Your customer says they'll buy it. Will their bank believe them?

There is a lovely moment in capital-intensive deeptech when sales and fundraising suddenly become the same job. You have built the pilot. Customers like what comes out of it. One of them tells you that, yes, if you build the commercial plant, they would happily buy 10,000 tonnes a year. Excellent. There is only one small circularity left to resolve: you cannot build the plant until somebody finances it, the financier would rather not finance it until there are customers, and the customers would rather not commit until they know you can build the plant. Welcome to the offtake agreement, one of the more useful commercial instruments deeptech founders rarely encounter until they suddenly need one very badly.

An offtake agreement is basically a promise about future purchases. That sounds unremarkable until the thing being purchased does not yet exist at commercial scale. Then the promise can become part of the financing architecture of the company. The OECD’s 2026 Climate Club Financial Toolkit explains the mechanism unusually clearly: long-term purchase agreements can underpin future project revenues and thereby help secure external financing. Fixed prices, price floors, indexed pricing and take-or-pay provisions can all reduce revenue and volume uncertainty. In other words, the customer contract is no longer merely the thing you get after building the factory. In the right form, it becomes part of what allows the factory to be built in the first place.

This is already normal in energy, mining and large infrastructure. It is becoming increasingly relevant to climate tech, critical materials, industrial biotech, new chemicals and other deeptech businesses where commercial scale requires tens or hundreds of millions before the first serious production revenue arrives. And it changes what “traction” should mean. Twenty interested customers can be less valuable than one customer willing to put its balance sheet behind future demand.

Avantium provides an almost absurdly useful case study because you can watch this progression happen across several years. When the company was preparing its first commercial FDCA plant in Delfzijl, it did not wait until the factory existed and then start looking for buyers. In April 2021, Avantium announced that five conditional offtake agreements already represented commitments for more than 50% of the planned plant’s output. The company explicitly described that as an important step towards the final investment decision for constructing the plant.

That is the first important distinction. A customer can validate your product. An offtaker can validate your capacity.

Those are not the same thing.

The customer running 50 kilograms through its process tells you whether the material works. The customer agreeing today to buy meaningful future production tells investors something else: if you spend the money required to scale this, somebody intends to absorb the output.

Avantium then built different flavours of this commitment with different customers. Carlsberg Group signed a conditional agreement securing a fixed volume of Avantium's PEF for packaging applications. Plastipak later signed another conditional offtake after evaluating PEF in beverage and food packaging. Other agreements opened textile, industrial-fibre and insulation applications.

There is a strategic lesson hiding in that portfolio. An early industrial plant does not only have a volume problem. It has a concentration problem. Signing one enormous customer feels fantastic until that customer's product launch slips twelve months and half your factory suddenly has nobody to sell to. Several offtakers across different applications can therefore be more valuable than the same nominal volume concentrated in one program.

The OECD calls this offtake risk. Even with a contract, financiers still care whether the buyer is creditworthy, whether the agreement is enforceable, how long it lasts and what happens when demand or prices move. It explicitly lists diversification across several offtakers as one way of reducing that risk.

So merely announcing an “offtake agreement” tells us remarkably little.

This is where founders, and frankly investors too, need to become much more annoying about the details.

Is it binding? Is it conditional? What conditions have to be satisfied? How much does the customer actually have to buy? Is there a minimum annual volume? What determines the price? When do purchases begin? How long does the agreement last? Can the customer simply decide it no longer wants the product? What happens if your plant starts six months late? And does the commitment cover enough of the future capacity to make any difference to the financing case?

The difference can be enormous.

Avantium's 2023 agreement with Origin Materials is particularly interesting because it stacks several commercial mechanisms on top of each other. Origin paid an upfront amount, then a €7.5 million milestone fee when the industrial technology licence was signed, with additional licence payments tied to development milestones and royalties due once production begins. Alongside that licence sits a conditional offtake agreement under which Origin agreed to purchase gradually increasing minimum annual volumes of FDCA on a take-or-pay basis.

