The Immense Scale of Carbon Capture

6 min readFeb 12, 2021


Just this week, Elon Musk (the genius, visionary, mogul, troll, or provocateur, depending on who is describing him, the day of week, and what he just tweeted) offered up $100 million in prizes for the best carbon capture technology. The prize notes that alongside vastly reducing emissions in the near future, we will likely need to drastically reduce current atmospheric CO2 levels using carbon capture and sequestration technology. The prize indicates no preferences for the approach used; a plan to plant immense numbers of trees could win just as a high-tech solution could. This prize does not value theoretical solutions, it requires real hardware proven at the 1 ton/day scale, with plans to economically ramp up to a 10 gigaton/year scale by 2050. They also note that the systems will be compared based on their “fully considered cost per ton,” meaning they will include the environmental impacts of the systems themselves (i.e. manufacturing, transportation, installation, and operation.)

Great. Fine. But what is 1 gigaton? Just how large must these systems scale? Why don’t we just start planting trees? I thought that too. While carbon capture technology is outside my scope of expertise, as an engineer I can at least do high-school algebra, so I started doing some rough scaling calculations. This story will not discuss possible technological solutions, but rather focuses on the scale of the challenge, and the need for novel, outside-the-box thinking about the problem.

The “giga” prefix indicates 1 billion. In a world with $1 trillion+ government spending bills, and Google Chrome tabs digging deep into your gigabytes of RAM, it is easy to misunderstand just how large these numbers are. There are roughly 7.8 billion humans (7.8 gigapeople?) alive right now, with a world average weight of 136.7 lbs. This means that all of the humans alive right now combined only weigh 0.54 gigatons! While estimates vary, it is likely that around 35 billion humans have lived since 1200 AD. With humans being roughly 18% carbon, this challenge aims to sequester an amount of carbon per year by 2050 equal to 23X the total combined carbon in every person who has lived in the last 800 years! Truly, this is a mind boggling amount.

Keeping in mind the astronomical scale of this challenge, let’s consider some of the commonly discussed approaches to it.

The World’s Biggest Vacuum

Direct air carbon capture, such as the process developed by Carbon Engineering, pulls CO2 directly from atmospheric air. It is an interesting concept. Such systems could be located anywhere in the world, presumably in lower-cost, remote locations for economical reasons. Unfortunately, they do have some major challenges as well. The CO2 portion in the air, while very high compared to historical levels and significant for the climate, is quite small. In 2019, the average was 408.9 parts-per-million. No carbon capture technology is expected to be 100% efficient, nor would it be good if it were. A massive scale, fully efficient carbon capture system could be bad for the nearby environment by reducing the local CO2 levels too much. For argument’s sake, let’s consider a system which can remove 5% of the CO2 in any air that passes through it. To extract 1 gigaton of CO2, this system would need to process 4.4 quadrillion kg of air. If we assume air pressure at sea level and room temperature, this is roughly 3.6 quadrillion cubic meters of air. If we consider all of the air up to 1 km (3,280 ft.), this volume would cover 14 million square miles. The entirety of North America covers only 9.355 million square miles, so we would need to process significantly more air than covers all of North America up to 3,280 ft. high! Hitting the eventual goal of 10 gigatons per year by 2050 is even harder. The concentration will increase by then, which modifies this calculation, but at current levels this would require processing all of the air covering 71% of the Earth up to 3,280 ft. every year.

But, you might say, why would we extract the CO2 from air once it has already mixed? Why not just capture it at the source, for example at natural gas or coal fueled power plants or during concrete or steel production? Many companies and research groups indeed have focused on that. The concentrations of CO2 are much higher, so the processing can be more effective. However, I would argue that capturing new CO2 at the source is not the intention of this challenge, but rather it is designed to inspire technologies to reduce the CO2 levels we already have.

Just Plant Trees, Dummy

Perhaps the simplest approach to this problem is to use nature. Trees and other plant life are Earth’s natural carbon sequestration technology. How hard can it be to plant some trees and walk away? Surely they will grow, processing CO2 and producing oxygen. Many scientists and environmentalists have also pushed for this. Not only does it sequester carbon, reforestation does not pollute and has a variety of other benefits (biodiversity, erosion resistance, drought resistance, etc.) if done properly. The process is truly solar powered. In the words of Jeremy Clarkson, “how hard could it be?”

The main factor in this equation is how much carbon can be sequestered by reforestation by a certain area over a certain time. There is no one answer for this, as it varies based on the climate and what is being grown. One congressional report puts this value at 1.1–7.7 metric tons of CO2 per acre per year. I will use the upper bound in the analysis, for the most optimistic case. This is a simple multiplication problem. If we want to sequester 10 gigatons of carbon in 1 year (the stated 2050 goal), we would need to have an area of 1.6 million square miles actively growing at the high end of the EPA’s rate. This is equal to 48% of the land area of the United States, and over 11X the size of California.

We do have experience sequestering very large amounts of carbon every year, at least temporarily. In 2020, the US produced roughly 0.4 gigatons of corn. In 2016, the US produced 0.13 gigatons of soybeans. Crop production is a highly optimized, mechanized process, using massive parts of the accessible land across the US. This is well below the lower limit of the scale that reforestation would need to reach.

It seems a stupid question to ask, but is planting trees even sustainable? I don’t mean TV-commercial-selling-soap-sustainable. Humans have done a large amount of deforestation, and some of the CO2 in the atmosphere is from burning biomass. But today, much has come from carbon that was previously locked away as coal, petroleum, or natural gas. Even if we fully reforest (which seems very unlikely), we will only have reverted the CO2 from biomass combustion. The net CO2 from coal, petroleum, and natural gas, is not changed. We will not have pumped oil back into the ground nor will we have unburnt any coal. It seems we will need the technological methods as well.

What About The Ocean?

So far, we have focused on carbon sequestration from the air. The prize also covers technologies or approaches to pull carbon from the oceans. The oceans absorb carbon from the atmosphere to the tune of 34 gigatons of CO2 between 1994–2007. That stat alone shows the immense scale of the task — absorbing more CO2 per year than all of the world’s oceans do.

Closing Thoughts

If nothing else, I hope this story is not discouraging. I hope that thinking about the scale brings out new, fresh ideas. Elon Musk is injecting a massive amount of money into a critically important field. He has a long history of leading real progress on ideas that many think are crazy at the time. Landing a rocket on an autonomous boat is stupid, until it works. One company launching more of its own satellites than the total number currently in orbit is laughable, until they just start doing it. Directly interfacing with the brain is not possible, until it works. Electric vehicles are not practical or profitable, until they are. Digging tunnels is way too expensive and slow, until it’s not. If you asked Bell Labs engineers to make giga-transistors when they were building them by hand in 1947, they would probably start crying. Today, a single Cerebras’ AI chip has 1,200 giga-transistors. I am hoping that the same story will be told of carbon capture technology.