Climeworks and Carbon Engineering: Two approaches to scaling Direct Air Capture (DAC)
Climeworks is taking on the role of prime contractor, while Carbon Engineering is working oil and gas firms like Occidental Petroleum to build many projects, fast
You’re reading Terraform Now, my newsletter on carbon capture and other planet-altering technologies. To support my work, you can subscribe here:
Two leading Direct Air Capture (DAC) firms, Carbon Engineering and Climeworks, are taking different approaches to the problem of sucking CO2 out of the sky. As I wrote last time, the two companies are using very different technology. But as important as the tech itself is how it is deployed — the business model. Here too Carbon Engineering and Climeworks differ.
The way we’ll do this is by looking at the DAC plants that Climeworks and Carbon Engineering are building, Mammoth and Stratos:
Before going any deeper on these projects, I’d like to introduce the idea of a prime contractor (prime from now on), borrowed from the aerospace and defense industry. Primes are companies that specialize in project-based work, like building a line of fighter jets, developing an oil field, or producing a movie. They are called ‘primes’ because they are the main contractor, responsible for sub-contracting out smaller parts of a project.
Primes are companies with a lifespan that is longer than the project itself. The concept is useful for our purposes because DAC facilities are discrete projects with heavy up-front capital, just like fighter jets, oil fields, and movies.
Back to Climeworks and Carbon Engineering: the most important difference is that Climeworks wants to be a prime, while Carbon Engineering is choosing to be a sub-contractor to experienced chemical engineering primes like Occidental Petroleum (Oxy below).
Here’s more detail on the roles that Climeworks and Carbon Engineering are playing on their next-gen plants:
There are many choices for these firms to make, from construction to geological storage to marketing. But the decision to be or not to be a prime has the most far-reaching consequences.
Risk / reward: Primes raise capital and line up partners to design, build, and maintain the project. To return to our movie analogy, primes are like producers, who find a screenplay, line up financing, and pick the director, actors, etc. Like a producer, primes are paid if the project succeeds, and lose money if it fails. As some recent box-office flops demonstrate, there is real risk to being a prime. Instead, a DAC firm could just license its tech like Carbon Engineering. While this is less risky, it also means limited upside.
Skills: By sticking to technical development, Carbon Engineering focuses scarce talent on fewer problems. It can continue to hire top-end engineers capable for prototyping, and design-oriented engineers to architect new plants. By contrast, Climeworks will need to build a whole new organization, including hundreds or thousands of employees in project management, quality control, and automation. Their leadership will be spread across everything from invention to day-to-day plant operations.
Control: The prime makes the key decisions. When it comes to DAC, the key decision is whether to use vs. store captured CO2. Climeworks will be able to choose what it does with captured CO2, balancing the environmental advantages of storage with the commercial upside of utilization markets, like recycled fuels. Carbon Engineering has left this decision up to Oxy, an oil company which may use it for things not all environmentalists would approve of. More on that later.
Time-to-scale: In the short-term, Carbon Engineering’s licensing strategy allow it to work on more plants at once, and with partners capable of making bigger plants. Fast forward to 2025 or 2027, and Carbon Engineering will have an order of magnitude more DAC capacity than Climeworks. But the outcome a decade from now is less determined, since Climeworks could master scale and leapfrog licensing-based competitors.
The implications of scale are the main focus of this article, because the relationship between lower cost and more scale is central to the long-term success of any DAC business. Going one click deeper on upcoming plants for both Climeworks and Carbon Engineering will enable us to explore DAC scaling dynamics in the real world.
EVERYTHING’S BIGGER IN TEXAS
Occidental Petroleum (Oxy from now on) broke ground on what will be the largest DAC plant in the world, Stratos, extracting about 500,000 tons of CO2 each year. Oxy will be using Carbon Engineering’s liquid-based DAC tech to capture CO2, and though the details aren’t public this is presumably through a licensing agreement with additional engineering services.
