Last updated February 25, 2024
A few ground rules before beginning. I don’t like the term “deep tech,” and so will not use it. There are many tech companies that exist in the physical world and sell products other than software. There are a subsection of these companies that employ a number of different technologies, systems, and processes to rethink large scale manufacturing of a wide variety of products. I call these companies “Industrial Tech” companies. Climate companies, like Bowery Farming, which I used to work at from 2017-2022, often fall into this bucket, but don’t always. Companies like Senra Systems, Locus Robotics or Hadrian also fall into this category.
There are a few core phases in the life cycle of a industrial tech company. Different people have different views, and phases can certainly fracture depending on the vertical, but in general I think most industrial tech company lifecycles can be viewed in 4 core phases: 1) R&D / Formation, 2) Pilot Phase, 3) Commercial Demonstration, 4) Large Scale Deployment
This phase is characterized by the lab and the development of a science and foundational IP. Aside from developing the initial critical science and engineering behind the concept, teams are developing their technological milestones or technological “proof points” that they believe will be required to validate and minimize the technology risk inherent in the venture. This is also where teams are developing a techno-economic analysis, but it is just that, analysis and a roadmap. Typically this phase is incubated at universities, or through grants, and incubators. Depending on the access of the founders, this phase may even be too soon for early stage VC. We had a version of this at Bowery Farming, which where one of the co-founders did the first ~30 day grow cycle with a mini vertical farm he built in his basement.
Funding Type: Grants, Early stage VC, Incubators
Funding Size: $0-2M
This phase is defined by a deployment of initial technology at an uneconomically sized facility. In the pilot phase, teams have clear technological milestones that they set out to achieve. The key word here is technological. These are not often true operational milestones, which can only be achieved at a larger scale. Technological milestones are still largely built around proving out the initial science and engineering hypotheses of the team. Teams trade at-scale replicability for greater control and lower risk and a smaller capital outlay.
In many ways, this is the closest phase to traditional venture backed startup lifecycles. Most software businesses raise a seed to go develop the product toward true Product Market Fit. The round is for testing, and further proving out a given hypothesis. Unsurprisingly, this stage is typically funded by typical VC investors. They can deploy $2-5M into a round at a hopefully cheap price. The difference here is that the industrial company is far from product market fit. If a software company doesn't have clear PMF, it’s unlikely that they will be able to raise an A. That PMF is demonstrated through revenue growth and customer engagement, with customers seemingly pulling the product out of the team. For industrial companies, the Pilot Phase has none of this. If anything, there is likely zero revenue, and if there are any commercial conversations, they are defined by 1-2 anchor partners, and they are decidedly not the small software companies that b2b SaaS companies have the luxury of iterating with or failing fast with. Instead they are often massive, potentially slow moving institutions.
Today, this is where most of the VC dollars go. But the journey beyond this point does not look anything like traditional software VC investing. For an industrial tech company, true PMF comes after successfully surpassing Phase 3, whereas in a SaaS company, PMF happens somewhere between Phase 1 and 2.
Funding Type: VC
Funding Size: $5-10M
This is when things get tricky. Commercial demonstration is defined by development of 1-2 large projects that more closely resemble what an at scale project is going to look like. You can think of it like V1 of the scaled up model. No longer is a company demonstrating technological feasibility. They are demonstrating technological, operational, and commercial feasibility all at once, and the latter two are likely for the first time. And anyone who has operated in an industrial, physical world company before, science and technology behaves differently at this scale, no matter how sure you were about it in the pilot phase.
Depending on the company, this phase can resemble a 10-50x increase in scale compared to the pilot phase, and it typically requires $50-100M to pull off. Most VCs do not have the capital to back such an endeavor. The risk is enormous, potentially even higher than before because the team will have to ramp up their operations and build new operational and supply chain skills that were wholly unnecessary at the pilot phase. This phase will also have to prove that there is legitimate commercial demand for the product, and if there is not, the economics of the Commercial Demonstration phase will falter even with flawless operations. One might argue that this is actually the riskiest phase for industrial tech.
But if done correctly, the other side of this phase is an incredibly bankable business in a massive market with an incredible technological moat. Despite this being the riskiest phase, one could argue that most of the activity after this is about repeating the model and finding scale. This is the true PMF stage.
Funding Type: Grants, DOE, ???
