How to start an advance market commitment

A practical guide from the founders of Frontier

May 31, 2024
  • Support The Works in Progress Newsletter

    Since you liked this post, why not subscribe?

Don’t miss Issue 15, which we released two weeks ago. We have had hundreds of applications for Invisible College, our week-long residential seminar in Cambridge in August. Applications close at the end of today if you want to add yours, or know someone who does.

Here is a sneak preview of Issue 16: an article by Stripe Climate’s Nan Ransohoff. You can read it on our website here. The topic is so timely that we decided to release it right away.

In 2007, a group of governments and philanthropists pledged $1.5 billion to accelerate the development of vaccines against Streptococcus pneumoniae infections – pneumococcal vaccines – to prevent one of the most common causes of death in children in poor countries.

They did this with an advance market commitment. Advance market commitments, or AMCs, are promises to buy or subsidize something in the future, if someone can invent and produce it. Their purpose is to guarantee enough future demand (in this case nearly two billion dollars) to encourage suppliers (in this case pharmaceutical companies) to try to build something that should exist, but doesn’t (in this case a pneumococcal vaccine for people in poor countries).

There have been two notable AMCs for vaccines.

That first AMC was launched by Gavi, a public-private partnership largely funded by the UK government and the Bill and Melinda Gates Foundation. And it worked. By 2011, Pfizer and GlaxoSmithKline had developed a vaccine that met the target specifications, and both had signed contracts to start producing the vaccine at large volumes. By 2020, the Serum Institute of India had as well. Hundreds of millions of vaccine doses were purchased and distributed throughout the world, accelerating the development and distribution of the vaccine by five years, saving an estimated 700,000 lives.

A second AMC, launched as part of Operation Warp Speed in 2020, comprised US-guaranteed purchase orders for 900 million vaccine doses against Covid-19. The concern from pharmaceutical companies was: what if we spend time and money developing a Covid-19 vaccine, but someone else gets there first and the market for ours disappears? With these guarantees, even if Covid-19 had been addressed by the time some of these vaccines were authorized, the vaccines would still be purchased. The companies could invest in vaccines they knew their competitors were developing too, safe in the knowledge that their investments would not be totally lost if they weren’t first.

In 2022, I helped launch another AMC, Frontier, modeled after the vaccine AMCs. This time, the AMC was to pay carbon removal companies $925 million to permanently remove carbon from the air. While carbon removal and vaccines do not have identical market dynamics, they face a similar challenge: uncertainty about whether there will be customers who will pay for a product if it can be developed.

AMCs can send a strong and immediate signal that there is a market for a product, and do so without picking winning technologies at the start. They might help us solve many other important problems. And yet there have only been a handful to date. Fortunately, that’s starting to change. In May 2023, the University of Chicago announced a Market Shaping Accelerator with a focus on the global challenges of climate change, biosecurity, and pandemic preparedness. In July 2023, the US government announced that it would consider demand-side interventions, such as AMCs, to accelerate the development of hydrogen fuels, which can be used for transportation, electricity generation, and energy storage.

There could be more still. This piece is written for those starting or considering starting an AMC. It draws on my experience from Frontier, and attempts to describe what it was like for us to go from theory to practice. This piece tries to abstract a generalizable approach that may be helpful to others, without glossing over the messiness of reality and the imperfectness of our solution.

My goals for this piece are twofold:

1. To make highly motivated people working on important global problems aware of another potential solution available to them.

2. To help individuals who already want to start an AMC (or are seriously considering it) save time in doing so.

To help make this piece even more tactical, attached is a worksheet. If you fill it out and send it to my colleagues and me at Frontier, we’ll read it and help if we can.


Tools for market shaping

Most innovation is the result of natural market forces, meaning no special intervention is required. But there are many cases where innovation ‘should’ happen in the sense that it would result in a big social benefit if it did, but doesn’t because the would-be innovators don’t have commercial incentives to act. Neglected diseases, climate change, and pandemic preparedness are all fields where this type of market failure frequently occurs.

Market-shaping mechanisms – of which AMCs are one – are activities and tools designed to address this kind of challenge. For a problem to be a good fit for market-shaping, the commercial incentives for developing a solution need to be insufficient for a solution to either be invented or reach the scale required to meet the social need.

Let’s take carbon removal and the pneumococcal vaccine as examples.

In the case of carbon removal, the societal benefits of inventing and scaling ways to remove billions of tons of carbon from the atmosphere affordably are extremely high. The most reliable climate models say the world needs to permanently remove gigatons of carbon from the atmosphere every year by 2050 to avoid the most catastrophic effects of climate change. And yet, to date, less than ten thousand tons of CO2 have been permanently removed from the atmosphere – one million times short of the scale needed every year.

