Direct air capture: which startup is ahead?

Which direct air capture startups are worth comparing?

The useful direct air capture comparison today contains 12 independent startups with meaningful funding, operating equipment, contracted removals or a credible commercial project.

We focus on companies developing their own DAC process and aiming to supply permanent carbon removal, even when they also sell captured CO₂ for fuels, drinks or materials. Funding totals are minimum publicly disclosed amounts. Private startups often announce one round while keeping earlier grants, strategic investments or project finance confidential.

We exclude 1PointFive because it belongs to Occidental Petroleum. Carbon Engineering and Holocene have been acquired, so they no longer compete as independent startups. Deep Sky provides shared project and storage infrastructure rather than relying mainly on its own capture technology. Skytree and Aircapture are more focused on supplying CO₂ for industrial use, while this comparison gives greater weight to permanent removal.

Startup	What it is building	Minimum publicly disclosed funding or backing
Climeworks	Solid-sorbent DAC plants connected to geological storage	More than $1 billion in equity
Heirloom	Limestone-based DAC using electric kilns and passive carbonation	At least $203 million, plus later undisclosed strategic investments
CarbonCapture	Factory-produced modular DAC machines using replaceable sorbent cartridges	At least $115 million
Avnos	Water-positive DAC powered partly by low-grade waste heat	At least $53 million disclosed; more than $100 million in combined backing claimed
Spiritus	Passive-air DAC using a low-cost proprietary sorbent	At least $30 million, plus earlier undisclosed funding
Mission Zero	Electrochemical DAC requiring electricity but no high-temperature heat	£21.8 million Series A, plus earlier funding
Greenlyte	DAC combined with hydrogen and renewable-fuel production	More than $26 million
Noya	Modular, water-positive DAC using moisture-responsive chemistry	About $22 million
Phlair	Electrochemical DAC designed to run around intermittent solar power	€14.5 million seed round
RepAir	All-electric electrochemical DAC for permanent removal	$15 million Series A
Octavia Carbon	Modular DAC using Kenyan geothermal power and local geological storage	$5 million seed round
Airhive	Mineral-sorbent DAC using fast fluidized-bed equipment	Funding undisclosed

Is there a clear direct air capture leader today?

Climeworks is ahead today. The race is still mainly Climeworks against Heirloom, but Climeworks’ recent plant data has widened the gap a little.

The gap between sales and delivery explains why the leadership remains unsettled. CDR.fyi’s latest detailed DAC review counted 2.47 million tonnes contracted between 2022 and the first half of 2025. Only 1,186 tonnes had been delivered. Buyers had therefore ordered more than 2,000 times the volume that suppliers had completed.

Climeworks owns most of the field’s operating history and delivered removals, and it is the only startup that has published a large improvement at a plant above 10,000 tonnes. Heirloom owns the strongest combination of pure-DAC contracts and near-term construction. CarbonCapture and Airhive form the next group because both now have systems operating at around 1,000 tonnes or more in annual nameplate capacity. Which direct air capture companies lead and which trail is laid out in the market grids of our report on carbon capture.

Mission Zero has deployed more separate systems than most challengers. Phlair has attracted a large contract at a lower implied price. Avnos has a well-funded commercial demonstration under development. Spiritus has ambitious cost and scale targets but little verified operating evidence.

Which direct air capture startup has actually removed the most CO₂?

Climeworks has removed and stored the most CO₂ among DAC startups. Mammoth is now producing much more than in its first year, but still far less than its 36,000-tonne nameplate suggests.

CDR.fyi estimated that Climeworks supplied 81% of DAC deliveries recorded by the middle of 2025. That implies roughly 960 tonnes from the industry total of 1,186 tonnes. No independent startup was close.

Mammoth produced 873 gross tonnes during its first thirteen months, according to figures reported by the Financial Times. After emissions from the removal process were deducted, net removal fell to 205 tonnes. Against roughly 39,000 tonnes of proportional nameplate capacity over thirteen months, that equals about 2.2% gross utilization and 0.5% net utilization.

Mammoth was still adding collectors and moving through commissioning in that period. In September 2026, Climeworks said the plant removed 675 net tonnes in the first half of 2026, against 119 tonnes in the first half of 2025, a gain of about 5.7 times. Some reports rounded it to fivefold and others to sixfold, but both describe the same company figures. Annualized, the first-half figure is still below 4% of the 36,000-tonne rating, and the figure is company-stated rather than independently audited.