Funding Size: $50-100M
The model has been proven, the TAM is large, the economics are demonstrated, and the projects are financeable. It is not easy to rapidly scale any large deployment of technology, but the “playbooks” have been written, and it is now about pouring fuel into the machine so that the machine can achieve its potential. This is where hundreds of millions of dollars or billions of dollars are deployed, with a significant portion being debt. You could argue that there is probably a middle ground here of “repeatable playbooks” but where the upside isn’t so “obvious” to creditors. This is where a firm like Generate Capital in the climate world plays a lot, financing and operating things like anaerobic digesters.
Funding Type: Bank credit, private credit, PE equity, Growth VC
Funding Size: $300M-5Bn
It should probably be obvious from the description above that the industrial tech valley of death is Phase 3, Commercial Demonstration. While this isn’t a new idea, this also isn’t something that a lot of people are talking about. In climate circles, however, this dynamic is well known. Here are several different publications describing the different industrial tech life cycle phases:
They all pretty much say the same thing. The phase between the Pilot and the Commercial Demonstration is rife with danger. What I’m arguing here is that this is because it is structurally difficult from an operational and growth perspective, but ALSO because there is a complete gap in financing available to carry companies through this phase, whereas there are clear pools of capital for every other phase.
So why is this such an unfinanceable phase? Because Phase 3 usually requires $50-100M in capital to hit the relevant milestones AND there is still significant risk in the business. When venture firms, especially in this market, deploy $50-100M in capital, they are typically financing a well oiled machine with clear playbooks operating in an economics model they have underwritten 200 times before (e.g. underwriting to the Rule of 40 or others in b2b SaaS).
Industrial tech companies, however, at this phase are still proving out all the core ingredients that will make the company successful. Industrial tech companies have not yet reached PMF at this stage, and those who have the capital to finance this stage, rarely invest industrial tech, although places like A16Z (American Dynamism) and General Catalyst (Industrial Resilience) are more explicitly doing to so now. They are the exceptions, not the rule.
For climate, most of the venture capital is concentrated in the seed stage. Firms like Voyager, USV, Prelude, Congruent, Lowercarbon - all the well known brands in climate are ALL early stage VC. Only Lowercarbon will occasionally write a $20M+ check. Most VCs can’t write large enough checks, and are simply not incentivized to take on the risk inherent at this stage, particularly given the price they could get at the seed stage.
Physical industrial tech needs to graduate from VC**,** whereas software never really does.
As I alluded to above, when VCs deploy $50-100M checks, it’s usually a question of pouring fuel on a fire. It’s a matter of “doing more and doing it faster.” However, as I learned very clearly at Bowery (and this is the case with most physical infrastructure), the Commercial Development Phase (Phase 3), is NOT a question of “just doing it bigger.” A facility 10-20x the size of your last one requires a completely different set of operational considerations, project management, and supply chain management.
A really simple example of one I experienced at Bowery was the maintenance capex relationships in our first, second, and third farms were all very different from each other. The same dynamic existed with the supply chain, training staff, and ramping the farms. In your pilot facility, “ramping” doesn’t mean much. But in your Development facility, ramp time will have a massive impact on your ability to prove out the cash flow profile of your facility. This requires operational excellence, and frankly probably operational and supply chain considerations a founder hasn’t experienced before. Even at Bowery we had to cycle through veteran operators until we found the right fit.
In the ZIRP world, a lot of VCs didn’t understand this difference. Traditional VC unfortunately brings little expertise when it comes to financing infrastructure, let alone building and ramping operational and supply chain expertise. They’re not underwriting companies like this anymore anyway, but I think it is important to note that traditional VCs are also of minimal value-add to industrial tech companies at this stage.
The CEO of Generate Capital (courtesy of CTVC) has a great (long) quote on why the climate tech capital stack specifically looks completely different from early stage VC, and how there is a major gap in the market today (highlights are my own):
There are certain players in the ecosystem now taking on new risks than they have in the past. Twenty-five years ago, I was in the venture world for technology companies, primarily focused on software, and a late-stage venture investment was a company that was producing a lot of revenue. If it wasn't producing a lot of revenue, it was an early-stage venture capital investment. Well, unless you've built a bunch of projects, or sold your technology into a bunch of projects as a technology company in climate tech, you don't have a lot of revenue. There may be a need for rethinking venture capital stages when you've got project risk involved in the proof of the technology's viability. And it's exactly here where there’s the capital gap. But there's a capital gap in large part, in my view, because there's an operational gap. Most developers' incentives are to build projects they can sell. That's how they make money.