The problem comes down to commercial incentives. A market for carbon removal doesn’t exist because the end buyer doesn’t get direct value out of it, unlike, say, paying for energy. Why would an entrepreneur start a company if they won’t have customers? Why would an investor invest in a company with no path to continued revenue?

Similarly, a pneumococcal vaccine for low-income countries had the potential to save hundreds of thousands of lives. But pharmaceutical companies weren’t motivated to develop and produce it at large scales because the end customers (governments of poor countries) could only afford ultra-low prices.

The pneumococcal case had one other interesting nuance. At the time of designing the AMC, a version of the vaccine already existed, but it didn’t include the bacterial strains that affected people in poor countries. It’s possible that the pharma companies would have eventually gotten around to adding the extra strains and scaling the vaccine up, but it would have happened much more slowly, and many people would have died in the meantime. Delay is a legitimate type of market failure that AMCs can be designed to shorten.

What are some of the available tools to solve these kinds of market failures?

Market-shaping tools are often called demand-pull mechanisms. Whereas supply-push mechanisms reward inputs and activities (like a research grant from the National Institutes of Health that funds a scientist to pursue a particular project, regardless of whether it results in a scalable product or idea), demand-pull mechanisms reward outputs and outcomes (as when the Board of Longitude in Britain awarded prizes after 1714 for those who could calculate longitude at sea to within one degree). Demand-pull mechanisms ‘pull’ innovation by providing financial incentives for valuable new inventions. A big advantage of market-shaping tools is that they provide incentives for the private sector to innovate and ultimately scale, but they avoid the risk of funding projects that don’t end up succeeding.

An AMC is one type of market-shaping tool. It’s a promise to buy or subsidize a product or service – usually at a certain price and quantity – in the future, if it’s successfully invented. Fundamentally an AMC says, ‘If you develop a product that looks like this predefined spec, we promise to buy it.’

The rest of this piece is a step-by-step guide for starting an AMC. Each step draws heavily on my colleagues’ and my experiences starting Frontier. The steps are intended to help make this process less abstract and more tactical. But bear in mind that starting something, AMCs included, doesn’t actually happen in linear steps. You are encouraged to ignore, add, or significantly edit steps as you deem fit for your use case. Use these steps as a rough guide, but rely first and foremost on your judgment.


Step 1: Determine if an AMC is a good fit for the problem

We can use three criteria to help determine if a problem is a good fit for an AMC.

Criterion 1: Those who could develop a solution are blocked by uncertainty around whether there will be a buyer and/or if that buyer can pay enough.

AMCs are well-suited to problems that stem from demand uncertainty. Put yourself in the shoes of a potential inventor. If they’re asking themselves questions like, ‘If I’m able to invent a product that solves this problem, is anyone going to pay for it? And will they be able to pay me enough for me to recoup my costs?’, then an AMC might be a good fit.

Let’s take carbon removal. Despite the high societal benefits of carbon removal, very few people were working to develop carbon removal approaches until quite recently. This is because, as we’ve established, there are no ‘natural’ customers. There are lots of theoretical customers for carbon removal: almost every company and government with a net-zero target will need to remove carbon to meet it at some point. But new technologies – like solar panels, hard drives, TVs – tend to be expensive at the start; costs only come down over time as they scale. Carbon removal is at the beginning of a similar trajectory. And because there are cheaper options that satisfy climate commitments as they are written today, companies aren’t rewarded for buying carbon removal. Without customers, carbon removal companies can’t scale to reduce their costs. Carbon removal has been stuck.

In the case of the pneumococcal vaccine, governments in affected countries were willing to pay per dose to provide vaccines to poor people in their countries, but not enough to make it compelling to the pharma companies. As a result, pharmaceutical companies didn’t have enough commercial incentive to prioritize working on a vaccine.

Both carbon removal and the pneumococcal vaccine meet this criterion.

Criterion 2: We can (at least partly) describe the end product that would solve the problem, but we don’t know exactly who is best positioned to invent it or how.

A big advantage of an AMC is that it can credibly signal to innovators that there is real demand for a product without the AMC designers having to pick a winning technology at the start. As long as we can define the important elements of the end product we’d want to buy, we can be agnostic as to who will develop it and how. This makes AMCs a particularly useful tool in cases where we aren’t sure who is best positioned to invent something or how they’d do it.

If we do know who the likely inventor is, and exactly what it is we want them to invent, then it’s much simpler to write individual contracts directly with that supplier rather than go to the effort of setting up an AMC.

In the case of carbon removal, we wrote our target product profile to outline what was missing in solutions that we saw today. We wanted to pay for carbon to be permanently removed from the air. We wanted those tons to be out of the atmosphere for more than 1,000 years, which we call permanence. And, very importantly, we wanted the technologies to have a path to being low-cost and high volume in the future, which we defined as costing less than $100 per ton and removing more than half a gigaton annually.

