Sucking CO₂ out of the atmosphere costs 20× more than not emitting it

The article analyzes Direct Air Capture (DAC) technology for removing CO₂ from the atmosphere and shows that it is orders of magnitude more expensive ($230–1,000/t) than preventing emissions ($0–50/t). Current global DAC capacity captures as much CO₂ in a year as humanity produces in under a minute, and scaling it to the level needed would require trillions of dollars and terawatts of clean energy. DAC has its place in offsetting unavoidable emissions, but it must not serve as an alibi for postponing systemic change — ending the burning of fossil fuels, protecting forests, and transforming industry.

Imagine your neighbour floods your cellar with water every day. Instead of going over to talk to your neighbour, you keep building ever more powerful pumps. This is exactly what the current debate about carbon dioxide storage looks like. Direct Air Capture (DAC) technology promises to suck CO₂ straight out of the atmosphere and store it underground. But the numbers behind that promise tell a different story — a story about how vast amounts of money and energy are flowing in the wrong direction.

The Earth's atmosphere today holds roughly 3,340 billion tonnes of CO₂ — half as much again as before the Industrial Revolution. Every year another 42 billion tonnes is added from burning fossil fuels, industrial processes, deforestation and agriculture combined. To bring the concentration back down to at least 350 ppm, which the climatologists around James Hansen consider the upper bound of long-term safety, we would have to remove about 600 billion tonnes from the atmosphere. This article shows why such a task, though technically feasible, is economically and energetically absurd — and why there are paths that are orders of magnitude more effective.

The total global capacity of DAC technology at the end of 2024 amounted to about 59,000 tonnes of CO₂ per year. The largest existing plant, Climeworks Mammoth in Iceland, has a nameplate capacity of 36,000 tonnes — but over its entire first year of operation (2024) it captured a mere 105 tonnes, because only 12 of the planned 72 collector containers had been installed and the new generation of filters failed sooner than the manufacturer expected. Meanwhile the world emits 42 billion tonnes a year. Even if the entire current DAC capacity ran at full power, in a year it would capture as much CO₂ as humanity produces in under a minute.

The International Energy Agency's (IEA) scenarios for reaching net zero by 2050 envisage DAC capacity of over a billion tonnes of CO₂ per year — that is more than 17,000× more than we have today. And that is merely the volume needed to neutralise hard-to-abate emissions (aviation, cement, agriculture), not to lower the overall atmospheric concentration.

If we wanted to actively draw the atmosphere down to 350 ppm over 50 years, we would need to remove 12 billion tonnes per year. Where do we put it all?

Liquefied CO₂ cannot be stored on the surface for the long term — at atmospheric pressure the liquid phase does not exist, so surface tanks would require permanent overpressure and cooling. On the scale of hundreds of cubic kilometres that is technically untenable.

In practice, two options come into play: geological storage at depths greater than 800 metres, where the natural pressure and temperature keep CO₂ in a supercritical state without any active cooling, and mineralisation — converting CO₂ into solid carbonates in basalt formations. Both work, as demonstrated by the Sleipner project in Norway (running since 1996, with an estimated 19–23 million tonnes stored — although a 2024 investigation by the DeSmog website revealed that the operator Equinor had overstated the declared volumes by roughly 28% over the 2017–2021 period) and CarbFix in Iceland (95% of the CO₂ turns into rock within two years).

The problem is scale. A typical injection well can take about a million tonnes of CO₂ per year over 30–40 years. To remove 12 billion tonnes per year we need roughly 12,000 simultaneously active wells. That in itself does not sound insane — the oil industry operates millions of wells worldwide. But those 12,000 wells require a pipeline network, compression stations, geological surveys, permits and monitoring lasting centuries. The London Register of Subsurface CO₂ Storage from 2025, the first audited global overview, states that since 1996 humanity has stored a total of 383 million tonnes of CO₂ underground. We need a pace roughly 30× faster — and to sustain it for half a century.

Capturing one tonne of CO₂ from the atmosphere with DAC technology consumes 1,500 to 3,000 kWh of energy depending on the technology — solid-sorbent systems (Climeworks) sit at the lower end of the range, liquid-solvent systems (the former Carbon Engineering, now part of 1PointFive) at the upper end. That is 12 to 23× more than the thermodynamic minimum allows (~130 kWh/t). Adding compression, transport and injection, we reach 1,900–3,200 kWh per tonne.

To remove 600 billion tonnes we need a total of roughly 4,200 to 6,900 exajoules — seven to twelve times the world's annual primary energy consumption. Spread over 50 years, this means a continuous draw of 2.6 to 4.4 terawatts of clean energy, that is 14–24% of today's global production.