Heirloom’s Tracy plant has a 1,000-tonne annual rating and has fulfilled customer purchases, but Heirloom has not released a comparable full-year production figure. CarbonCapture’s Tamarack project achieved first capture at a 2,000-tonne site. Airhive started operating its 1,000-tonne Alberta system. Neither has yet published a full year of independently verified removal.

Startup	Approximate operating nameplate	What has been publicly demonstrated
Climeworks	About 40,000 tonnes a year	Clear delivery lead; Mammoth reported 873 gross and 205 net tonnes over its first 13 months, then 675 net tonnes in the first half of 2026, more than five times a year earlier (company-stated)
CarbonCapture	2,000 tonnes a year	Tamarack achieved first capture in October 2025; sustained annual output undisclosed
Airhive	About 1,060 tonnes a year	Alberta system opened in November 2025 and Teesside pilot online; sustained annual output undisclosed
Heirloom	1,000 tonnes a year	Tracy operating and supplying customers; Louisiana plant not yet declared operational; full-year verified production undisclosed
Mission Zero	About 550 tonnes a year	Three systems operating across the UK and Canada; annual removal totals undisclosed

Which DAC startup has learned the most from building and running real plants?

Climeworks has learned the most from building and operating complete DAC plants, while Mission Zero has accumulated the broadest range of smaller deployments.

Climeworks has run Orca since 2021 and Mammoth since 2024. Those plants combine air capture, heat, vacuum equipment, CO₂ processing and Carbfix’s underground mineral storage. The company has years of data on weather, maintenance, filter cycles, sorbent degradation and storage integration.

Mammoth’s weak first year exposed the problems that appear when DAC grows from several thousand tonnes to tens of thousands. Fixing them is now part of Climeworks’ lead. In September 2026 it said two upgraded collector containers had run for more than six months at around 40–50% capacity factor, and it plans to upgrade all 12 containers in one Mammoth module by the end of 2026. Competitors working at pilot scale have not yet had to learn these lessons.

Climeworks has coordinated construction, geothermal energy, CO₂ handling, measurement and permanent storage across two Icelandic sites. Project Cypress, its planned Louisiana hub, received an initial $50 million US government award shared by the project team, although federal support for DAC hubs has since become less certain.

Mission Zero has taken a broader small-system route. Its Sheffield plant supplies CO₂ for fuel research, its Norfolk system connects DAC with building materials, and its Alberta system connects with geological storage. The three deployments expose the same platform to different climates, customers and downstream equipment.

Airhive and CarbonCapture are learning inside Deep Sky Alpha in Alberta, where several DAC suppliers share transport, measurement and storage infrastructure. CarbonCapture’s modular Leo equipment has also moved from a factory concept into a 2,000-tonne operating site.

Heirloom has operated its Tracy facility for more than two years and brought in Chiyoda as an investor and engineering partner for its Louisiana expansion. The next test is whether it can move limestone, run electric calcination and automate the process reliably at 17,000 tonnes.

Which direct air capture startup has the strongest customers and contracts?

Heirloom has the strongest pure-DAC contract book today, while Climeworks has the largest and most diverse carbon-removal customer network.

Microsoft agreed to purchase up to 315,000 tonnes from Heirloom. Frontier buyers contracted another 26,900 tonnes for $26.6 million. United Airlines Ventures obtained the right to purchase as much as 500,000 tonnes, although that agreement is a purchase option and carries less weight than a firm delivery commitment. Firm contracts versus options show what buyers will really pay for, a question we study in our market report about carbon capture.

Microsoft’s agreement alone is close to the total annual capacity Heirloom eventually wants to build across its two Louisiana facilities. The demand and construction plans fit together unusually well.

Climeworks has signed prominent buyers across technology, finance and industry. Its disclosed agreements include 40,000 tonnes for Morgan Stanley and 31,000 tonnes for Schneider Electric. Climeworks Solutions also announced 14 new agreements covering roughly 450,000 tonnes during the first half of 2026, then a CORSIA-focused purchase agreement with Japan Airlines in September 2026, with volume and price undisclosed.

The newer Climeworks figure includes portfolios containing several carbon-removal methods. Some tonnes may come from biochar, enhanced weathering or other suppliers, so the agreements say more about Climeworks’ commercial reach than the demand for its own DAC machines.

CarbonCapture also deserves attention. Frontier buyers agreed to pay it $20 million for 45,500 tonnes. Phlair later secured $30.6 million for 47,000 tonnes. Both contracts came from experienced buyers and include defined delivery volumes.

Which direct air capture startup is gaining ground fastest right now?

