The money that takes a deep tech company from a working prototype to a first production line does not come from people who assess technology. It comes from project finance, and project finance underwrites demand. Before that round, the question that decides the outcome is not whether the science works. It is who has agreed to buy it.
I heard this put more plainly than I have managed to put it myself, by an investor describing the different kinds of money a hard tech company needs at different stages. On the last stretch, he said, they care a lot less about your technology and a lot more about who is going to buy it.
That sentence should change how a deep tech company plans its spending, and for most of the companies I speak to, it has not.
What actually changes at the capital wall
The costs in deep tech do not rise smoothly. They rise in one step.
Take a company improving the cost curve on something that already exists, a battery anode, a solar coating, a heat pump component. Getting from an idea to a validated result in a laboratory is expensive, but it is fundable expense. The investor looks at the science, the team and the patent position, and makes a judgement about whether they see the financial upsides.
Then the company needs to make the thing at volume, and the number required stops being a research budget and becomes an infrastructure budget. The investor who backed the science cannot write that cheque, and neither can the fund behind them. The money has to come from somewhere that finances plants, and the people who finance plants have a completely different job.
A project financier is not assessing whether your coating is elegant. They are assessing whether the facility will produce a return, which means they are assessing whether the output has buyers. Your technology is an input to that question rather than the question itself.
Why project finance cares less about the technology
Because by the time a company reaches that conversation, the technology risk is supposed to be largely retired. That is what the previous rounds paid for. What remains is market risk, and market risk is the thing that can sink a facility.
A plant with no buyers is a liability that costs money every month it stands still. A plant with signed offtake is an asset with a revenue line attached. The difference between those two outcomes is not in the laboratory. It sits in a folder of commercial agreements that somebody had to spend two years building.
This is also why the same logic applies well below plant scale. A Series B investor is asking a softer version of the same question. They want to know whether anyone outside the building wants this, and how they can tell. Companies that can answer with names and dates raise more easily than companies that answer with performance data, even when the performance data is better.
What counts as demand evidence
Not all of it weighs the same. Roughly, from strongest down:
A signed offtake agreement, where a buyer has committed to purchase a stated volume at a stated price. A letter of intent, which commits nobody but says publicly which company is interested and in what. A paid pilot, which proves somebody valued the work enough to move budget. A named company in active technical evaluation, with a real contact and a real timeline. And underneath all of that, a documented pipeline showing the right roles at the right companies know who you are and have engaged.
The bottom of that list is where marketing lives, and it is worth being honest that the bottom of the list is the weakest evidence on it. What the bottom of the list does is feed the top. Offtake agreements do not appear from nowhere. They come out of conversations that started somewhere, with someone who had heard of you and thought you were credible enough to take the meeting.
What we can and cannot do about this
Here is where I should be straight about the limits, because this argument is easy to oversell and I have seen agencies oversell it.
We can’t get you an offtake agreement. That is a commercial negotiation between your company and a buyer, involving your terms, your volumes and your chief executive. Anyone who suggests a marketing budget produces signed offtake is either confused or selling.
What the work does is narrower and still worth doing. It identifies which companies in the chain actually have the problem you solve, which is a research job that most deep tech companies get wrong on their first attempt. It gets you in front of the specific people at those companies. And it makes the company look, on inspection, like an organisation a large buyer can sign with, which matters more than founders expect, because a procurement team that finds a thin website and no visible track record has to justify the risk internally.
There is also a timing point that argues against buying anything from us. If you are three years and two technical milestones away from that raise, spending on demand generation now is premature. Get the milestones. If you have no clear view of which industry you are selling into, spend the money on finding out instead, which you can do yourself and which I have written about separately in the piece on running 100 customer interviews in 10 weeks. The work described here pays off when the target is roughly known and the round is roughly visible.
When to start
About eighteen months before the round it is meant to support.
That sounds long until you count backwards through an industrial procurement cycle. A large manufacturer takes months to move from a first conversation to an internal champion, months again to get technical evaluation scheduled, and months again to get anything signed. Three or four of those running in parallel is a year and a half of work, and it cannot be compressed by wanting it more.
The alternative is what most companies do, which is to notice six months out that the deck has no commercial traction slide, and assemble something. Investors read a lot of decks. Evidence assembled in a hurry looks assembled in a hurry, because the dates all cluster and the names are all early stage conversations dressed as pipeline.
What this means for how you spend
If demand evidence is what unlocks the round, then the budget for building it does not belong in the marketing line, competing against laboratory equipment. It belongs against the round.
That reframing matters practically, because it changes who approves it. A marketing budget is argued over with whoever guards operating costs, and in a deep tech company that person is right to guard them. A capital raising cost is a board conversation, and boards understand spending money to make a raise go well.
It also changes the sequencing. Website, film and campaign work is not decoration to be added once things are going well. It is the machinery that produces the evidence, and it needs to exist before the eighteen months start, not during them.
Peter Collins ran Permasense, an Imperial College spinout in industrial sensing, and took a version of this to its logical end. Rather than raising equity to fund development, he got BP to fund early pilot deployment as a customer. His line was that with revenues you have a business, and without them you have a project. Very few companies can do exactly that. But the instinct behind it, that a paying customer is worth more than a slide about a market, is the same instinct that makes a buyer list the most valuable document a pre-raise company owns.
If you want to talk about what building that would involve for your company, a 20 minute call gets you a straight answer, including whether it is too early. Most of what we do sits under LinkedIn campaigns and demand generation, and the Albotherm project shows how the website and photography work underneath it fits together.
Frequently asked questions
What is the capital wall in deep tech?
It is the point where the money needed to keep going jumps by two or three orders of magnitude, because the next step is a production facility rather than a laboratory. Reaching a working prototype might cost single millions. Building the first plant that makes the product at cost can run to hundreds of millions. The funders change at that point too, from investors who assess technology to project finance, which assesses demand.
What counts as demand evidence for investors?
In rough order of weight: signed offtake agreements, then letters of intent, then paid pilots, then named companies in active technical evaluation, then a documented pipeline of the right people at the right companies who know who you are. The strength comes from how much the buyer has committed and how specific the commitment is.
When should a deep tech company start building demand evidence?
Around eighteen months before the round it is meant to support. Named buyers in active evaluation take that long to accumulate in industries with long procurement cycles, and evidence assembled in the three months before a raise looks assembled, because it is.
Can a marketing agency get me an offtake agreement?
No. An offtake agreement is a commercial negotiation between your company and a buyer, and it is signed by your commercial lead or your chief executive. Marketing identifies who the right buyers are, gets you in front of them and makes you look like a company worth signing with. The negotiation is yours.
Does this apply to companies that are not building factories?
The extreme version applies to capital intensive manufacturing. A softer version applies everywhere in deep tech, because every investor at every stage is trying to work out whether demand exists outside the founder’s conviction. The evidence is smaller and the timelines are shorter, but the question being asked is the same one.