People who don’t work in construction or engineering — myself included — underestimate the difficulty of scaling up. It seems like scale just involves replication, but of course it’s not that simple. Invention, prototyping, and design are all vastly different from scale in terms of how people are hired, trained, and organized. Where invention involves a few dozen engineers, scale requires thousands of people working in the same way. Scale requires standardization, quality control and an agile management that can make adjustments. This means that reliable and concise communication up, down, and across the organization are a must.
To illustrate the point, consider the US car industry. In the last 100 years, hundreds of American car companies have built prototypes and sold a few hundred or even thousand cars. But only one company, Tesla, managed to scale to a million cars per year, and it was so hard that Tesla almost went bankrupt multiple times. In the car industry, Ford-like scale is a wide moat. It is so hard to do well that it keeps new entrants from even trying.
Oxy, like Ford, is over 100 years old and specializes in scale; its area is chemical engineering. From Carbon Engineering’s perspective, Oxy is playing the part of committed Prime. Many oil and gas players are making noise about negative emissions technology, but Oxy is arguably doing the most to make it a reality. It hopes to build a carbon capture business unit that does ~$2B revenue per year by the early 2030s. This has attracted far-sighted investors, with heavy hitters like Warren Buffet increasing their holdings as recently as June 2023.
Carbon Engineering sees many advantages to working with Oxy. With all its experience, Oxy has the reputation to finance massive projects. It has pre-existing relationships with companies like Worley, a world-class construction firm for chemical engineering that says it will build the Stratos plant in just 18 months. And Oxy has whole teams of permitting experts who can get every bit of these projects approved fast — from the DAC facility itself, to transport infrastructure, and even geologic storage.
Speaking of storage, Oxy also has plenty of geologists on hand to help solve the tricky problem of permanently storing captured CO2 underground. In fact, Oxy already stores more CO2 underground than any other company.
Because such a strong scaler like Oxy exists, Carbon Engineering is not currently specializing in standardization, quality control, and all the rest. That means it can focus on improving its core technology and planning the next project; Carbon Engineering already designing another massive plant with Oxy.
NOT ALL SUNSHINE AND RAINBOWS
The main downside to working with a prime like Oxy is that Carbon Engineering will have less control, which means it will have less say in important decisions:
Energy source: Oxy produces natural gas, and will be tempted to use this as a power source instead of purely renewable options. There’s nothing wrong with this, since the system as a whole will still likely be carbon negative. But it’s not ideal: a natural-gas based system is worse for global warming than a renewable-based system1.
CO2 utilization vs. storage: Oxy has not shared whether CO2 it captures will be stored underground vs. used for other purposes, like recycled fuels and industrial CO2 use cases. I mentioned earlier that Oxy is the world leader in storing CO2 underground, but it does this almost exclusively with Enhanced Oil Recovery (EOR), a process that sends CO2 into almost dry oil wells to pump more oil out of the ground. Depending on your point of view, this is either fine or very bad2.
Learning: Carbon Engineering will outsource many technical problems to Oxy and other primes, which means they will not learn as quickly.
The key question Carbon Engineering will have to grapple with is how these primes will behave as long-term partners. There is a clear tension between Carbon Engineering’s learning goals and the profit-maximizing goals of primes like Oxy.
Carbon Engineering and its competitors are focused on bringing the cost to capture a ton of CO2 down from $500-$1000 today down to ~$100 (or lower) in the next 5-10 years.
Oxy is less focused on the goal of cost declines through learning, at least across projects. It wants the Stratos plant to be profitable for as long as possible, and if more profitable plants come online, Stratos may be forced to sell carbon offsets at a loss. Put differently, as Oxy learns how to improve DAC processes and thereby reduce costs at Stratos, it may hesitate to relay those learnings back to Carbon Engineering for use in future plants, and especially with other primes. Chevron, for example, is said to be interested in working with Carbon Engineering.