You can't sell a project if everybody who is in the buyer universe thinks it's too risky to buy. The technology companies often try to build that development capability themselves out of necessity, but will their venture capitalists fund that? That's still early-stage venture because they don’t have revenue. So the cost of capital for those first projects, not just the first project, but for the projects, is quite high. Do the economics for those projects then work with that high cost of capital? Far too often, I hear from an entrepreneur or a venture capitalist, “We've proven it at pilot scale. It's ready. It's ready for Citibank to make a loan and for Generate to take the equity stake.” That's the wrong conversation to have.
The conversation needs to be, “How many hours has it operated successfully, continuously in the exact form we're talking about proposing for this next iteration of it? How many blue-chip players have been willing to put their seal of approval on it, whether it's an insurance provider or an EPC or a customer willing to buy the output?” We still haven't really addressed this exact point that once you're at pilot stage, how do you get to an at-scale roll-out of the solution that mainstream capital will find safe enough to put their money at risk for? That's both the capital and human capacity gap that really represents the continuing valley of death for many of these solutions.
In commenting on this funding gap, CTVC claims that there are some who are filling the gap specifically in the climate space:
The good news is that there are more vehicles that have emerged to finance and support startups through this gap along with Elemental - such as Breakthrough Catalyst and LACI Market Transformation. These are largely powered by philanthropic dollars, and the vehicles are specifically designed to enable companies to scale to the point that debt vehicles (i.e. Generate Capital with $3B to deploy), government programs (i.e. the Loan Program Office with $40B to deploy), and other capital can get in the game.
How many companies that reach Phase 3 over the next 10 years will be able to rely on philanthropic grants or even government programs, which are often slow to deploy cash?
Even the few companies that have successfully bridged the valley of death have been heavily reliant on government subsidies.
For example, H2 Green Steel announced this month a $4.5Bn debt funding round paired with $325M in equity and $260M in EU government grants. H2 Green Steel is one of the largest industrial tech successes of the last several years in terms of production scale. But when we look a little closer, it becomes clear that even this seemingly “Phase 4” funding round is actually not very reliant on traditional credit. A little over 50% of the debt is almost entirely guaranteed by the Swedish Government. This is of course great for H2 Green Steel, but it is not practical to assume every company can reach scale via this path.
LanzaJet, another at scale industrial tech success story, recently launched its largest facility to date, which was buoyed by a $50M philanthropic grant from Breakthrough Energy Ventures which came at a critical time during construction and unlocked more favorable economics for the project.
Neither of these stories are negative in their own right, but they indicate that even our most successful first of a kind climate projects of the last several years have largely been propped up by government subsidies or grants. While this is critical in the earlier stages, it is not sustainable for driving the rapid deployment of new industrial technology.
Inherent in this entire discussion is the belief that there will be an enormous number of large scale physical industrial tech infrastructure deployments over the next 30 years, and that they all will need to be financed into the deployment phase. So how much demand do we really think there is?
There is a lot of room to go deeper here, but I did a very high level bottoms up approach to scoping out the amount of dollars that actually need to be deployed in the near future.
Let’s say that there are a certain number of climate tech industries (focusing on climate makes the data more manageable) that require significant capital deployment at the Commercial Development phase. Those industries are:
Direct Air Capture, Carbon capture and storage, sustainable aviation fuels, electric aviation, nuclear, green steel, alternative proteins, water treatment, decarbonized cement, biofuels, hydrogen, recycling, chemicals, alternative plastics, large scale geothermal, medium scale geothermal, alternative energy storage, indoor agriculture (greenhouse, algae, vertical farming), ocean mining, alternative textile manufacturing, EVs (shipping, motorcycles, cars), micromobility, robotics, built environment energy efficiency infrastructure (HVAC)
Currently, based on the NetZero database, there are ~18,000 companies across these verticals in the world that exist today. As we know, the vast majority have not reached scale yet.
Let’s assume that 5% make it past the pilot phase and into the Commercial Development Phase. Their riskiest and largest projects will be their first 2 “at scale” projects. Let’s say those go for an average of $50M a project.