Airhive is gaining ground fastest among the DAC challengers currently trying to catch Climeworks and Heirloom.

Airhive entered operational ramp-up at its 1,000-tonne Alberta system, started commissioning a 60-tonne pilot in Teesside and joined Mission Zero in the UnionDAC consortium. UnionDAC targets an initial 20,000 tonnes of annual capacity, followed by another 40,000 tonnes.

Airhive then acquired Carbyon, adding a second technical platform and a team working on fast-cycling thin-film sorbents. The company now combines operating hardware, mineral-sorbent expertise and another route for improving capture speed and energy use.

Phlair is also accelerating. Its 10-tonne outdoor pilot has been operating since early 2025, Frontier is financing a 47,000-tonne offtake, and the company recently added WISAG as a customer and project partner. Phlair is also involved in a phased Norwegian project that could eventually reach 500,000 tonnes, although most of that scale remains on paper.

Spiritus raised a $30 million Series A, established a New Mexico pilot site and continued development around its proposed Wyoming Carbon Orchard. Its passive-air design and sub-$100 target are eye-catching, but public plant data remains limited.

Which DAC startup is closest to making carbon removal affordable?

Nobody has proved cheap DAC yet. CarbonCapture has the lowest implied large contract price, while Airhive, Phlair and Climeworks’ upgraded Mammoth offer the most interesting newer cost evidence.

BloombergNEF has placed current industry capture costs around $900 per tonne. Airhive says credits from built DAC plants have recently remained near $1,000. These numbers remain far above normal carbon-credit prices and the support available from most public incentive programs.

Dividing CarbonCapture’s $20 million Frontier contract by 45,500 tonnes produces an implied price of about $440 per tonne. Phlair’s $30.6 million contract for 47,000 tonnes implies about $651. Heirloom’s $26.6 million agreement for 26,900 tonnes comes to roughly $989.

Contract prices can include financing, verification, storage, delivery risk and margins, so they should not be confused with production costs. CarbonCapture’s price is still notable because an experienced buyer accepted it for a large, defined volume.

Climeworks said in September 2026 that Mammoth’s operating costs had fallen by more than half, but it did not publish a cost per tonne. Airhive expects its Alberta system to capture CO₂ for less than $500 per net lifecycle tonne after ramp-up, excluding transport and storage. Climeworks targets $250 to $350 per captured tonne by 2030 after missing earlier cost-reduction goals. Spiritus claims a pathway below $100, but its public evidence currently comes from modelling and technical targets.

Company or benchmark	Approximate public price or cost	How much confidence should we place in it?
Current DAC industry	Around $900–$1,000 per tonne	Based on market estimates and existing DAC credit prices
CarbonCapture and Frontier	About $440 per contracted tonne	Strong buyer evidence, though future delivery risk remains
Airhive Alberta system	Below $500 per net captured tonne	Company calculation excluding transport and storage
Phlair and Frontier	About $651 per contracted tonne	Strong buyer evidence for future deliveries
Heirloom and Frontier	About $989 per contracted tonne	Strong buyer evidence from an earlier first-of-a-kind contract
Climeworks	$250–$350 per captured tonne targeted	Future target that still needs plant-scale proof
Spiritus	Below $100 per tonne targeted	Early company projection with little operating validation

Which direct air capture technology uses energy and water best?

No DAC startup has proved the best full-system energy and water performance, although Phlair, Avnos and Airhive currently have the clearest differentiated designs.

Climeworks has produced the strongest large-scale test data. Its Generation 3 equipment completed more than 1,300 cycles using structured adsorbents developed with Svante. Climeworks reported twice the capture capacity and half the energy use of its previous design. The company is aiming for roughly 1.5 megawatt-hours per tonne.

Phlair also targets less than 1.5 megawatt-hours per tonne. Its electrochemical process can separate regeneration from air contact, allowing parts of the system to follow cheap solar generation while stored liquids keep the capture process running. That flexibility could matter as much as the headline energy number. Energy use per tonne is one test of which capture technologies work today and which remain on paper, as we show in our carbon capture market report.

Airhive publishes unusually specific resource figures. Its 1,000-tonne system is expected to use around 2 megawatt-hours per tonne across capture and regeneration. Water consumption reaches a maximum of 0.6 kilograms per kilogram of CO₂ in dry air and falls as humidity rises. The company says no additional water is required at 100% humidity.

Avnos has the strongest water proposition. Project Cedar is designed to capture 3,000 tonnes of CO₂ and produce more than 6,000 tonnes of clean water each year. Its process can also use low-grade waste heat, which creates a possible fit with data centers and industrial sites that already need cooling and water.