Take-or-pay is where the promise starts acquiring teeth. Rather than saying “we expect to buy 10,000 tonnes”, the buyer agrees to take an agreed quantity or incur a payment obligation if it does not. The exact structures vary enormously, and this is firmly lawyer territory when you actually draft one, but commercially the principle is beautifully simple: some of the demand risk moves from the producer to the customer.

Why would a customer agree to that?

Because they are getting something too. Secure supply. Preferential access to scarce early capacity. Price certainty. A strategic source outside a problematic geography. First-mover advantage. Perhaps influence over specifications or development. The same reason a startup should not give exclusivity away for free applies here: real commitments need value on both sides.

And when the underlying technology requires serious capital, those commitments can become enormous.

In March this year, critical-metals refining company Nth Cycle signed a binding ten-year, approximately $1.1 billion offtake agreement with commodities trader Trafigura. The deal covers future nickel and lithium carbonate output and sits alongside Nth Cycle's planned expansion in the United States and the Netherlands. That is not somebody politely telling the founder that the technology is interesting. It is future demand being turned into a commercial asset.

Carbon removal has pushed the idea even further. Frontier was deliberately created as an advance market commitment, essentially organising buyers to promise demand for technologies that still need to scale. In March 2025 its buyers agreed to pay Eion $33 million for 78,707 tonnes of future CO₂ removal between 2027 and 2030. The point is not simply to purchase carbon removal once it exists. The purchase commitment itself is intended to help create the market in which those technologies can become commercially viable.

That gives us a more useful way to think about the ladder of commercial evidence in capital-intensive deeptech.

At the bottom is interest. A customer likes the technology.

Then comes technical evidence. They tested it and it worked.

Then intent. They sign an LOI or tell you what they expect to buy.

Then conditional commitment. They agree to future purchases if defined things happen.

Then contracted demand. Volumes, duration, pricing mechanics and obligations become increasingly real.

And somewhere near the top sits the thing a project financier can actually underwrite.

These stages should not be collapsed into the word traction.

There is a nice contemporary example of why. In October 2025, REalloys announced an LOI for a ten-year agreement covering 15% of planned Phase 1 production from the Tanbreez rare-earth project. Useful signal. Seven months later, in May 2026, that LOI was replaced by a definitive long-term offtake agreement covering 15% of monthly Phase 1 production. Same underlying commercial relationship. Very different evidentiary value.

That distinction matters when raising money.

If I were looking at the commercial slide of a deeptech company that needs €80 million to build a first plant, I would not be terribly impressed by “€300m pipeline” unless I understood what sits underneath it. How much is a CRM opportunity somebody created after a conference? How much has been technically validated? How much has an LOI? How much is conditional offtake? How much has minimum volume? How much is take-or-pay? How creditworthy are the buyers? When does the demand actually start?

A €20 million binding commitment from a credible industrial buyer may tell me more than €500 million of Excel.

There is an implication for sales strategy too. Once you know that commercial commitments can help finance scale, the job of the first sales process changes. You are no longer only trying to maximize near-term revenue. You are trying to assemble enough credible future demand to unlock the next stage of the company.

That may mean accepting several smaller offtakers rather than chasing one giant customer. It may mean targeting a famous, creditworthy buyer because their commitment carries financing value beyond the revenue itself. It may mean negotiating minimum volumes rather than celebrating a huge but non-binding forecast. It may mean giving an early customer preferential access to capacity in return for committing before the plant exists. And it may mean designing commercial milestones explicitly around the next financing decision.

The uncomfortable question for a capital-intensive deeptech founder is therefore not:

“Do customers want this?”

You probably established that several funding rounds ago.

It is:

“What are they willing to commit before we build it?”

Because somewhere between a successful sample and a billion-dollar factory, enthusiasm needs to become paper.

And not all paper is equal.

If you're trying to turn early industrial demand into something strong enough to support a scale-up decision, that is exactly the kind of commercial architecture SLSCRW likes working on.