Commercializing University Research with AlbionVC's Dave Grimm

Dave Grimm is a partner at AlbionVC, a London-based venture firm that backs early-stage, research-heavy deep tech spinouts.

Tell me a little about yourself and AlbionVC.

I'm a partner at AlbionVC, a venture fund manager in London, and I look after our deep tech investing. My background is maths and chemistry and then quantum physics at university, followed by a couple of years in management consultancy. After that I went to China for a few years to run a dairy startup, which got me into the world of startups. For the last 14 years I've been backing very early stage, research-heavy spinout companies.

AlbionVC has been around for 31 years. We have about £1 billion under management, and the majority of that sits in retail funds, which is fairly unusual. They're listed on the stock market, retail investors buy shares, and we use the capital to buy equity in startups. Those funds are evergreen. Alongside them we run early stage deep tech funds that work directly with research institutions, and I joined Albion 10 years ago to build out that part of the business. We go very early into universities and research labs, find interesting technology, and connect it with capital and expertise. There are about 20 of us who are investors and about 50 in total, covering software, digital healthcare, and deep tech.

AlbionVC tries to write the first check into companies that are still at the research stage. Why does Europe lag the U.S. at commercializing its research, and is that starting to change?

We typically want to be the first check, and the supportive one. One of the problems in Europe is that we don't do anywhere near as well as the U.S. West Coast at commercializing our research and creating value from it. Here's a stat to illustrate it. If you take the Nature Index, which measures the quality and quantity of research, London and the West Coast come out roughly similar, and University College London (UCL) and Imperial more or less rival the West Coast on output. Yet the number of unicorns created from that research is around 10 times higher on the West Coast.

There are lots of reasons, and some are starting to change. Our universities used to lack entrepreneurial cultures, but that is shifting massively. The year before last was the first year that more students leaving university wanted to start their own business than go into a traditional career path. You're also seeing far more academics who want to turn their work into something exciting. The universities are more supportive, and we have more capital willing to take risks than ever before.

There's still a knowledge gap around how you take something sitting at the bottom of a windowless lab and turn it into real value. A lot of the time those ventures get set up badly. The capital structure is wrong from the start, they take investment from the wrong people, they don't know what they're trying to build, or they bring in the wrong team. We've spent the last 10 years working in that space, and we've built companies from zero to unicorn scale, so we're happy to start right at the beginning with founders and take it all the way through. That hands-on approach has become our differentiator. We'll put real effort into things most venture firms would tell you to come back with once they're ready, because we think we can help get them ready.

How do you diligence a technology when it's still effectively sitting in a lab and the commercial potential isn't obvious?

We're somewhat reliant on the fact that we work with very good institutions and with academics who are usually top in their field. So we're not there trying to out-science them or find the holes in their work. We take a reasonable bet that the people and the science are good, and we'll always have other experts from the field look at it. That isn't our unique edge, though.

What we try to bring is a real focus on massive market outcomes. A lot of academic research is genuinely interesting, but its outcome tops out in the hundreds of millions, and that isn't suitable for venture. So we spend a lot of time thinking about how the world changes and about future constraints. 

Back when "Attention Is All You Need" was published in 2017 and the transformer architecture was starting to look like the way things would go, we asked ourselves what constraints would emerge if that played out. One of the biggest was memory. It looked clear that memory technologies wouldn't be good enough, and memory was the least fashionable area imaginable. You mentioned it back then and people would roll their eyes, because plenty of investors had lost money there. But we were convinced it would be a bottleneck, so we invested in a memory company.

For the first five years that business was a real slog, and now it's one of the hottest companies we have. That's the value of future-constraint thinking. If you're technical enough to spot the constraint that something like the transformer will create, you can act on it.

Having technical expertise on the team is what makes that possible. We have machine learning experts, neuroscientists, and synthetic biologists, and we even have a former brain surgeon who decided brain surgery was too dull and moved into venture. You need that depth, because when you can't speak someone's language on physics or machine learning, they don't trust you as quickly. For a long time the ecosystem hasn't had that expertise at the coalface, and instead business experts were trying to help startups they didn't fully understand. Deep tech is also a team sport. You don't want to be the only investor on the cap table when it might be five, six, or seven years to revenue.

There's skepticism about turning researchers into founders. The worry is that researchers like to keep exploring and asking questions rather than putting a stake in the ground and shipping a product. What's your take?

It's a fair criticism, and it is one of the failure modes for these businesses. So we look for a few archetypes. The first is the academic who came into academia from industry to solve a particular problem, solved it, and is ready to go. We have a business in the photonic networking space exactly like that. The founder was a professor who spent 15 years in academia doing it singularly to solve one problem, and once he solved it he was ready to leave and build. Those people are rare, and they don't suffer from the same issues. They might need a bit of coaching, but they're always well worth backing.

The second archetype is one Marc Andreessen described on a podcast years ago, and I've come to agree with it more and more. The best thing you can often find is a world-class professor running a research group, paired with a PhD student or postdoc in that group who has spent a few years in academia. They've not been there so long that they've picked up bad habits, and they're hungry, entrepreneurial, and technical enough to carry the work out. Sometimes we'll meet a group, think the work is cool, ask to get to know the wider team, spot someone, and ask whether they want to come and do something fun.

The third archetype brings in someone with domain experience, and it's genuinely complex. At such a low technology readiness level, it's hard to convince someone who has already built businesses to spend three or four years just getting something to the point where you're sure it's exciting. So you sometimes set that person up to lead the business through the early stages as an interim, with a shared understanding that you'll bring in a superstar CEO once it's ready. That requires people with low ego, and it's hard to do, but I've done it a few times with success. The point is that each situation is bespoke. It isn't cookie cutter. You look at the pieces you have and play them down the right pathway, and that pattern recognition is where we see our edge.