But while we knew exactly what we wanted, we didn’t know which carbon removal approaches would work or who would develop them or how. An AMC allowed us to invite anyone to try to come up with a solution.

The pneumococcal vaccine end product was also clear – doses of a vaccine that met a dozen or so requirements, such as being effective and safe in children under five, as well as a cost target. While it was clear that pharmaceutical companies would be the inventors, it wasn’t clear which one(s) would actually invent it or how.

Both carbon removal and the pneumococcal vaccine meet this criterion.

Criterion 3: Even after a product gets invented, further incentive is needed to bridge the gap to a long-term market.

AMCs are particularly good at helping bridge the gap between the world today and a world where the long-term market for that product can take over. If all we need is the thing to be invented, after which existing market forces will help it scale, then a prize or a patent may be a better and simpler choice than a standard AMC. But if further incentives are needed even after the product is invented to get the solution to the point where the long-term market for that product can take over, AMCs are a good fit.

Carbon removal only sort of meets this criterion. It’s certainly true that even after a particular approach gets invented, it will still need to get much cheaper for there to conceivably be long-term customer demand. The open question is, ‘What is the long-term market for carbon removal?’

There is a theoretical long-term market for carbon removal because scientists say we need it at huge volume. But in practice, no such market exists today and it’s unclear when one will materialize and at what scale. For carbon removal to reach climate-relevant scale, governments around the world will need to create a long-term market for it, either through direct procurement or through policies that force private entities to buy carbon. Without that, Frontier risks being a bridge to nowhere. For this reason, Frontier is quite a bit riskier than a textbook AMC. (This is also why we spend a lot of our time advocating for policies to catalyze that long-term market.)

In the case of the pneumococcal vaccine, there was long-term demand but the end buyers – poorer countries – weren’t able to pay a high enough price for the product at that moment, and would only be able to pay for it in the future if it was extremely cheap. If not, permanent subsidy would be needed. The pneumococcal AMC was designed to incentivize enough scale that eventually the marginal cost of producing another dose would become equal to or less than poor countries’ willingness to pay per dose. In other words, once the AMC ran out, the pharma companies should still have had an interest in supplying the vaccine.

Carbon removal aspires to meet this criterion, but doesn’t fully today. The pneumococcal vaccine does.

Step 2: Define the end product you want to exist

Now that we’ve determined that an AMC is an appropriate tool for the problem, the next step is to more crisply define the end product that ought to be invented and scaled. What would have to be true about a vaccine dose or a ton of carbon removal in order for us to buy it in large quantities?

AMC economists call the desired end product the target product profile. There are two considerations to take into account when defining a target product profile.

Consideration 1: If achieved, the target product profile should appeal to the long-term market of buyers when the AMC is over.

An AMC helps bridge the gap between today and the time when the market can take over, so it’s important to make sure the end product you’ve defined is something the long-term market actually wants. In the case of carbon removal, Frontier did this by defining what we believe is important to long-term buyers: that it is permanent, low-cost, measurable, takes advantage of carbon sinks less constrained by arable land, and is net negative when considering all of the activities that go into the removal.

Consideration 2: The target should describe what great looks like while being solution agnostic.

Frontier cares that the technology permanently removes carbon from the atmosphere and has a path to being low-cost and high volume. But we are open to that being accomplished in many ways, whether through giant fans (direct air capture), rocks (enhanced weathering), or by turning waste biomass (like corn stover) into bio-oil and injecting that underground.

Source: Stripe Climate.

Similarly, the pneumococcal vaccine target profile asked for a vaccine that met standards of safety and effectiveness against strains of the bacteria that affected people in poorer countries. But it was open to the type of vaccine that could be used – protein vaccines, inactivated vaccines, and recombinant vaccines were all eligible.

Here are two examples of actual target product profiles: the first is Frontier’s target product profile for carbon removal, the second is the target product profile for the pneumococcal vaccine.

Example 1: Carbon removal
Example 2: Pneumococcal vaccine (full TPP here)

Step 3: Figure out who the potential suppliers might be and what would motivate them

Now that we’ve defined the thing we want to exist, who is going to invent it?

While we won’t know exactly who will invent it, identifying the types of organizations that are well-suited to work on the problem – and what’s prevented them from doing so to date – is a critical input into the AMC’s design. If you don’t work out who your potential suppliers are, then you might fail to tailor the AMC’s structure to their needs.

Ask yourself:

  • Is there anyone already working on this today? Is there anyone who is particularly well suited to work on this but isn’t yet?

  • What’s preventing them from working on this? What would have to be true for them to work on this?

The most common failure mode of this step is being too abstract. This is the time to go from 10,000 feet to 10 feet. Doing this well means meeting dozens of specific individuals in each category so that you’re fairly certain you understand why they haven’t done something to date, and what it would take for that to change.