There’s real potential that these divergent incentives hobble Carbon Engineering’s ability to learn from scale, since this pattern will be repeated at just about every plant. The only way out of this dilemma is to work mostly with one prime, but that would limit Carbon Engineering’s ability to grow.
A final note on Carbon Engineering before we move to Climeworks. Carbon Engineering is the only DAC company with liquid-based tech, which is more like the large chemical engineering processes that Oxy, Chevron, and other firms already do. That means that Carbon Engineering may have many primes aiming to license its technology in the coming years.
CLIMEWORKS: COMMAND AND CONTROL
Climeworks is in a different position than Carbon Engineering, and again this comes down to technology. Air-based DAC doesn’t have much in common with normal chemical engineering processes — it would feel very unfamiliar to a potential prime3. This makes the ‘be my own prime’ a much more attractive option for Climeworks, Global Thermostat, Octavia Carbon, and everyone else working on air-based tech.
Even if Climeworks’ tech was very attractive to large primes like Oxy, it’s unlikely that Climeworks would want to sub-contract to an oil and gas firm. Culturally, Climeworks seems more focused on creating an environmentally sound solution than on hyperscaling. That makes its’ approach less pragmatic, yet more in tune with the long-term direction of DAC — renewable energy only, responsible sourcing, and (mostly) geologic storage.
This, however, will not be easy. Climeworks will need to get good at a lot of things, all at the same time:
Invention, prototyping, and design of the core technology
Scaling up, including project management, quality control, etc.
Learning at scale (more on this later)
Sourcing only renewable energy
Geologic storage, potentially in continued partnership with Carbfix
Marketing and selling carbon offsets
So Climeworks has to compete with Carbon Engineering’s technology development teams and Oxy’s project management storage teams and 1PointFive’s marketing and sales machine. That’s a lot of things to do well!
There is real risk to messing up any one of these steps, but there’s also a large payoff for Climeworks if they manage to get each bit right. Climeworks, unlike Carbon Engineering, can claim a truly environmentally friendly solution, and will be able to credibly market its carbon offsets as the highest quality offsets. This is already proving to be of real value, as recent deals with JP Morgan and others show.
Being a prime also gives Climeworks the benefit of control. By tightly controlling the process end-to-end, Climeworks has the opportunity to learn and drive costs down faster than competitors. The question is how well will Climeworks grasp this opportunity. Can it develop top tier quality control? And can it move up the value ladder to Toyota-style production-line innovation?
The complexity of Direct Air Capture means that there are many opportunities to improve the process and to make mistakes. Balancing strong quality control and production-line innovation is hard, and requires trust in individual line workers, quality of middle management, and the wisdom of company leaders all have to be top notch* Organizations with these qualities are not built overnight.
The good news is that Climeworks’ process is complex, but not as complex as, say, making cars. All the same, Climeworks should study what made a company like Toyota so revolutionary — root cause analysis, trust in line workers to pause the process, etc. — and apply it to building and maintaining modules of their Mammoth plant. This is the most reliable path to leapfrogging Carbon Engineering and other air-based DAC players. Becoming the DAC volume leader, known for low emissions and quality offsets, is the only recipe for sustained interest from customers and investors.
SCALE-COST VIRTUOUS CYCLE IN DAC
I’ll end with some observations about scale and costs in Direct Air Capture. First, some basic principles:
the first company or technology that can deliver DAC at low cost (say $100 per ton) will attract massive investment and customer interest, allowing them to get big fast
the first technology that scales rapidly is likely to deliver lower costs due to learning curves (sometimes called experience curves)
So there’s a virtuous cycle here — more scale and lower cost reinforce each other. That’s the obvious part — what’s hard is sorting out the rate of decreased cost that comes from scale. Anyone that tells you they know how quickly DAC will get to $100 per ton (or any other number) is likely making things up.
