Will Direct Air Capture Ever Be Affordable? The Rise of DAC 3.0
By Jonte Boysen & Torben Schreiter
Direct Air Capture (DAC) has been a prominent topic in mainstream media for over two years now and has often been portrayed through a fairly narrow lens focusing mainly on two specific technologies: Climeworks’ solid sorbent and Carbon Engineering’s liquid solvent systems. This narrow focus, however, overlooks the broader, dynamic landscape of over 80 companies innovating in this field. This article aims to dispel some common, oftentimes overly pessimistic, misconceptions about DAC and its potential. We aim to offer a more optimistic yet grounded and comprehensive view on the topic and — hopefully — inspire new angles for looking at DAC.
Apart from the abovementioned misconception that DAC systems are actually only being developed by two companies, common criticisms can be segmented into two camps:
A) Critics who categorically oppose Carbon Capture, Utilisation and Storage (CCUS), of which DAC is a part, overlooking the state-of-the-art climate change mitigation research in the most recent IPCC report, who often cite a higher (short-term) carbon impact of using the electricity needed by DAC to instead contribute green electrons to decarbonising the grid. A secondary argument focuses on current high DAC tonnage prices, which allegedly delegitimize DAC as a viable solution indefinitely. We will not spend much effort responding to these criticisms as they ignore basic cost curves and developments in grid intensities.
B) Critics who consider CCUS as a viable piece to the decarbonization puzzle, but do not believe that any DAC system will ever reach the now-famous $100/tCO2 long-term cost target. They conclude that DAC is therefore merely a distraction in the fight against climate change. But how did $100/tCO2 become such a magic number amongst insiders? What other DAC technologies are out there? And can these novel approaches reach the elusive $100/tCO2 goal, if we even have to reach it?
A brief history of the magic $100/tCO2 number
The genesis of the $100/tCO2 target traces back to the late 2000s when the current DAC incumbents were founded. At the time, many in the scientific community felt that (nascent) DAC was fighting entropy and was therefore never going to be economically viable. A 2011 report assessing potential costs using known technologies drafted by the American Physical Society (APS) estimated a minimum(!) cost of $600/tCO2, bolstering this criticism. The industry’s public rebuttal took a full seven years to arrive when Carbon Engineering published a peer-reviewed 2018 paper projecting long-term costs for their technology to between $94/tCO2 and $232/tCO2. The company publicly announced this research, stating it proved “CO2 can now be captured from the atmosphere for less than $100 per tonne”, sparking a universal benchmark for the industry. Our own historical investigation suggests that this was the moment the magic $100/tCO2 number was born.
All competitors back then, and all those that have emerged since, were subsequently under pressure to confirm they, too, could hit $100/tCO2 or risk being shunned by customers and investors.
Also in 2018, Prof. William Nordhaus received a Nobel Prize in Economics for his work on estimating the social cost of carbon (SCC), including a 2017 paper that estimated the SCC in 2050 at $102/t (obviously he and the laureates were off by a sizable margin, but that is a different story). The social cost of carbon refers to future damages associated with one tonne of carbon emissions. With the projected cost of DAC and the estimated SCC meeting for the first time at $100/t, the number’s prominence grew. It was finally set in stone in May 2020, when the ARPA-E (an agency within the US Department of Energy (DoE)) issued a funding opportunity announcement for DAC technologies that could hit this mark (see DE-FOA-0001953, Appendix P). They arrived at this number through first principles engineering and economic considerations (particularly a 30% efficiency target, corresponding to less than 500 kWh/t energy consumption). ARPA-E also benchmarked the $100/tCO2 price point against the average selling price of CO2 for industrial use at the time, which also was around $100/t (considering a wide mean variation of prices) and against emissions trading schemes prices in the US and elsewhere at that time.
What is interesting with the ARPA-E analysis — and what aligns well with how we are looking at the offtake markets here at Extantia as well — is that price sensitivity in the market is actually quite comparable with regards to both green CO2 as a feedstock for utilisation pathways (thus circular solutions such as for our portfolio company INERATEC) or permanent sequestration (thus removal). Both markets will be highly relevant for DAC companies going forward.
The Commodification of CO2: A Market Perspective
Comparing the $100/t target cost for CO2 to tonnage prices of other basic materials and traded commodities yields various insights. It is very hard to find any commodity at all that trades below $100/t. Only very few commodities, such as coal, cement, iron ore and barley, are close contenders to this price point (all somewhere between $100–200/t). Each of the first three are today produced at gigatonne scale globally — a size insiders expect the negative emissions industry to reach one day (ideally minus most of the logistical challenges of the mining industry). The only thing we know of (other than drinking water), that you can get for substantially less than $100/t, is something we also bury underground at gigatonne scale: gravel and sand (selling at around $35–70/t). It becomes clear that when we talk of green CO2 to sell at sub-$100tCOe, we quite literally mean “dirt cheap” carbon removals at a profound scale.