18,000 * 5% = 900 companies
900 companies * $50M * 2 projects = $90Bn
This of course can flex up significantly based on how much capital one can feasibly deploy alongside the growth of a company. For example:
900 companies * $50M * 3 projects = $135Bn
Assumptions on the upside:
Assumptions on the downside:
This bottoms up analysis is very high level, but it should give you at least a direction that there will be many tens of billions of dollars that need to be deployed at this specific phase, and the model is already constrained by just using climate company data.
I’m currently building a list of companies that could be financed today. The criteria here is:
Right now I think the list will net at around ~450 companies, and that’s coming all from one database, and only climate related companies. This filter is also highly susceptible to missing a fair number of companies that actually have reached beyond the pilot phase with less than $30M of equity capital invested.
Email me for the list at [email protected]
It’s clear that VC cannot and will not fill the gap. The economics don’t work in their favor and there simply is not enough capital to be deployed. Admittedly, this is an area that requires a significant amount of detail and analysis and actual modeling of outcomes, which will come in a v2 of this. But for now, let’s just outline the a few going high level hypotheses:
Structure
A credit facility with high warrant coverage can fill the gap. Venture debt [warrant coverage](https://flowcap.com/warrants-in-venture-debt/#:~:text=to be exercised-,Average Warrant Coverage on Venture Debt Deals,anywhere from 10-20%25.) is usually 10-20% which ends up being 1-2% of the company. This debt typically has high teens interest rates. Development stage industrial tech businesses would be better off with slightly lower interest rate terms, longer duration than they would get with venture debt (which has largely dried up too), but also much higher warrant coverage, perhaps 30-50%, so that as a fund you can participate significantly in the upside. While I don’t yet have the numbers to back up the fund math, I know from Bowery experience that this would have been preferable. We were able to raise debt from some of the best lenders in the world (KKR), but most won’t be able to do that, and also those times are largely gone. In a more sober world, that line of credit may have instead been contingent on making one project profitable rather than being contingent on building 3 farms as it was.
Investing like this requires real clarity about the full cash flow profile of your customer, and must avoid any company at a stage of maturity where interest is covered with equity dollars.
A credit fund would likely still need to purely finance on a project basis to begin, and would be best suited to create structure such as a ROFR on financing the next X projects so that the credit fund can grow with its biggest hits.
Return Profile / Fund Dynamics
A credit fund typically looks for a 1.3-1.6x multiple in 2-5 years. Commitment periods are around 2-3 rather than PE or VC at 5+ years. During that time frame, the key for a credit fund is to recycle any cash that comes in. Principal pay down or interest will be rolled into other deals, which helps make the returns a little juicier. Many credit funds will also lever their funds, for example at a 1:1 match with the existing fund. This can also help boost returns, but is likely not where we would operate with the first fund.
Top tier, 90th percentile returns are in the 1.5-1.6x returns range. In the heyday of SVB and ZIRP, venture debt was yielding in the 2x range because of their warrant coverage. Of course, like all venture, the lionshare of the returns from that warrant coverage came from a few companies.
Although credit funds usually attract LPs like endowments or pension funds, given the higher risk of this type of fund and the potential addition of climate-related portfolio, higher risk tolerant LPs such as sovereign wealth funds might be better suited to fund this structure.
Credit funds don’t operate with a 2 and 20 model, and typically operate at a 0.5-1 and 15 model. That said, if the model proves successful, credit funds can become very large asset gatherers and do not require large teams. So the goal is to prove out the model and move the AUM to the many $ billions with a relatively lean team.
Structure
The most popular model here is the growth equity + equity project finance + syndicate debt model. This is the model used by Springlane and Vision Ridge (see below). This model is more hands on, slower to deploy capital, but has significantly higher upside.
The basic model is to find a company that has completed a pilot and has a few large offtake agreements in place with large scale buyers. The fund will then invest 10-20% of a total equity investment into the holding company (HoldCo), and invest 80-90% into a specific project (typically deploying anywhere from $40-70M altogether. They will also syndicate the debt for the project.
Critically, they will own 100% of the project, and will bring in experts to help manage construction and operation of the project, and will divide some portion of the revenue with the HoldCo. They will also have a ROFR on investment in the next several projects, allowing them to invest more deeply in their winners.
The pitch is that they take on the risk of the “Commercial Development” stage as well as the upside, and in doing so derisk the HoldCo and bridge the gap between pilot and bankability, while having significant equity upside at the HoldCo and at the project level given their ability to continue to finance subsequent projects.