For any academic weighing this, my single piece of advice is to get out and spend time in the commercial domain you're targeting. Build that network intentionally by going to the industry conferences and meeting people, because that network is where you'll eventually sell, recruit, and find the partners who help you develop the technology. Speak to future customers early so you understand whether your technology actually solves their problem. If you stay too heavily focused on the academic world, you'll never get out of it. I've seen academic founders do this deliberately and have it work out very well.

How does university tech transfer in the UK compare with the U.S., where institutions often take a large cut and many universities are slow at it?

It's funny, because everyone in Europe looks at the U.S. and assumes you're much better at this than we are. The truth is that it's a handful of outliers, like MIT and Stanford, that do it really well, and plenty of people in the U.S. feel exactly the same way about them.

In the UK we now have much better terms, partly because of those examples. We worked closely with UCL for a few years and convinced them to adopt a flat 5 percent model for all software spinouts, which was a first. It took more than a year of conversations, endless committee meetings, and navigating university politics, but eventually they did it. The first business to spin out under that model was a medical AI imaging company. We put its pre-seed investment in, and it spun out far more quickly than it otherwise would have. It landed a grant it would have missed for being too slow getting out of the university, and then it exited without taking any more capital, making the university several million on a small stake. That result made the case for the model far better than any argument could, and it never got questioned again.

London universities are now genuinely good at this, with Imperial and UCL leading and Cambridge doing some of it too. One reason this took so long is that a lot of the UK's early success was in biotech, which is a very different model. Biotech universities take large chunks, royalty streams are well understood, and everyone is comfortable with it. Apply that to a software business and it doesn't work. Universities are now seeing better returns from the physical sciences, and that's forcing a healthy rethink about how they engage.

Deep tech is exploding in a lot of directions right now. Are there areas you think are overinvested, and areas you think are underinvested?

The easy answer is AI. Generative AI is dominating the investment scene right now in a mildly unhealthy way. Most of the value still looks likely to accrue to a very select number of companies, so it's going to be hard to find space in that zone. AI infrastructure is getting very hot, with a lot of money going into chips, memory, and photonics. Is that overinvested? Not if you believe we're going to build out the level of data infrastructure that's coming, and I think we probably will have to.

One of our businesses exited to Anthropic last year, and I catch up with the founders now and again. It's been useful to get that inside view, because it's all real. They're selling to customers who come back wanting more because they're seeing a return, and the business is heading toward profitability. Even setting aside whether we get to AGI, the change is significant enough today that a huge infrastructure buildout has to follow.

I'm more skeptical about space. There's so much risk in getting a technology working on the ground, and then you put it on top of a massive controlled explosion and send it up. It feels like compounding risk. SpaceX is about to go public, which will pull more people toward the sector, but I doubt we'll see huge value from earlier/smaller investments  in the near term. There are a handful of huge outliers and the rest honestly aren't that investable. It's a very capital-intensive category, so a few companies raise enough to do something real and a lot never raise enough to compete.

Defense tech has become very hot outside the U.S. The rest of the world has suddenly realized it might have to look after itself from now on, so a lot of money is flowing in. But your customers are governments, defense agencies, and primes, and those are complex customers to sell to. There will be a few good successes, but it may be an area that struggles.

Even companies that never set out to build for defense are getting pulled into dual-use territory. A quantum company suddenly has defense applications and the Ministry of Defence comes calling. Are you seeing that in your portfolio?

Yes. I worry about it slightly, because there's a real pull. The agencies say they need you, the company agrees to do a bit more in that direction, and then it turns out the agency is very slow and very demanding. They make a lot of noise about how badly it's needed, but whether they turn out to be a good customer is another question, and I find that unlikely in a lot of cases. So I'd always advise a startup to engage, for sure, but not to put all their eggs in that basket for what might be a fleeting spike of interest.

Europe is good at seeding companies, but they hit the growth stage and head to the Middle East or, more often, the U.S. Is that an accurate read, and what could keep these companies growing domestically?

It has definitely been true, but I think it's changing rapidly. The Saleup Europe 5 billion Euro fund just launched, the British Business Bank is doing a lot to put more money into growth capital, and more growth funds are appearing. So the solution is happening.

I'm also not too worried from a sovereignty perspective. You can take funding from other nations, as long as you don't end up with all of it coming from one. We have a photonics business that's interesting as a sovereignty play for the UK, and I don't think it weakens that argument to have some investors who aren't UK based. What matters is that there's enough capital that the company doesn't get pulled away and relocate its headquarters to San Francisco. 

Once a company is big and interesting enough to be in the growth-capital conversation, and it can generate competitive tension, a U.S. investor isn't going to make moving 50 or 100 people to the U.S. a condition of the deal. They want to invest because it's clearly working. So in general I think we're getting to a place where that problem is being solved.

How do you define deep tech?

For me it's something with a technical moat that buys it the time to get to market. Technical moats don't last forever. You might have some IP, but someone will eventually work around it or out-innovate you. The question is whether the thing is difficult enough, and whether there's enough complexity or protection, that you have the time to get to market and to revenue before someone else displaces you. So that's what I'm looking for: a technical moat that will last long enough, knowing that none of them last forever.

SPONSORED

Infinite Frontiers is made possible by HAUS, the leading public relations agency for deep tech startups and scaleups

Learn more