And this is exactly where the fundamental snag lies: if we generate that energy from fossil fuels, we emit more CO₂ in producing it than we capture. The US electricity mix, with an emission intensity of ~0.4 kg CO₂/kWh, would make the process purely counterproductive. DAC makes sense only with low-carbon energy — solar, wind, nuclear, geothermal.

But those very sources are scarce. And if we have terawatts of clean energy at our disposal, there is an incomparably more efficient way to use it: to replace fossil fuels directly. Every kWh of clean energy used in DAC is a kWh that did not replace a coal or gas power plant.

The current cost of DAC ranges between $600 and $1,000 per tonne of CO₂. Climeworks co-founder Jan Wurzbacher admitted in May 2024 that Mammoth's real operating costs are "closer to $1,000 per tonne than $100." A study by ETH Zürich (Sievert et al., Joule, 2024) offered more sober long-term projections: upon reaching a cumulative capacity of 1 Gt CO₂/year, it estimates costs at $230–540 per tonne — not the $100 promised by marketing materials. The industry will not reach such capacity before around mid-century.

At a mid-range estimate of $200 per tonne, removing 600 billion tonnes would cost $120 trillion. That is more than the world's annual GDP. For context: total global investment in renewable energy in 2024 amounted, according to BloombergNEF, to $728 billion, and according to IRENA to as much as $807 billion.

And what do we get for that money? Cooling of about 0.8 °C — a return to roughly the temperature level of the turn of the millennium.

Comparing the costs and capacities of the individual measures reveals chasmic differences. The transition to renewable sources (replacing coal) costs $0–50 per tonne of CO₂ and can cover over 15 Gt per year — it is scalable today. The energy efficiency of buildings actually saves money (−100 to 0 USD/t) with a potential of around 5 Gt per year. Halting deforestation costs 5–50 USD/t and likewise covers about 5 Gt — it rests on political will. Afforestation and landscape restoration, at a similar price (5–50 USD/t), can fix 3–5 Gt, but require decades of growth. BECCS (bioenergy + CCS) sits at 100–200 USD/t with a potential of 2–5 Gt, still in the pilot-project phase. And at the bottom of the ranking stands Direct Air Capture: 230–1,000 USD/t, current capacity 0.00006 Gt per year, an early stage of development.

The world's forests — tropical, temperate and boreal combined — absorb around 14.4 billion tonnes of CO₂ a year, while deforestation, fires and other disturbances release on average 9.2 billion tonnes. The net uptake is thus roughly 5.3 billion tonnes of CO₂ per year — still more than double what the entire European Union emits. And the forests do it for free, without electricity, without wells, without pipelines. On top of that they add water-cycle regulation, biodiversity and soil protection.

But this system is breaking down fast. WRI/Global Forest Watch data from 2025 show that in 2023–2024 forests absorbed only a quarter of the usual amount of CO₂ — massive fires in Canada, the Amazon, Siberia and Bolivia released over 4 Gt of greenhouse gases per year, two and a half times more than in an average year. The south-eastern Amazon has already become a net source of emissions. Some boreal forests in Canada have likewise shifted from sinks to sources.

Even so, we lose roughly 10 million hectares of forest a year. Halting deforestation would cost a fraction of DAC's cost: the REDD+ programme estimates $5–15 per tonne of CO₂ retained in the forest stand.

Restoring degraded agricultural land and afforestation can, over the coming decades, fix another 3–5 billion tonnes of CO₂ per year. It is not a substitute for cutting emissions — trees grow slowly and have limited capacity — but the cost/performance ratio is incomparable with any industrial technology. And as 2023–2024 show, not even forests are a reliable guarantee: climate change is threatening the very sinks we rely on.

It would be intellectually dishonest to say that DAC is useless. There are emissions that cannot be eliminated even with maximum effort: cement plants (the calcination process releases CO₂ chemically, not just by combustion), long-haul aviation (batteries for a transatlantic flight are beyond the physical possibilities of current technology), agriculture (methane and nitrous oxide from livestock and fertilisation).

These "residual" emissions amount to an estimated 5–8 billion tonnes of CO₂ equivalent per year even in maximum-decarbonisation scenarios. Some form of capture — whether DAC, BECCS, or enhanced weathering — will be needed to offset them. IPCC scenarios consistently assume negative emissions on the order of 5–15 Gt CO₂/year in the second half of the century.

