Burning coal, oil and gas, making cement and clearing forests together release roughly 40 billion tonnes of CO₂ a year. A billion tonnes is a gigatonne (Gt), so that is about 40 Gt CO₂ per year. The ocean and land plants soak up roughly half of it. The rest stays in the air.

Think of the atmosphere as a bathtub. Emissions are the tap; the ocean and land are a drain that can't keep up. The water level is the amount of CO₂ in the air, and it is the level, not the flow from the tap, that sets how warm the planet gets. Before the Industrial Revolution the level was about 280 ppm (parts per million of air); it is now over 420 ppm.

The tub also drains very slowly. Some of a pulse of CO₂ is absorbed within decades, but the ocean and rocks take far longer to mop up the rest. Roughly a fifth to two-fifths of it is still in the air after a thousand years. CO₂ is not like smoke that clears after a few days. It accumulates.

Because warming tracks the total CO₂ ever added, scientists can work backwards from a temperature limit to a remaining carbon budget: how much more CO₂ the world can emit, in total, and still have a given chance of staying below that limit. The budgets are uncertain and get revised, but the message is stark. By the mid-2020s, the budget for an even chance of holding warming to 1.5 °C was down to only a handful of years of current emissions, and the 2 °C budget to roughly a quarter of a century's worth.

The cheapest tonne of CO₂ to deal with is the one never emitted, so cutting emissions comes first. But cuts alone can't finish the job, for two reasons.

First, some emissions are very hard to eliminate: parts of farming, aviation, shipping, cement and steel. These residual emissions have to be balanced by removing an equal amount of CO₂, otherwise the level never stops rising. Reaching net zero means the CO₂ going in equals the CO₂ taken out. Second, if the world overshoots a temperature limit, the only way back is to remove more than is emitted, called net-negative emissions, and lower the stock itself.

That is why every pathway assessed by the UN's climate science panel, the IPCC, in its latest (sixth) assessment report that limits warming to 1.5–2 °C includes carbon dioxide removal (CDR): deliberately taking CO₂ out of the air and storing it. Removal is not a substitute for cuts; the pathways need both, at the same time.

The figures come from the 3rd edition (2026) of the "State of Carbon Dioxide Removal" report, an independent assessment that tracks the field. Total removal is about 2.2 Gt a year, and its most ambitious 1.5 °C scenarios reach around 8.75 Gt by 2050. A megatonne (Mt) is a million tonnes, so novel removal, at about 2 Mt a year in 2026, has to grow several thousand times over in about 25 years. That gap is the opportunity, and the challenge, for every removal company.

Carbon cycles through the Earth system on two very different clocks.

In the fast carbon cycle, carbon swaps between the air, plants, soils and the surface ocean within years to centuries. A leaf takes in CO₂ in spring and its carbon is breathed back out by microbes a year or two after it falls. A tree might hold its carbon for a century, until it burns or rots.

In the slow carbon cycle, carbon moves between the surface and rocks over hundreds of thousands to millions of years. Volcanoes release CO₂; silicate weathering turns it into bicarbonate that rivers carry to the sea (see Weathering, erosion & soil); marine life and chemistry bury it as limestone and organic-rich mud. This is the thermostat that has kept the planet habitable through Earth's history.

Fossil fuels are slow-cycle carbon: plants and plankton buried for millions of years. Burning them moves that carbon into the fast cycle in a matter of decades, far faster than the slow cycle can take it back.

That is what durability means: how long the removed carbon stays out of the air. Storing a tonne in a forest helps, but if the forest burns in 40 years the tonne comes back. Turning a tonne into dissolved bicarbonate in the ocean, or into carbonate rock, returns it to the slow cycle.

Every method has two halves: capture (getting CO₂ out of the air) and storage (keeping it out). Plants are superb at capture but store carbon in fragile forms. Rocks and the deep ocean are superb at storage but slow or costly at capture. Each method below is a different trade-off. Costs are given only roughly, because they change quickly and vary between projects.

There is no single winner. Biological methods are cheap and available now but can be reversed. Engineered methods are durable and easy to measure but expensive. Geochemical methods like ERW sit in between: durable and potentially cheap, but slow and hard to measure. Most experts expect the world to need a portfolio of all of them.

Most removal today is paid for by selling carbon credits: certificates that each represent one tonne of CO₂. A buyer can use a credit to claim that tonne against its own emissions. Because a credit is a claim about something invisible, a small vocabulary has grown up to say exactly what is being claimed.

Durable removal is mostly sold in the voluntary carbon market, where companies buy credits by choice to meet their own climate targets, rather than because a law requires it. Four kinds of player are involved.

A removal credit goes through a lifecycle. The details differ between registries, but the shape is the same everywhere:

The middle steps together are called MRV: measurement, reporting and verification. For ERW, MRV is the hardest and most important part, and it gets its own module: Measuring removal (MRV).

On price, only a cautious statement is safe. Durable engineered and mineral removal has been selling for hundreds of dollars per tonne in early offtake deals. That is far above cheap avoidance credits, which have often traded for a few dollars a tonne. Buyers pay the premium because a durable, well-measured removal tonne is a much stronger claim, and because they hope early purchases will drive costs down, as happened with solar panels.

ERW takes the slow-cycle thermostat and turns the speed up. Everything you learned about weathering applies: choose a mafic rock that weathers fast, grind it to increase surface area, and put it where warmth, water, roots and microbes drive the reaction. It has some real strengths, and some real difficulties.