One factor that is often overlooked in the $100/tCO2 discussion, however, is inflation. The 2018 Joule paper that led to Carbon Engineering’s $100/tCO2 claim used 2016 dollars. In today’s (2024) money, the famous number would therefore be closer to $130/tCO2. Looking ahead to 2030 and 2050 and assuming a very modest 2% annual inflation, the nominal figure rises to $145/tCO2 and $215/tCO2, respectively. Further justification of a higher viable cost is provided by recent calculations from the US and the German Environment Agency (UBA) putting the SCC today at $190/tCO2 and €237/tCO2, respectively. Also see Figure 8 down below for reference of an inflation-adjusted view on different CO2 target prices.
Coming back to the Nobel Prize-winning price of $102 that was originally suggested for 2050, we have already observed the EU ETS cap-and-trade market effectively reach this level 25 years early (high in 2023 was €100.23/tCO2). Keeping in mind future inflation along with likely additional developments resulting in potentially higher carbon prices, it is fair to ask whether we need to get down all the way to the magic number of $100/tCO2. We can certainly afford to engage in a more nuanced discussion about how all these factors influence the acceptable cost of DAC over time. From today’s standpoint (2024), it seems quite likely that the (inflation-adjusted) $100/tCO2 is actually more of a lower bound on the viable price range for DAC, rather than an absolute maximum price.
What is clear is that DAC will take decades to (really) scale and that costs will gradually decrease as more capacity is deployed. Using our crystal ball, we can envision a future where DAC never actually reaches the $100 price tag due to the combined and ever-changing effects of scale-up/cost curve effects, inflation and higher SCC market expectations. In simpler terms, we don’t expect to see a $100 price tag for DAC-sourced CO2 in the next few years (despite very meaningful cost improvements), because we are still too far up the cost curve. In the late 2040s, we will approach the bottom of the cost curve, but potentially still not see a $100 price tag, because both inflation and higher carbon market price expectations might have driven tonnage values up. This certainly would neither be a negative indicator for DAC/CCUS at this point nor disprove the intent and thinking behind the magic $100/tCO2 number. An economically viable price for DAC-sourced CO2, and therefore the highest-quality CDR, would still be achieved.
But which technologies will actually be capable of moving down the cost curve to this ballpark of economically viable DAC prices?
Exploring DAC’s Technological Diversity
Where we agree with the DAC critics, is that we also think the $100/tCO2, or even the inflation-adjusted $130/tCO2 figure, is likely out of reach for Climeworks’ and Carbon Engineering’s technologies. Climeworks itself recently suggested that a cost as high as $300/tCO2 (in 2023 dollars) is feasible for 2050. Where we disagree with the critics, however, is that $100/tCO2 may very well be in reach for entirely different DAC technologies.
New technology platforms — few people outside the DAC community are even aware of — allow for substantially steeper learning curves that can drive costs down to $100/tCO2 or even lower. More of these novel technology platforms exist than most people think and new ones continue to emerge.
The number of DAC startup companies has grown from 5 in 2013 to at least 67 by the end of 2022 (the year of the almighty US Inflation Reduction Act as a defining moment for CCUS). In 2023 the number of DAC startup companies grew again to around 80.
This explosion in the number of companies was also accompanied by a vastly increased diversity of approaches. Although DAC technologies vary in many ways, systems can be categorised by both their capture type and release mechanism. While there were only liquid solvent and solid sorbent approaches until as recently as 2018, we have since seen membrane-based and cryogenic solutions emerge. Furthermore, there is a rich diversity even within solid sorbent approaches based on innovation in novel materials capable of capturing CO2. These include not only amines, but metal organic frameworks, zeolites, hydroxides, and many more.
When categorising DAC technologies based on their CO2 release mechanism, we see the same trend: diversification only happened in recent years. Low-grade heat and high-grade heat release mechanisms, together called thermal approaches, have long dominated the DAC industry. Due to accelerated innovation in the last few years, however, the toolbox now contains at least seven strategies for dealing with captured CO2. We prefer some of these new release drivers, like voltage and pH, as they avoid the parasitic energy losses of thermal approaches and therefore often allow for a much lower energy consumption. At least as importantly, the novel capture and release approaches also open up new ways for startups to innovate on reducing capital expenditures (e.g. due to mass manufacturing of modularised key components).
