Thanks for your post. I would love to read more of what's behind your predictions on the cost curve of DACCS coming down. I worry that the underlying logic to claims like these goes no further than "it happened for solar and batteries." The fundamental technology behind DACCS is super mature (every part of it has been used in the oil and gas industry for a century). We're close to the thermodynamic limit of efficiency in those processes. If there isn't a specific rationale behind the cost reduction predictions I am very skeptical they'll be realized because DACCS isn't a widget made in a factory. It's a process that's governed pretty directly by thermo. And there is a clearly identifiable headwind to cost reduction in that we'll fill up the most accessible and cheapest reservoirs first.
Great point, and I know that DACCS faces very different inputs and constraints than solar or batteries. It won't necessarily follow the same cost curve. But you could argue solar was considered a mature technology just 10 years ago (the technology was invented decades before) and then proceeded to come down in cost 12x in the last decade. I think it has more to do with iterations than how long a technology has been around. Point-source carbon capture has been around a while, yes. But it has been deployed maybe a few dozen times? Modular DAC machines that could be manufactured as opposed to tailored for individual facilities holds promise to come down the cost curve.
It's a tough thing to predict, but what worries me is that the process itself is already quite close to the thermodynamic limit. It's at like 50%, which is pretty darn good in thermo terms, and means that the highest theoretical cost reduction you could *ever* get on the process is 2x. Maybe there are some reductions on the capital equipment, but that's where I worry that the maturity is the issue. Even if the underlying technology for solar was mature, nobody had ever made lots of panels. The factories making solar cells were *far* from mature. The factories are the thing that scales, so they're what's important. The widgets that go into a DAC system are already made at unbelievably massive scale for oil & gas. Sure, you may get some optimization around this new process. But again, the cost reduction potential seems meager. Layer on top of that the loss of site optimality as you scale up and you could have a super flat looking curve. I genuinely hope I'm wrong on this. But I worry that we too often treat cost curves and other Moore's Law-ish phenomena as laws of nature. The real laws of nature are thermodynamics. Moore's Law works because microchips are hilariously far from the theoretical physical maximum transistor density, just as batteries are hilariously far from the theoretical physical cost minimum. When you're far from theoretical perfection, there's lots of room to learn and that learning can be exponential with scale. But when we're already scraping up against the physical constants that truly do rule our world, we'll quickly find that Moore's Law and its cost curve cousins bend the knee.
Great overview. However, I think nature based removal deserves more buyer support. Lots of innovation underway here but we need to accelerate it. Check out the work Salk Institute is doing to augment the amount of carbon plants can draw down and and store in root systems through genetic interventions.
Great point. They're doing some amazing work. I would put them in a different category than the traditional forestry offsets. Theirs is basically a hybrid nature/engineered solution that can be really impactful.
Thanks for your post. I would love to read more of what's behind your predictions on the cost curve of DACCS coming down. I worry that the underlying logic to claims like these goes no further than "it happened for solar and batteries." The fundamental technology behind DACCS is super mature (every part of it has been used in the oil and gas industry for a century). We're close to the thermodynamic limit of efficiency in those processes. If there isn't a specific rationale behind the cost reduction predictions I am very skeptical they'll be realized because DACCS isn't a widget made in a factory. It's a process that's governed pretty directly by thermo. And there is a clearly identifiable headwind to cost reduction in that we'll fill up the most accessible and cheapest reservoirs first.
Great point, and I know that DACCS faces very different inputs and constraints than solar or batteries. It won't necessarily follow the same cost curve. But you could argue solar was considered a mature technology just 10 years ago (the technology was invented decades before) and then proceeded to come down in cost 12x in the last decade. I think it has more to do with iterations than how long a technology has been around. Point-source carbon capture has been around a while, yes. But it has been deployed maybe a few dozen times? Modular DAC machines that could be manufactured as opposed to tailored for individual facilities holds promise to come down the cost curve.
It's a tough thing to predict, but what worries me is that the process itself is already quite close to the thermodynamic limit. It's at like 50%, which is pretty darn good in thermo terms, and means that the highest theoretical cost reduction you could *ever* get on the process is 2x. Maybe there are some reductions on the capital equipment, but that's where I worry that the maturity is the issue. Even if the underlying technology for solar was mature, nobody had ever made lots of panels. The factories making solar cells were *far* from mature. The factories are the thing that scales, so they're what's important. The widgets that go into a DAC system are already made at unbelievably massive scale for oil & gas. Sure, you may get some optimization around this new process. But again, the cost reduction potential seems meager. Layer on top of that the loss of site optimality as you scale up and you could have a super flat looking curve. I genuinely hope I'm wrong on this. But I worry that we too often treat cost curves and other Moore's Law-ish phenomena as laws of nature. The real laws of nature are thermodynamics. Moore's Law works because microchips are hilariously far from the theoretical physical maximum transistor density, just as batteries are hilariously far from the theoretical physical cost minimum. When you're far from theoretical perfection, there's lots of room to learn and that learning can be exponential with scale. But when we're already scraping up against the physical constants that truly do rule our world, we'll quickly find that Moore's Law and its cost curve cousins bend the knee.
Great overview. However, I think nature based removal deserves more buyer support. Lots of innovation underway here but we need to accelerate it. Check out the work Salk Institute is doing to augment the amount of carbon plants can draw down and and store in root systems through genetic interventions.
Great point. They're doing some amazing work. I would put them in a different category than the traditional forestry offsets. Theirs is basically a hybrid nature/engineered solution that can be really impactful.