Thermodynamically, DAC works: the waste heat from the process is roughly 200× smaller than the cooling benefit of the reduced greenhouse effect. Physics is not the problem. The problem is economics and scale.

Hopes for a breakthrough do exist. Stratos, the plant of 1PointFive (a subsidiary of Occidental Petroleum) in Ector County, Texas, has a nameplate capacity of 500,000 tonnes per year and, after repeated delays, the start-up of the first phase (250,000 tonnes) is expected in the second quarter of 2026 — full operation in the second half of the year. It is currently supplied with energy by a freshly commissioned 500 MW solar park. If it reaches its planned output, it will be capturing nearly ten times more CO₂ than all DAC capacity to date. The question remains whether the gulf between project ambitions and operational reality can be bridged — Mammoth, which in its first year captured 0.3% of nameplate capacity, counsels caution.

Climeworks, meanwhile, has announced a third generation of the technology: new structured sorbents promise double the capture rate, half the energy consumption and triple the filter lifetime. The goal is to cut costs to $250–350 per captured tonne by 2030. The first deployment will take place under Project Cypress in Louisiana (1 Mt/year), construction of which is due to begin in 2026. It is an ambitious leap — but even if it succeeds, we remain far below the gigatonne scale that the climate scenarios require.

There is one more dimension that deserves attention. The largest DAC and CCS projects are financed by oil companies — Occidental (the $1.3 billion Stratos project, with a $550 million investment from BlackRock), Shell, Chevron, Equinor. Roughly 80% of all captured CO₂ today is used for Enhanced Oil Recovery — injection into old oil wells in order to extract another barrel of oil. Stratos too will use part of the captured CO₂ for EOR, as Occidental CEO Vicki Hollub confirmed.

That is not in itself a bad thing — if the CO₂ stays underground, it is better there than in the atmosphere. But it is important to realise that the CCS/DAC industry has a structural incentive to keep the fossil economy going, because without it it loses both its main source of revenue (EOR) and its main source of funding (oil companies). Some jurisdictions, such as California, have already banned the use of EOR in carbon-removal projects — a signal that this link is becoming ever more controversial. The debate about CO₂ storage is therefore not purely technical — it is also political and economic.

The numbers are uncompromising. Stopping CO₂ production costs on the order of $0–50 per tonne. Capturing it back out of the atmosphere costs $230–1,000. Stopping the felling of forests costs $5–15 per tonne and, on top of that, delivers ecosystem services worth hundreds of billions a year — although, as 2023–2024 show, even this capacity is fragile and threatened by climate change itself. Planting a tree costs a few dollars. Building an injection well with infrastructure costs tens of millions.

This does not mean that DAC has no place in the climate portfolio. It means that it cannot be an alibi for postponing real systemic changes: an end to burning fossil fuels, the preservation and restoration of forests, the transformation of agriculture and industry. Every dollar invested in DAC instead of these measures is a dollar wasted.

Twelve thousand wells into the Earth is an impressive engineering vision. But the cheapest, fastest and most reliable form of carbon storage lies in not emitting it at all. And for what is already in the atmosphere, we need every tool — forests, landscape restoration and technology alike. Only in the appropriate order of priorities.

Methodological note: This article draws on data from NOAA (CO₂ concentrations), the Global Carbon Budget 2024 (emissions and budgets), the IPCC AR6 (climate sensitivity and scenarios), the IEA Direct Air Capture 2022 and World Energy Outlook (technology and costs), the London Register of Subsurface CO₂ Storage 2025 (stored volumes), WRI/Global Forest Watch 2025 (forest carbon flux), the AlliedOffsets DAC Tracker 2025 (DAC capacities), the study by Sievert et al. (Joule, 2024; DAC cost projections), the IRENA Global Landscape of Energy Transition Finance 2025 (renewable investment), DeSmog (October 2024; the Sleipner audit), Latitude Media and C&EN (Mammoth performance), Hart Energy and Energy Intelligence (the status of Stratos) and the technical specifications of Climeworks, 1PointFive/Stratos and CarbFix. Data cut-off: 20 February 2026.

Disclosure of the creative process:

The concept, structure and editorial line of the article are the work of the author, who prepared the content outline, set the key theses and directed the entire creative process. Generative AI (Claude, Anthropic) was used as a technical tool for research, fact-checking and fleshing out the author's draft.

The author edited the outputs throughout, verified the key findings and approved the final wording. No part of the text was published without human review. All factual data were verified against the publicly available sources cited in the text.

The procedure complies with the transparency requirements for AI-generated content under Art. 50 of EU Regulation 2024/1689 (the AI Act). #poweredByAI