The Next Wave of Innovation
With so many new approaches, the obvious question is whether any room for true innovation remains. Historical trends suggest so and even give us an idea of where it will come from. DAC technologies can be divided into three generations. DAC 1.0 arose in the late 2000s and mostly employs well-understood and proven technical approaches with acknowledged energetic limitations — most famously Climeworks and Carbon Engineering. Starting around 2018, DAC 2.0 introduced novel and previously untested capture and release approaches. This includes membrane-based capture approaches like RepAir Carbon (no liquids) and Mission Zero Technologies (liquid solvent-based), which both recently worked their way out of lab-scale at light speed announcing pilot field deployments, reaching TRL 6 within three years from incorporation. Also notable is Verdox, which pioneered an entirely voltage-driven capture and release process, leveraging a proprietary redox-active compound.
DAC 3.0 — to Extantia’s observation — is a third wave of companies that started to emerge around 2021 when DAC 2.0 had seemingly more or less run out of entirely new approaches. The name of the game became driving innovation on a process level by creating new capture-release permutations that had never been tried before, or combining multiple capture or release approaches in one system. This generation of companies includes the likes of Greenlyte Carbon Technologies or Parallel Carbon, which have swapped out thermal approaches for a liquid solvent capture and electrochemical release approach (Greenlyte) or following a passive air contacting strategy and releasing via pH swing (Parallel Carbon). Maia — also a DAC 3.0 company — is combining multiple release mechanisms such as low-grade heat and voltage for favourable bottom line economics. These and other companies have been introducing a creative fusion of processes and mechanisms, pioneering unique approaches that sidestep conventional energy (or cost) penalties. Most interestingly, the vast majority of DAC companies have by now progressed in technological maturity (i.e. TRL) towards solving mainly engineering challenges rather than fundamental research challenges, which makes the technologies VC-investable.
Why Technological Diversity Enables $100/tCO2
The reason we are so interested in the technological diversity of DAC approaches is the associated increase in the aggregated likelihood of achieving an acceptable levelised cost of capture (a.k.a. $100/tCO2). Different technologies tend to follow different cost learning curves that determine a potential cost floor. The more shots on goal, the greater the likelihood of one solution having a sufficiently high learning rate. The learning rate of a technology is typically defined as the per cent decrease in cost achieved by every doubling of total installed capacity. We project a learning rate for DAC 1.0 similar to that of coal power at 8%, as both are derived from relatively established components put together in relatively bespoke engineering designs. This learning rate implies a gigatonne scale DAC 1.0 cost of roughly $250/tCO2.
DAC 2.0 and 3.0, on the other hand, tend to rely on more unique components and are built on less established technology platforms, suggesting learning rates closer to those of wind power at 12% and gas power at 15%. We believe their typical modularity and system complexity is closer to that of wind power than that of gas power, resulting in projected learning rates of 13% for DAC 2.0 and 14% for DAC 3.0. These learning rates both enable $100/tCO2 cost at gigatonne scale, though exact current costs and scales are difficult to determine. Importantly, however, this shows that reasonable learning rates allow the novel DAC technologies to achieve the magic $100/tCO2 target.
In our analysis, we are applying the same learning rate to both Capex and Opex, which seems a reasonable simplification for the model given the relatively high energy consumption at the current very small scale. Sequestration costs are also included with the same learning rate. In target locations for DAC (e.g the US Permian Basin), sequestration is already today at or below ~$10/tCO2, making it a comparably marginal cost item to total DAC/CDR cost (src1, src2).
The broadening array of DAC technologies, underpinned by varying learning rates, presents a mosaic of potential pathways to the elusive $100/tCO2 target. Despite the cost limitations of DAC 1.0, we also still see tremendous value in supporting incumbents come down the cost curve. They are pulling the rest of the industry up with them and provide a gradually decreasing ceiling for society on the price of durable carbon removal. We therefore hope all three DAC generations continue to innovate, collaborate, and prove the DAC doomers wrong. In this spirit, we also want to thank the DAC Coalition for its thoughtful and inclusive work in service of the industry.
If you are working on a novel DAC solution that can achieve the $100/tCO2 target, please reach out. At Extantia, we believe that technological pluralism not only fosters a rich environment for breakthroughs but also underscores the importance of supporting the continuum of DAC research and development to scale carbon removal. The journey ahead to scale is long, but — while it certainly is not a silver bullet — DAC is playing a crucial role as an important piece of the puzzle in averting the climate crisis.
Many thanks to Gregory Thiel and Benjamin Bronfman for helping us understand more about the history of the $100/tCO2 sound barrier.
We would also like to thank Jonte Boysen, who has been with us as Entrepreneur in Residence for the past few months. You have contributed significantly to our CDR activities and we wish you the best of success in your upcoming endeavour in the space.
UPDATE: We have added a paragraph to better explain our assumptions on the influence of different partial cost drivers on our total cost per tonne extrapolation. Thank you for your feedback.
