon in the first place. Of everything here, this is the part already decided.