Mission Zero avoids high-temperature regeneration and has demonstrated a system powered entirely by solar electricity in Alberta. The published figures remain too uneven for a clean ranking because companies use different boundaries for electricity, heat, construction and storage.

Has fixing Mammoth strengthened Climeworks’ lead in direct air capture?

Yes, partly. The Mammoth upgrades Climeworks reported in September 2026 make its lead more solid, but more than $1 billion of funding still buys a smaller lead than it should.

The money funded Orca and Mammoth, two commercial equipment generations, a large innovation center, certification work and relationships with many of the world’s largest carbon-removal buyers.

It has also funded expensive mistakes and slow ramp-ups. Mammoth’s first-year utilization stayed below 3% on a gross basis, and net removal was lower again. Climeworks cut a meaningful share of its workforce as capital became harder to raise and several government-backed projects faced uncertainty.

The company still raised another $162 million and used it to work on the plant it already had. In September 2026 it said Mammoth’s net removal in the first half of the year was more than five times the level a year earlier, and operating costs per tonne had fallen by more than half. That is the first public sign that the money is fixing real production, not only funding new designs. Generation 3 also performed well across more than 1,300 test cycles. The tension remains: a gain of that size from a very low base is still far from nameplate, and Generation 3 has yet to run inside a full commercial plant.

Climeworks’ expansion into managed removal portfolios provides revenue and keeps corporate buyers close, though it makes commercial growth less dependent on the performance of Climeworks’ own DAC technology.

Can Heirloom still overtake Climeworks before its Louisiana plant runs?

No. Heirloom cannot take first place on contracts alone; its first 17,000-tonne Louisiana plant has to run first.

Heirloom has presented that facility as coming online in 2026. It has not publicly announced the start of commercial operations, so the plant remains a construction milestone rather than operating capacity. Every month without a start date while Mammoth improves makes the catch-up harder.

A 17,000-tonne facility would be seventeen times larger than Heirloom’s Tracy plant and equal to roughly 42% of Climeworks’ current operating nameplate.

Heirloom then targets a 100,000-tonne phase in 2027, followed by another 200,000 tonnes. Reaching the first two phases would give it around 117,000 tonnes of annual capacity, nearly three times the combined ratings of Orca and Mammoth.

Microsoft’s 315,000-tonne agreement gives Heirloom a buyer for almost the same volume as its wider Louisiana plan. Few DAC startups have aligned one major customer with one major buildout so neatly.

The engineering risk remains high. Heirloom is attempting a seventeenfold first jump, followed by another increase of almost six times. Reliable operation at 17,000 tonnes would bring it close to a tie. Successful operation above 100,000 tonnes would probably put it first.

Is CarbonCapture the best modular DAC startup?

CarbonCapture is the strongest modular DAC startup today because Tamarack has moved the Leo design from a factory concept to an operating 2,000-tonne site.

Each Leo unit fits inside the dimensions of a shipping container and is rated above 500 tonnes annually. Tamarack therefore represents roughly four modules working as one project, creating a useful test of repeatable manufacturing, installation and maintenance.

CarbonCapture can change sorbent cartridges rather than tying every machine permanently to one material. Better sorbents could enter an established hardware platform without forcing the company to rebuild the whole system. Hardware that can swap sorbents is one business-model bet, and we compare which models keep their margins in our report on carbon capture.

We still need to know how often the modules operate, how quickly sorbents degrade, how much maintenance each container requires and whether the factory can produce dozens of identical units at a predictable cost.

Tamarack has achieved first capture. A full year of high availability would carry much more weight.

What can the leading DAC startups do that rivals cannot easily copy?

Climeworks has the strongest operating-data advantage, while Heirloom and CarbonCapture may have systems that are easier to industrialize.

Climeworks has years of maintenance records, weather data, filter-cycle information and storage experience from Iceland. A competitor can buy similar fans and study similar sorbents. Recreating years of failure data and operating adjustments takes much longer.

Heirloom’s raw material is ordinary limestone, yet the process around it is specialized. The company must control carbonation speed, particle handling, hydration, calcination, heat recovery and automation across millions of repeated cycles. Its partnerships with electric-kiln and engineering specialists add knowledge that cannot be copied from a patent alone.

CarbonCapture’s advantage could come from manufacturing. A standardized machine, qualified suppliers and repeatable servicing procedures can create a stronger position than one exceptional sorbent. Its cartridge approach also gives the company flexibility when materials improve.