In the case of the pneumococcal vaccine, the AMC creators considered a short list of large pharmaceutical companies, such as GlaxoSmithKline, Pfizer, and the Serum Institute of India to be their potential suppliers. As discussed above, these companies had been hesitant to work on this vaccine because they were unsure whether buyers could pay enough per dose to recoup vaccine development costs. Before spending resources working on the vaccine, these companies needed to be sure that they would be able to recoup those costs. For them to then be motivated to continue to scale the vaccine indefinitely, they needed to be sure that the price per dose was at least high enough to cover the marginal cost per dose.

In the case of carbon removal, Frontier had no ready-made industry players to interrogate. We were trying to kick-start a market that didn’t exist yet, and didn’t know which approaches would end up working best. We spoke to a lot of start-ups in the field, and drew some conclusions about what they would need for the AMC to work:

  • Prospective founders needed confidence that a large market would materialize so that it would be worth their time to try to start a company in carbon removal.

  • Existing founders needed us to meet them where they were in terms of price, which is to say, very early and expensive with the potential to be cheap in the future, to help them get down the cost curve. They also needed firmer demand agreements to help them secure the investment (in terms of equity and debt) they’d need to build and scale up.

At the end of 2023 we sent an anonymous survey to our latest cohort of prepurchase companies, and 71 percent of founders said Frontier’s launch played a role in their decision to start a carbon removal company, while 94 percent said Frontier had accelerated their company’s trajectory. These are early indicators that Frontier has had some success in meeting these needs.


Step 4: Estimate roughly how much the AMC will cost

We now have an understanding of who the potential suppliers could be and what it would take to motivate them to work on this. The goal of this step is to use that knowledge to get a rough sense of the size of the AMC: how much money to raise.

The first question is, what size should the AMC be? An AMC is intended to incentivize firms to bring an innovative product to market at a price where there will be real demand. Therefore it should aim to cover the costs that firms would incur between today and the point at which the marginal cost of the product is equal to or less than the long-term willingness to pay. (Additionally, the AMC should be less than the social value of the innovation: we shouldn’t pay more than the innovation is worth.)

One way to visualize this is with the pink section of the chart below, which shows a hypothetical cost curve for something, with the per-unit cost falling as production ramps up and as new technologies and practices get adopted that reduce costs. This area under the curve represents the incremental amount needed to cover the costs that wouldn’t be covered based on the market today.

Based on this analysis you’ll likely land on one of two scenarios.

Scenario 1: The problem is of a scale that can be wholly solved by a single AMC. In the case of the pneumococcal vaccine, a group of expert economists calculated that $1.5 billion, on top of the per-dose co-pay that receiving countries were willing to pay, would be enough in subsidy to motivate pharmaceutical companies to work on this. In Step 6 we’ll get into how they decided to distribute the $1.5 billion to maximize societal outcomes. But for now think about this as enough to cover the costs incurred – R&D, fixed costs, and so on – up until the point where the marginal cost of another dose was equal to or less than the willingness to pay of the recipient countries. In short, $1.5 billion (on top of the per-dose co-pay) was enough to bridge the gap to the point where it would have a long-term market.

Scenario 2: The problem is of a scale where a single AMC can only kick-start it, or provide a stepping stone to a self-sustaining market. It’s very possible that the answer to the previous question yields a number so big that it’s clearly not possible to raise this amount all at once. This was the scenario Frontier was in with carbon removal.

The problem was twofold. First, as we’ve seen, a large, long-term market for permanent carbon removal didn’t exist. Significant policy work would be needed to create that, and that would take time.

Second, even if a large market for carbon removal at, say, $100 per ton did emerge, carbon removal technologies were still early and expensive. Our (extremely rough) estimate suggested that it’d likely cost in the tens of billions of dollars to get a portfolio of carbon removal solutions – that could collectively stack up to gigatons per year – down to under $100 per ton.

So the question for us was: what’s the minimum amount that would motivate suppliers to start building?

Coming up with this number, in our case, was more art than science. At the time we launched Frontier in 2022, around $30 million total had been spent by anyone on carbon removal to date. We determined that a one billion–dollar commitment, equivalent to a 30-times increase in demand, would compel enough suppliers to start building. In parallel, it would also buy time for other corporations and governments to get involved. It was a ‘build the plane while flying it’ strategy.

We were well aware just how imperfect our one billion–dollar number was. In fact, we fiercely debated whether it was worth doing at all. We knew it was possible that our AMC might not be sufficient to give anyone enough confidence to build because even with one billion dollars, entrepreneurs and investors still have to take a leap of faith that a bigger market will materialize. But we decided our billion-dollar starting point was still worth doing, on the grounds that it would buy time for governments to create a large, lasting market for carbon removal either by buying it directly or requiring the private sector to do so.

