You've probably seen the photo: a shipping-container-shaped unit sitting on a black lava field in Iceland, steam curling off it, the caption promising that this machine pulls carbon dioxide straight out of the sky. It's a powerful image. It's also, on its own, almost meaningless.
The question worth asking isn't whether direct air capture works. It does. The question is whether the startups building it are actually helping the climate, or whether they're selling a very expensive story that lets everyone else keep polluting. After following this sector closely since the first commercial plants came online, I've come to a conclusion that will annoy people on both sides: direct air capture is real, necessary, and currently far too small to matter—and those three things are true at the same time.
Let's untangle what these companies actually do, what the numbers really say, and where the honest limits sit.
Key Takeaways
- Direct air capture (DAC) physically removes CO₂ from ambient air, but the net climate benefit depends entirely on the energy source and the full lifecycle.
- Costs have fallen but remain in the range of hundreds of dollars per tonne—orders of magnitude above the price of simply cutting emissions.
- Most startups sell carbon removal credits to fund operations, so their climate impact is inseparable from whether those credits are verifiable.
- The technology is best understood as a cleanup tool for legacy emissions, not a substitute for reducing new ones.
- Watch the measurement and verification (MRV) practices, not the press releases.
What direct air capture startups actually do
Start with the physics, because it explains everything that follows. CO₂ makes up roughly 0.04% of the air around you. Capturing it means pushing enormous volumes of air through a material that grabs onto those trace molecules, then releasing them in a concentrated stream you can store or use.
That dilution is the whole problem. A cement plant's exhaust is roughly 10–15% CO₂ by volume. The atmosphere is a fraction of a percent. You are chasing a needle in a haystack that is the size of the sky.
The two main technologies
Most companies fall into one of two camps, and the distinction matters more than any marketing deck admits:
- Liquid solvent systems — air bubbles through a chemical solution that binds CO₂. Reliable, well understood, but energy-hungry, because you must heat the solvent to release the gas and regenerate it.
- Solid sorbent systems — air passes over a solid material that captures CO₂ and releases it when heated or exposed to a vacuum. Often lower-temperature, which can pair better with renewable heat, but the materials degrade and need replacing.
- A hybrid question that rarely gets answered publicly: how much of the captured CO₂ is released again before it reaches permanent storage?
Climeworks, the Swiss company behind that Iceland plant, uses a solid sorbent approach and stores the captured CO₂ underground in basalt, where it mineralizes over time. That geological route is what makes the removal durable—more on why "durable" is the word that actually matters.
Why the energy source decides everything
Here's the part that gets glossed over. A DAC plant runs on electricity and heat. If that energy comes from a coal plant, you can burn more carbon running the machine than you capture. The net climate impact of the entire sector lives or dies on this single variable.
This isn't a hypothetical. It's the reason the credible players go to great lengths to source clean power and siting. When I first dug into the lifecycle accounting, I'll admit I expected it to be a footnote. It's the whole story.
What the numbers and costs really look like
DAC is one of the most expensive ways to deal with carbon ever invented. Current costs sit somewhere in the hundreds of dollars per tonne of CO₂ removed—commonly cited in the range of a few hundred dollars, with the most optimistic projections aiming for around $100 per tonne at scale. That scale does not exist yet.
Compare that to reforestation, which can cost tens of dollars per tonne, or to simply not emitting in the first place, which is often free or profitable. Anyone claiming DAC is competitive on cost today is selling you something.
The cost comparison that matters
| Removal method | Typical cost per tonne (rough order) | Durability of storage | Land / energy footprint |
|---|---|---|---|
| Direct air capture | Hundreds of dollars | Very high (geological) | High energy, low land |
| Reforestation | Tens of dollars | Low to moderate (fire, decay, reversal) | Very high land |
| BECCS | Moderate | High if stored geologically | High land and biomass demand |
| Biochar | Low to moderate | Moderate (centuries) | Moderate land |
Read that table honestly and a pattern jumps out: the methods that are cheap are the ones whose carbon can escape back into the air, and the methods whose carbon stays put are the ones that cost a fortune. You are paying for permanence. That's the trade, and it's not a flaw in DAC—it's the entire reason the technology exists.
How much has actually been removed?
Here's a sobering framing. Global DAC capacity, even counting every operating plant, removes a tiny fraction of what humanity emits in a single day. We're not talking about a meaningful slice of the problem. We're talking about a rounding error with excellent PR.
That doesn't make it worthless. It makes it early. A technology that removes a rounding error today is a technology whose cost curve you're trying to bend before you need it at scale. Whether it bends fast enough is genuinely unknown.
The greenwashing question nobody wants to ask
Most DAC startups don't make money selling CO₂. They make money selling carbon removal credits to companies that want to offset their emissions. That business model creates a direct incentive problem: the more generous your climate claims, the more credits you can sell.
Which is why measurement, reporting, and verification—MRV—is the battlefield that actually decides whether this sector helps or hurts. A credit is only worth the tonnes it verifiably, permanently removed.
What to check before believing a claim
- Is the storage permanent, or could the CO₂ return within decades? Geological storage and mineralization are strong; "we'll figure it out" is not.
- What energy powers the plant, and is that disclosed in the lifecycle accounting?
- Is the removal independently verified by a third party with a public methodology?
- How much of the captured CO₂ is sold for products like fuels or fizzy drinks—where it's released again almost immediately?
- Does the startup's own marketing distinguish between removal and avoided emissions? These are not the same thing and should never be blurred.
That last point trips up a lot of people. Using captured CO₂ to make synthetic fuel means it gets burned and re-released. Using it in a beverage means the same. Only permanent storage delivers a durable climate benefit.
So is direct air capture actually worth it?
Here's where I'll take a firm position, and you can disagree. In my view, DAC is worth funding precisely because it's expensive and hard—not despite it. There is a category of emissions we cannot eliminate: the CO₂ already in the air, the residual from agriculture, the hard-to-abate industrial processes. For those, you need a cleanup tool. Reforestation can't scale to cover them without competing for land we need for food.
But the framing that gets sold—"DAC lets us keep emitting"—is dangerous and wrong. If DAC becomes an excuse to delay cutting emissions, its net climate impact turns negative, no matter how many tonnes each machine captures. The technology is a complement to reduction, never a replacement.
The honest limits
- Cost must fall by roughly an order of magnitude to become broadly deployable.
- It needs abundant clean energy, which competes with everything else that wants clean energy.
- The sector's credibility depends on MRV rigor it hasn't fully earned yet.
- Scale today is negligible against global emissions.
What to watch from here
The story of DAC startups won't be decided by the machines in Iceland or the funding rounds in the press. It'll be decided by boring things: the price per tonne on the cost curve, the honesty of the credits, and whether governments eventually put a real price on carbon that makes removal economically rational rather than charitable.
If you're trying to judge whether any of these companies is genuinely helping the climate, ignore the lava-field photographs. Find the lifecycle report. Find the verification methodology. Find out where the CO₂ ends up and whether it stays there for a thousand years or a season. That's where the impact lives—buried, unglamorous, and far more revealing than any image of a machine that looks like it's saving the world.
