Decoding Microsoft's carbon removal strategy

The biggest buyer in carbon removal keeps on buying. What does it know that the rest of the market doesn't?

You’re reading Terraform Now, a newsletter on the business of carbon removal. To support my work, you can subscribe or share this article with interested folks:

Microsoft is single-handedly committing more to carbon removal than anyone else. It has purchased 56% of all high-quality carbon removal credits, and has outspent it’s Cloud and AI rivals, Amazon and Google, by 20x. Much has been made about Frontier, the carbon removal buying consortium used by many tech companies, but Microsoft is even outspending Frontier by an order of magnitude.

About a year ago, I wrote about why competition within the Enterprise Cloud market was having such a huge impact on the carbon removal market. Of relevance today is this passage:

For many years The Cloud was the fastest growing large market in enterprise technology. Now growth is slowing, with customers capping spend. This is also a unique market where the majority of large customers buy cloud instances from both Amazon and Microsoft, actively compare them, and make decisions quarter-to-quarter about where their next dollar should go.

All of this means that the hunt for differentiation is intense. Though it is unlikely to be a deciding factor, Microsoft is probably touting its carbon removal purchases in customer pitches, claiming to be more green than Amazon.

Amazon will need to purchase offsets just to ‘tick the box’ in pitches. Perhaps Google will move next. It will be hard for Amazon and Google to buy anything other than high-quality offsets.

Since I wrote this, Microsoft has decided to massively expand its data centers, as they try to keep up with OpenAI’s compute- and energy-hungry training needs. In the last year Microsoft announced $1B+ data center investments in Australia, Mexico, France, Kenya, the United Kingdom, Indonesia, and is expanding data centers all over the US.

This explains some of Microsoft’s behavior in carbon removal markets. It is building data centers with huge energy needs that will inevitably burn a lot of fossil fuels. So there’s an urgent need for Microsoft to maintain it’s climate leadership. But Microsoft is now so far ahead of its competitors that committing to additional high-quality credits does not give them even a marginal competitive advantage. And committing to credits is not free — Microsoft is on the hook for $2-$4B worth of CDR payments.

Of course Microsoft doesn’t have to spend that money until the credits are delivered, and 80%+ of these credits will probably be delayed beyond 2026. Still, we’re talking about capital that could be spent building out more data centers, hiring more AI engineers, etc.

And yes, Microsoft prints cash like no other business in the world. But you don’t become the most valuable company in the world by spending willy-nilly. There is clear strategic logic behind Microsoft’s splurge, one that suggests they know something about carbon removal markets that others have not seen yet.

Pre-empting the market

Most companies buying carbon removal credits — say Amazon or JP Morgan — are exploring the market, trying to understand it, and finding solutions that work for them. But not Microsoft — it is urgently buying as many credits as it can. Frankly, it looks a bit like hoarding, and when a company hoards the supply of something, it tells you that it thinks prices are going up.

My guess is that Microsoft has an internal model of the carbon removal market, and that model suggests that high-quality credits will be scarce in the second half of the 2020s. Scarcity will lead to price increases.

It must be hard for market newcomers and observers to imagine carbon removal prices going any higher. Biochar has risen from ~$100 to ~$250 per ton in the last year. Direct Air Capture (DAC) is typically 3x the price of Biochar.

But Microsoft is an extremely close observer of this market. Now that they have a 500,000 ton commitment to Oxy’ Stratos plant, Microsoft climate folks can go out to see for themselves the incredible complexity involved in buildling a large DAC plant, and appreciate why this project will probably be delayed. And after buying biochar credits from about a dozen suppliers, Microsoft will have first-hand experience with the logistical challenges of scaling Biochar.

Clearly Microsoft is hedging its bets across a number of technologies, including biochar, mineralization, and bioenergy with storage. But they have been laser focused on targeting supply from the first scaled DAC plants. Microsoft now has agreements with all the players that are building 500,000+ ton DAC plants in America — Climeworks, Oxy / 1PointFive, and Heirloom.

This makes sense. DAC is likely to be the most supply-constrained market for the next 5-10 years, but is also the most bulletproof from a quality perspective. And a guarantee of quality is so important. The winds have shifted in carbon removal in the past, and legitimate accusations of ‘greenwashing’ for one supplier of, say, afforestation, effects all suppliers for afforestation.

If the winds shift in the carbon removal market, and technologies like biochar, mineralization, and bioenergy with storage are credibly accused of ‘greenwashing,’ DAC will be fine. DAC production is just too hard to fake, and too easy to measure. And that’s why Microsoft is cornering the DAC market faster than any other.

So, Microsoft is moving early to lock in high quality credits at what they view as ‘low’ prices today. Microsoft modellers must believe that there will be many more new buyers in the market, that existing buyers will ramp up their purchases once they see technologies like DAC scaling, or that some suppliers will fail to deliver promised credits.

Creating the future you want

One interesting side effect of Microsoft buying so much is that the company is creating the scarcity that it is predicting. This might even be intentional. Being the biggest buyer gives you market power in the future, which means that not only will Microsoft lock in low prices today, but should be first in line for sweetheart deals in the future.

We’ll see if Microsoft’s prediction is right. The optimist in me wants carbon removal to scale very fast, for prices to fall, and for many companies to become regular buyers, and for the deliveries to grow at exponential rates. But the realist in me recognizes that Microsoft is probably right — the obstacles to scaling here are very real, and we are in the early innings.

---

Special thanks to cdr.fyi for their excellent data set on CDR purchases, which I used throughout this article, and to CIO Dive, which published a detailed article on Microsoft’s data center expansion plans.