Airhive now combines fluidized-bed engineering with Carbyon’s fast-cycling sorbent research. Mission Zero has patents and experience connecting its electrochemical process to three different downstream uses. Phlair is building around load flexibility. Avnos combines CO₂ capture, water production and waste-heat use inside one system.

Large industrial companies remain a serious threat to every startup. Occidental already owns Carbon Engineering and is building through 1PointFive. Oil, chemical and engineering groups have the capital, subsurface knowledge and project teams required for large plants.

Can we trust the numbers published by direct air capture startups?

We can trust verified stored tonnes. Nameplate capacity, cost targets and distant project pipelines require much more caution.

The strongest evidence is a measured tonne that has been captured, stored and independently certified. Full-year plant production comes next because it reveals downtime, energy consumption and maintenance under real conditions.

A binding offtake with a defined buyer, quantity, price and delivery period is useful commercial evidence. A purchase option, reservation or memorandum of understanding carries less weight because the buyer may never take the full volume.

Nameplate capacity only describes the plant’s intended output. Mammoth’s rating and first-year net production were separated by two orders of magnitude, and even a fivefold gain does not close that gap. The same caution should apply to every planned Heirloom, Phlair, Spiritus or CarbonCapture facility.

Cost figures also need clear boundaries. Capture alone excludes compression, transport, storage, verification, financing and some lifecycle emissions. A contract price can differ substantially from the supplier’s production cost.

Which direct air capture startup is actually ahead?

Climeworks is the direct air capture startup ahead today, with Heirloom the only rival close enough to take first place if its Louisiana plant works as planned.

We give Climeworks the lead because it has delivered the most removal, operated the largest startup-owned DAC fleet, learned from complete capture-and-storage plants and built the broadest commercial organization. Mammoth’s reported recovery in September 2026 strengthens that position, although the plant still runs well below its rating.

Heirloom comes second through its contracts, financing and credible path toward much larger capacity. It remains behind because its Louisiana plant has not yet been declared operational and its planned scale has not become verified production.

CarbonCapture takes third through Tamarack and a genuinely repeatable modular design. Airhive follows closely after a rapid sequence of deployment, technology development, project partnerships and acquisition activity.

Mission Zero ranks fifth because it has repeatedly put machines into different operating environments. Phlair ranks sixth because its commercial contract and energy design are strong, although its current pilot remains tiny. Avnos has a well-backed 3,000-tonne project under development. Spiritus has high upside but much thinner operating evidence.

Heirloom’s 17,000-tonne plant is the most likely event to change first place. A full year of upgraded Mammoth output, then a first Generation 3 plant, are Climeworks’ best chances to widen the lead.

Rank	Startup	Why it ranks here now	What would change the ranking
1	Climeworks	Leads verified delivery, operating experience, installed nameplate and customer access; Mammoth net removal more than fivefold higher year on year in the first half of 2026 (company-stated)	A verified full year of upgraded Mammoth output would secure the lead; a stalled recovery would threaten it
2	Heirloom	Strongest pure-DAC contracts and most credible large startup construction pipeline	Starting and running the 17,000-tonne Louisiana plant could create a near tie; 100,000 tonnes would probably put Heirloom first
3	CarbonCapture	Operating 2,000-tonne modular site, repeatable hardware design and strong contracted pricing	Needs a full year of reliable output and several replicated sites
4	Airhive	Fastest recent momentum, operating 1,000-tonne system, UnionDAC role and Carbyon acquisition	Independent validation of its sub-$500 cost could move it into the top three
5	Mission Zero	Three deployed systems across different uses and climates	Needs to convert small deployments into a plant above 10,000 tonnes
6	Phlair	Large Frontier contract, load-flexible electrochemistry and attractive implied price	Must move from a 10-tonne pilot to commercial delivery
7	Avnos	Well-backed 3,000-tonne Project Cedar and a distinctive water-positive design	Project Cedar must enter operation and publish energy, water and output data
8	Spiritus	Strong investors, passive-air design and ambitious cost targets	Needs independently measured pilot performance before its megatonne plans carry weight
9	Octavia Carbon	Strong geothermal and geological-storage position in Kenya	Commissioning of its 1,000-tonne project must become verified annual removal
10	Greenlyte	Commercial partnerships and a differentiated DAC-plus-hydrogen system	Needs more evidence of permanent removal rather than fuel-focused CO₂ use
11	Noya	Credible funding and a water-positive modular approach	Recent commercial deployment evidence remains limited
12	RepAir	Promising all-electric electrochemical process	Needs a larger operating system, binding demand and public performance data