The second question we asked ourselves was whether we could actually raise one billion dollars. It’s a lot of money. But Stripe, which helps to run Frontier, had already pledged some of its own revenue to support carbon removal and had also set up Stripe Climate, which lets companies that use Stripe direct a fraction of their revenue to carbon removal. Even before the AMC, Stripe and Stripe Climate users were on track between 2022 and 2030 to commit a meaningful chunk of the billion dollars we thought we needed. Raising the rest would be hard, but not impossible.

In the end, we actually didn’t quite get to one billion dollars at launch: we got to $925 million. We debated waiting until we got to the full one billion dollars but decided the cost of waiting longer to launch wasn’t worth it.

Step 5: Figure out who the potential buyers are and what would motivate them to fund this

Now that we’ve roughly quantified how much money is needed to incentivize prospective suppliers, we now need to figure out where that money is going to come from.

Buyers pay for the product. Without someone to buy or subsidize the product, there is no AMC. But identifying the right buyer is challenging because the whole reason an advance market commitment is needed in the first place is the lack of natural customers at this stage. If you’ve identified an AMC as the solution, a central part of your job is to create enough demand to fund it.

There are a few different paths here, as illustrated by carbon removal and the pneumococcal vaccine.

Path 1: Try to find a buyer (or collection of buyers) who can be convinced to pay higher prices while the product is still early.

This is the path we took with Frontier.

In standing up Frontier, we spoke with nearly 100 companies to see if we could convince them to be buyers in an AMC for carbon removal. Most weren’t interested at the time. But we didn’t need everyone to be on board, we just needed to get to one billion dollars.

Those who were interested had a long-term interest in carbon removal scaling and getting cheap. Frontier’s founding buyers are companies that have already made some version of a net-zero climate commitment, of which the basic tenets are: measure your emissions, reduce as much as you can, remove the rest. The ‘remove the rest’ part is squishy for most companies (as it is for most of the world) in part because carbon removal is so nascent. For these buyers, a carbon removal AMC was an opportunity to help create the supply that they will need to meet their existing climate commitments, many of which require significant volumes of carbon removal after 2030.

But this alone isn’t enough to explain why the founders got involved. Climate commitments are, at this stage, mostly voluntary, and compared to other ways companies can technically check the box, Frontier is still very expensive. Our founding buyers had a few other unique traits.

First, they had a track record of approaching climate with a long-termist mindset. Google, for example, had done pioneering work decades earlier paying a premium for renewable energy suppliers to help them get down the cost curve. Our buyers saw both the good for the world in accelerating promising climate technologies down the cost curve, as well as the possibility of brand benefits.

Second, they were able to fund carbon removal at high prices so that the rest of the world can benefit later.

Finally, each had an insider who believed in this idea. These insiders became our champions who successfully advocated for an atypical opportunity for any organization. Finding such an insider takes time but is absolutely critical. An insider can build buy-in in a way an outsider simply can’t.

After spending significant time with prospective buyers (including Stripe itself), we had a rough set of requirements to design around.

Path 2: You can’t find a buyer at a higher price, in which case you may need to add another type of funder.

Buyers for the pneumococcal vaccine looked a bit different. Because the countries that needed the vaccine weren’t able to pay higher prices, someone else needed to fund the gap. In the case of the pneumococcal AMC, this was a combination of philanthropists (the Bill and Melinda Gates Foundation, which put in $50 million) and governments (the UK, Italy, Canada, Russia, and Norway, which put in the remaining $1.45 billion).

A final observation: finding an anchor buyer is an important step in building momentum. If you’re a builder, do whatever you can to find even just one buyer early on: it can dramatically accelerate your efforts – ideally that buyer would be able to help get peer organizations on board. If you’re a philanthropist or a company, consider playing that role yourself to help get a promising new idea off the ground.

Step 6: Come up with a design that works for both suppliers and buyers

At this point you’ve done a lot of work to understand both the problem and the requirements a good solution would need to embody. The next step is to design the AMC itself. What design will (1) motivate potential suppliers to act, (2) be attractive to (and work within the financial constraints of) buyers, and (3) help solve the overall market failure?

There is no one-size-fits-all design. I’ll give a little detail on the pneumococcal AMC design, much more detail on the mechanics of Frontier, and then offer some overall observations.

For the pneumococcal AMC, Gavi first worked with beneficiary countries to approximate vaccine demand, estimating that $1.5 billion would be sufficient to subsidize up to 200 million doses a year for ten years.

Gavi then established a long-term price cap of $3.50, set at the estimated long-term marginal cost of production. The $1.5 billion was then used to pay suppliers an initial subsidy over and above that $3.50 price cap. Gavi agreed to front-load the subsidy to help suppliers recoup their development costs more quickly, in exchange for suppliers committing to adhere to a long-term price cap even after AMC funds were exhausted. Specifically, each supplier could get paid up to seven dollars total per dose for the first 20 percent of doses.