Besides general uncertainty, there are two choices that Carbon Engineering and Climeworks have made that mean their learning curves will have different slopes:
Technology: water-based (Carbon Engineering) and air-based (Climeworks) technologies are just very different, so we should expect different rates of cost decreases
Business model: Carbon Engineering and Climeworks are taking more- or less-hands-on approaches to how their tech is deployed, which means their pace of learning will be different, which means we should expect different rates of cost decreases
Again, if one company claims to have a faster learning curve, consider that marketing. By coincidence, they might look very similar, with cost declining at the same rate as scale increases. If this is close to the actual state of the world, that would be bad for Climeworks, since Carbon Engineering’s next plant is more than 10x the size of Climeworks’ Mammoth.
That’s not to say Climeworks should be counted out — we already discussed the many reasons that Mammoth, despite being <10% the size of Stratos, could yield much faster learning if managed right. But learning much faster might not be enough. Take two simple scenarios* — one where scenario where Climeworks learns twice as fast as Carbon Engineering, and another where it learns 20x as fast:
Climeworks can’t just reduce costs faster than Carbon Engineering — it has to reduce costs an order of magnitude faster.
Of course this analysis is full of unknowable assumptions. I assume that Carbon Engineering and Climeworks are starting out with a similar initial cost, which is unlikely given how different their technologies are. And it assumes returns to scale that, right now, are impossible to predict. This means we can’t make either of the following claims:
Carbon Engineering will be able to lower costs faster due to its 15x scale advantage
Climeworks will be able to lower costs faster due to its tight control over its Mammoth plant
The implications for each company are profound.
Climeworks should accept that they aren’t going to match Carbon Engineering in terms of scale in the next 2-5 years. All things being equal more scale leads to lower costs (which leads to more scale by attracting investment). Climeworks needs to make sure that all things are not equal by bending the learning curve with the goal of radically reducing cost. Proof of costs reducing rapidly will allow them to raise significant capital and leapfrog the competition.
By contrast, Carbon Engineering should recognize its early lead and carefully capitalize on it, continuing to license to primes and further extending its capacity lead. Yet Carbon Engineering risks being spread too thin, across too many projects, and outsourcing the learnings that come from scale to primes. If these relationships are poorly structured, those learnings will not flow back to Carbon Engineering. If it becomes clear that air-based companies are learning faster, Carbon Engineering’s early lead in Direct Air Capture could evaporate4 in just a few years.
DAC does not seem like a ‘winner-take-all’ market to me — there will be many big DAC companies in the future. But given the virtuous cycle between scale and cost, and the first company (or technology) to prove that they are in that virtuous cycle will attract outsize funding, both from investors and customers. Climeworks and Carbon Engineering, and other firms all have opportunities to benefit from that virtuous cycle. I’m excited to see how it all shakes out.
The opinions in this article are my own and do not reflect the position of my employer, Bain & Company
This is true even if they promise to install carbon capture tech in the smokestack of the natural gas plant, since these plants require more energy as a result and will still emit some CO2 — emissions’ source tech can only capture ~90% of the CO2 in any smokestack. And of course there’s the opportunity cost of this energy.
EOR is distasteful to some environmentalists simply because it pumps more oil to come out of the ground, but this is a complex topic. I’ll write more on EOR at a later date, since it is a beautiful example of how skills in oil & gas are essential for mitigating climate change
Climeworks’ and other air-based DAC players’ tech most resembles Emissions’ Source (often called point-source) carbon capture. But even Emissions’ Source is a sub-scale technology without any major primes, and DAC is different enough that these players may not be comfortable taking on Climeworks’ technology. For more on the difference between DAC and Emissions’ Source technology, check out Terraform Now’s Carbon Capture Primer
Pun intended
Opinions expressed on Terraform Now do not represent the views of my employer, Bain & Company. Facts presented here are publicly sourced unless otherwise noted.
This is an excellent analysis. Thank you! How does the acquisition of Oxy impacts this analysis? On one side Carbon Engineering will have access to capital from Oxy (coming from their core fossil fuels business). On the other side, there might be a negative impact to the entrepreneurial culture of Carbon Engineering. What else?