The equity target returns, fund life cycle, and fee structure are much closer to PE style returns. Although this model clearly has more upside and control than a debt driven one, I am a bit dubious of it being attractive to the best companies out there. For example, would a Solugen give up this level of operational control but retain greater HoldCo control, or would they prefer to continue to raise equity as they have.
Regardless of the fund structure, the fund must be not only a source of cash but also one of operational expertise. This is particularly apparent in a credit fund or an equity + project finance fund. Bridging the valley of death, as I mentioned, is not just about understanding the science. It’s about being able to scale up operations and supply chain for the first time and creating a company centered on operational excellence. That is hard, and it’s something no other capital provider understands how to do. A fund can fill that gap by creating a team of infrastructure operators and supply chain advisors or a network of added services to our capital.
The CEO of Generate has a similar quote about this:
So if you're willing to do the work on a project-by-project basis, and you have the team in place to not only understand the technology risk or the financing risks, but you actually have the team to do the work to mitigate the risks—venture firms don't have that. Banks don't have that. This is work that either development companies, infrastructure operators, or engineering firms have to do. If you do that work, you can start mitigating these risks and then you can bring the financing along because you've mitigated those risks.
VC can’t fill the gap:
This isn’t unusual. The climate tech renaissance is not that old. It took 10 years from Cleantech 1.0 for the new climate tech wave to begin. You could argue the modern incarnation of climate tech investing has really only been around for 10 years, and if we’re being really intellectually honest, I think the proliferation of climate focused funds is about half as old. So it is natural that new capital will concentrate into the highest risk highest reward most flexible forms of capital that exist (early stage VC). When you start looking beyond climate specifically, it feels like industrial tech broadly has only become truly popular over the last 3-4 years, particularly in areas like aerospace and defense and robotics.
Interest rate hikes have changed the financing landscape, and we’re not going back to ZIRP until the next economic crisis.
During ZIRP 1) traditional venture was the better place to find returns, and 2) the Demonstration phase could actually be financed by venture dollars. It wasn’t uncommon for $100M+ rounds to happen at this time. So providing a credit facility with high warrant coverage for industrial tech infrastructure plays didn’t make economic sense. But now, it might be the only sensible option.
For climate tech specifically, there has been enough capital deployed into the space that now we’re at the stage where there are a good number of companies that have raised Series A or even Series B and need to raise $50-100M for the Development phase. They’re not in the Deployment Phase. They’re at the Commercial Development phase RIGHT when the market has dried up. I believe there will be many many more companies over the next 1-3 years that had a very similar funding time horizon to Bowery (maybe offset by a few years) that will be at the Commercial Development phase soon and VC won’t be there for them.
People are still underestimating this market and the general trends around reshoring and developing infrastructure in the US.
There’s no reason why this model couldn’t work for any industrial tech company. If you believe that there will be many more companies at the intersection of technology and industry in the coming years, then you believe the addressable market is way larger than stated above. There will be way more than 18,000 companies across all of these climate sectors in the next 30 years. We’re only at the beginning.
There’s not that much competition, but there isn’t zero competition.
Credit Focus
I don’t think you want to go into this space being the only one. There are a few companies I’ve seen that are doing this model, and they’re doing it very quietly.
Galway Capital is a pureplay credit fund $350M AUM, most of which came recently from a $265M investment from Oaktree. See below from their Impact Report. They target 15% returns, which feels like the right target for infrastructure.
Equity + Project Finance
Springlane Capital is viewed as a pioneer to the Growth equity + project finance approach. They’re based in Boston and have ~$400 AUM, which they raised in 2019 and 2021. I find their website interesting in that they seem to feel the need to advertise that the space they’re investing in is bigger than one would think, which I think speaks to the relative immaturity of this part of the capital stack. They lay out their approach here:
Vision Ridge Partners is another pioneer in this space, and likely the largest fund at ~$2.4Bn AUM, with their first fund raised in 2016 and their latest in January of 2023. I consider them a little more later stage and less focused on technologically risky areas, but their ability to deploy massive amounts of capital into infrastructure that is too early for traditional debt financing is clear. Their latest investment was a $1Bn round in Terawat Infrastructure, which they led in September 2022. Keyframe Capital (below) was also involved in this deal.
Keyframe Capital is also a prominent player. They are a fairly young firm, and were only founded in 2019. They have a diagram of how they invest (below), which indicates that they do both credit and equity, but they also move between the project level and the HoldCo level, so I view them as more similar to Springlane and Vision Ridge.