Frontier is a commitment to buy at least one billion dollars of permanent carbon removal between 2022 and 2030. Its founding buyers are Stripe, Alphabet, Shopify, Meta, McKinsey, and tens of thousands of businesses using Stripe Climate. Here’s the gist of how it works:

1. Buyers decide how much they want to spend on carbon removal and pool their commitments through Frontier. When buyers join Frontier, they sign an agreement that includes both their total commitment amount as well as an annual schedule for how much of the commitment amount can be deployed each year. For example, a $25 million commitment might look like this:

Defining an annual demand schedule gives buyers financial predictability. It’s also an important contracting constraint for the Frontier team.

Critically, there isn’t a specific number of tons or price-per-ton target. The funds are flexible. This allows us to make purchases we believe will be most catalytic for the field rather than optimize around price in the short term.

2. Frontier’s team of technical and commercial experts find and vet the most promising suppliers, and then facilitate purchases between buyers and suppliers. Frontier purchases carbon removal in two ways. The first is through prepurchases. This track is targeted toward promising early-stage companies that are just getting started. Prepurchases are low-volume contracts (typically $500,000) where we pay the company up front, before tons have actually been delivered.

But the bulk of Frontier’s spend will be spent via our offtake agreements. Offtakes are legally binding contracts to buy a specific amount of carbon removal at an agreed price if and when tons are delivered. Offtakes have a few important properties. First, they help derisk purchases for buyers because they’re pay-upon-delivery: if the supplier doesn’t deliver, buyers aren’t on the hook. Second, they help unlock project finance from lenders. When a carbon removal supplier goes to a bank to ask them for capital to build an expensive new carbon removal facility, one of the first questions a bank will ask is, ‘Why would I give you money to build this? Is anyone going to buy the tons produced by it?’ An offtake gives a lender hard evidence that there will in fact be a buyer, helping to derisk the project for the bank.

Trying to assess which carbon removal companies to offtake with takes up the lion’s share of our team’s time. For each deal, Frontier provides buyers with a purchase memo, a pre-negotiated contract, and how much money the buyer is willing to pay each year. Buyers then sign identical, pre-negotiated carbon removal offtake agreements directly with each supplier (i.e., Frontier does not hold funds in escrow on behalf of buyers).

A standard AMC offers contracts for the final product. Frontier needed to offer contracts for an interim product – we’re buying tons from an approach that has the potential to be low-cost and high volume in the future, but today is expensive and small-scale (because we couldn’t raise enough money to get them all the way down the cost curve). As a result, we have to carefully review each company to determine if they have the potential to be very large and very low-cost in the future – even if they’re not there today.

In this sense, Frontier is almost more of a compilation of milestone contracts for interim, not finished, products. When we were initially designing Frontier back in 2021, we had biweekly working sessions with some of the economists who had invented and developed the concept of advanced market commitments: Rachel Glennerster, Chris Snyder, and Susan Athey. At one point, one asked, ‘Is this even an AMC?’, given we were buying carbon removal on the path to being low-cost, rather than just the ‘final product’ of carbon removal at under $100 per ton. (We determined it was enough of an AMC.)

3. When carbon gets removed and verified, suppliers are paid and tons are issued back to buyers. Frontier buyers can choose to count these toward their company’s climate commitments. Buyers pay an annual fee to Frontier to cover costs.

A few things to consider as you’re designing your solution.

First, don’t worry if the solution doesn’t perfectly match the theory. Your job is to make sure whatever you design helps solve the overall problem as well as the specific needs of suppliers and buyers, regardless of whether that matches the theory.

Second, be careful when copying and pasting from past AMCs. The differences between Frontier and Gavi are significant. It’s very likely that in your case big edits will also be necessary.

Third, get fast feedback from buyers and suppliers and alter your plans accordingly. Throughout the design of Frontier, we were on a texting basis with roughly ten potential buyers and five suppliers. The more tactical and specific we made Frontier, the better feedback we got. For example, early on we created a scenario of how a $100 million commitment might play out over eight years from the buyer’s perspective: in what year would the funds be spent, the range of what tons might cost in different years, etc. Giving potential buyers something specific to react to usually elicits more actionable feedback than discussing the idea at a high level.

Step 7: Launch

Ultimately a demand signal – like promising to buy thousands of tons of carbon out of the air – only matters if the right ears hear it. The launch strategy of an AMC is as central to its success as the design.

In general, an appropriate go-to-market plan largely depends on two key questions. First, who needs to hear the demand signal? Second, how can they best be reached?

Because our audience of potential founders is very diffuse (great prospective founders could be anywhere!), our go-to-market strategy felt almost like launching a consumer product. In contrast, Gavi needed to reach a finite number of pharmaceutical companies and likely thought about its launch strategy quite differently. In short, the ‘right’ launch strategy depends on the target audience(s) and how to reach them.

What we did was straightforward, but effective:

  • Named the initiative ‘Frontier’, came up with a simple brand for it, and built an initially single-page website and video for launch.

  • Hosted a 20-person private briefing the day before launch with influential academics, policymakers, and journalists to help spread the word to their large audiences.

  • Continued to host regular briefings on both the supplier side (top universities, research labs, and start-up accelerators) and the buyer side (top VCs and banks).

More than a year into Frontier we are still working on getting the word out. Last year we realized that our demand signal wasn’t heard as well outside of the US, and we’re working to change that in order to reach the many potential carbon removal founders globally.

Step 8: Operate

It may be tempting to feel like you’re done at this point – after all, the money has been raised and announced. But, at least for us at Frontier, this is where the work begins. Now you have to run it.

The shape of this step will depend on the design decisions previously made. It will also depend on where you are in the journey (the shape of our work looks different today than it did a year ago, as does the team). It will also depend on how developed the market is – in the case of carbon removal, the entire industry was incredibly nascent, so we spend time not just on the tech development itself, but the market foundations as well (e.g., what do the contracts look like, how do you accept delivery, and so on). In the spirit of making this more concrete, here are a few examples of how we currently spend our time.

  • Sourcing and doing due diligence on suppliers for offtakes. Frontier’s success hinges on spending our billion dollars wisely. Finding and vetting the most promising carbon removal technologies occupies the vast majority of our time. In 2023 we contracted $157 million in offtakes. Each deal requires deep technical and commercial diligence, as we’re largely assessing a company’s future potential to be a meaningful part of the world’s carbon removal portfolio.

  • Defining (and refining) our capital deployment strategy. How do we spend one billion dollars to be maximally catalytic to the field? How much should we allocate per pathway (i.e., enhanced weathering versus direct air capture versus biomass versus our other options)? Within each pathway, what does a great company look like? To inform the deals we do, we spend a lot of time developing (and refining) our point of view by pathway and for the portfolio overall.

  • Contracting. For Frontier, we decided to align on a single offtake form for all of our deals, which the Frontier team uses as a template to pre-negotiate each carbon removal deal for all buyers. While on the one hand this should lead to a more streamlined deal process, at the beginning this took significant effort. Getting buyers, suppliers, and banks to agree on a single offtake agreement can be very time consuming. Finding lawyers who are creative, resilient, and forward-leaning has been paramount. And, regardless of the form you use, getting organizations to sign multiyear, multimillion-dollar deals simply takes time and energy. (It’s possible that contracting may take less time for other AMC designs – perhaps if there are fewer stakeholders, or if the product is easier to assess than carbon removal, or if there are existing precedents for how the product is purchased, like energy and cement. Though even Gavi had thousands of pages of supply agreements.)

  • Making prepurchases. We currently run one round of prepurchases a year, which is a significant undertaking. The 29 prepurchase companies in our portfolio today are the end result of evaluating hundreds of applications.

  • Supporting the portfolio. Carbon removal companies need customers but they also need many other things to be successful. What else can we do to maximize the likelihood they’ll succeed? This includes things like helping them fill critical roles, writing support letters for government grants, talking to potential financing partners, and helping them work through unique policy challenges.

  • Verifying and accepting deliveries. This might sound straightforward, and in some cases it can be. But for Frontier, defining and verifying a ton of CO2 removed from the atmosphere is a work stream unto itself. And yet, much of this market infrastructure is still emerging – we do significant ad hoc analysis for each deal delivering for the first time.

  • Adding new buyers. The initial $925 million was a good start, and we need demand for carbon removal to keep growing. Since launch, new buyers like JPMorgan Chase, Autodesk, H&M, Workday, and Watershed have joined Frontier.

  • Catalyzing demand-side policy. Carbon removal won’t succeed without governments creating a large-scale market for it. We spend a lot of time on this, in part through a policy organization we helped to establish in 2023.

  • Administrating. As with any organization, we also have our fair share of administrative work – creating individual contracts for each buyer for each deal, collecting dues, planning, budgeting, and hiring.

The team that does this is about sixteen employees: seven ‘quarterbacks’ who own outcomes for specific projects (like deals), seven subject matter experts who work across those projects (expertise includes science, finance, policy, legal, and communications), and a two-person product team that build things like our knowledge gap database, the website, dashboards for Frontier’s buyers, and tools to make the Frontier team more efficient.

More AMCs

AMCs are a generalizable tool that have the potential to help solve a wide range of global challenges. And yet there are still only a few real-world examples.

That’s hopefully on the cusp of changing. In the last six months, I’ve had conversations with people thinking about how AMCs might be good fits to solve a number of challenges. What unites them is that they’re areas where a product should exist but doesn’t because there’s uncertainty around who will pay for the product, or if they’ll be able to pay enough.

Here are a few products that an AMC could help pull into existence and scale.

  • Low-/zero-carbon cement. Cement made out of recycled industrial by-products like fly ash or calcined clay, referred to as green cement, can be 70 percent less carbon intensive than regular cement. But right now this cement alternative is expensive and small-scale. Scaling up these new solutions quickly would have enormous societal benefit, but would also be costly. An AMC that paid the ‘green premium’ as these technologies are getting down the cost curve could incentivize faster scale-up.

  • Low-/zero-carbon steel. The case for a green steel AMC is similar to the case for green cement.

  • Greenhouse gas removal. Frontier focuses on carbon removal, but increased concentrations of methane and nitrous oxide are also major contributors to warming. Yet how to remove methane and nitrous oxide already in the atmosphere (or even methane destruction at non-flarable concentrations) is unclear. Sending a technology-agnostic demand signal could motivate researchers and entrepreneurs to spend time developing solutions.

  • Climate-resilient crops. Like green cement and carbon steel, climate-resilient crops are a technology with a huge mass market, but which will cost a lot to develop. An AMC could pay for that investment.

  • Health products where demand in high-income countries is uncertain, like strep A vaccines and hepatitis C vaccines. Most kids get strep throat, but vaccine companies aren’t sure whether the Centers for Disease Control and Prevention would recommend that a strep A vaccine be included in childhood regimens if one existed. Separately, many people who inject drugs are reinfected with hepatitis C multiple times – but would they receive a vaccine to prevent reinfection if they’re not insured? In America, the Centers for Disease Control and Prevention, probably in consultation with the Food and Drug Administration (which approves drugs) and the National Institutes of Health (which funds most public health research in the USA), could convene an advisory group to set a target product profile for these two vaccines, confirming under what conditions they would recommend kids get the vaccine and how much they’d pay per vaccine course. That would set a clear demand signal for vaccine inventors.

  • Health products where the need in lower- and middle-income countries is disproportionately high, like vaccines for tuberculosis, syphilis, and malaria. Gavi could be a natural home for these AMCs when they are for vaccines that don’t exist yet like adult tuberculosis and syphilis (which, through infected mothers, still kills up to 100,000 babies each year); the President’s Emergency Plan for AIDS Relief (known as Pepfar) could be the host when the product is improved HIV/AIDS drug regimens; and the Global Fund could host AMCs for as yet uninvented malaria, TB, and HIV products, like monoclonal antibodies for malaria.

And there are many more areas with promise. The Market Shaping Accelerator recently published the ideas they’re most excited about. They think AMCs might be useful for driving the invention of broad-spectrum antivirals and personal protective equipment, and repurposing generic drugs for new uses.

If you are starting or considering starting an AMC, Frontier would love to be helpful if we can. In an attempt to make this piece as actionable as possible, attached is a worksheet. If you fill it out and send it to us at [email protected], we’ll read it and, if we think we can help you, get on a call to discuss it.

In the meantime, I’ll leave you with a few final thoughts.

First, don’t worry about fitting your solution perfectly into a ‘theory’ box. Reality is messy. At the end of the day, the right design is the one that solves both the overall problem as well as the specific needs of buyers, suppliers, and any other key players – regardless of how theoretically clean it is.

Second, speed matters. It took us about seven months to go from coming up with the concept to publicly launching Frontier (though we had a head start from our Stripe Climate work). Speed matters for a few reasons:

  • Big, important problems demand we move fast. The cost of waiting is high.

  • The world around us changes and what may be impactful or even possible one year may no longer be the next year. For example, the economy changed drastically just a few months after we launched Frontier, resulting in layoffs at many companies. We got lucky with the timing. I suspect if it had taken us longer, things might have fallen apart before we’d even launched.

  • There is a real risk of getting stuck in the theory when so much of the learning takes place in the doing. As in most things, the faster you ‘do’, the faster you learn.

Finally, the rate limiter to more AMCs is likely not money, but great teams that both have a vision of what could be and can execute like hell. There’s a lot of philanthropically minded capital available, and my anecdotal observation is that donors are increasingly open to and actively looking for creative, nontraditional ways to improve the world. Similarly, companies are spending substantial amounts of money on climate in particular, especially activities that are brand differentiating. And yet there have still only been a small handful of AMCs tried at all. I suspect that the primary thing standing between many more of these existing is great people willing to run through walls to make them happen. If this resonates with you even a little bit, consider deciding to be one of those people.

Nan Ransohoff leads Stripe Climate as well as Frontier, a $925 million advance market commitment to accelerate carbon removal.

Subscribe to The Works in Progress Newsletter

Launched 4 years ago

Get the latest updates on our magazine, featuring new and underrated ideas to improve the world. Visit our website:

  • Support The Works in Progress Newsletter

    Since you liked this post, why not subscribe?
A guest post by
Head of Climate @stripe. I also lead Frontier, an advance market commitment for carbon removal (

Ready for more?

© 2024 Works in Progress
Substack is the home for great culture