Module Description:
In this session, Ubiquity Extended Team member Natalya Bailey shares her lessons from life as a founder/CEO of a deeptech startup (Accion Systems). She dives into the nuances of making deeptech problem solving work with traditional venture capital.
Full Transcript:
Welcome everyone to our Ubiquity University session on managing a deep tech startup. Ubiquity Ventures is a seed-stage venture capital firm, investing in startups that are moving software beyond the screen of computers into the real physical world, so often involves smart hardware and machine learning. Today, we're lucky to have Ubiquity extended team member, Natalya Bailey. She was previously founder and CEO of Axion Systems and has a PhD in aerospace from MIT. Today, she'll be talking to us more about managing a deep tech startup and some of the nuances involved. Thanks for being here, Natalya.
Thank you, Sunil. Great, so, today, like you just said, I'm gonna talk about solving hard problems, particularly with venture capital. So like you just mentioned, I founded and ran a space propulsion company building ion engines for about 10 years. Spent a year at a fuel cell and electrolyzer company, currently consulting in semiconductors and the energy space, And then have another five years investing in deep tech, and along the way, doing a lot of mentoring and work like that with these types of companies. And what I've realized throughout most of my experiences is that there just aren't a lot of resources out there to support these types of companies. So existing resources, like books and incubators and just people out there with knowledge, most of it's focused on more traditional startups, but there are a lot of really important problems to solve as humans that are of the deep tech variety, you know, like climate change, curing cancer, nuclear energy, space travel, I have to throw that one in there.
And so, I hope to, today, you know, point out some of the ways these companies are different and need their own playbooks so that entrepreneurs can solve these deep tech problems, hopefully, faster and more successfully going forward. So there are different definitions out there for deep tech, So I'm gonna start with what I mean. So by contrast, a traditional startup has attractable development path with, you know, little to no technological uncertainty. You're pretty certain it can be built, it's more a matter of execution and market risk, whereas a deep tech startup is usually trying to bring a new scientific idea or invention to market where there's still a lot of uncertainty that requires trial and error. And what the end solution will look like is still pretty unknown, and getting to market requires R&D.
So if you look at the, you know, life cycle, at least of the early stages of both of these types of startups, they generally follow, you know, roughly the same path. You identify a problem and make sure there's an exciting opportunity to go after, prototype the product and the business model and iterate on that with customers, release an MVP and demonstrate traction, iterate with customers again, go on to find product market fit and grow and expand. So here's really where, you know, the meat of the difference is. A deep tech startup has to do scientific R&D to get to its MVP, and by MVP, I mean an early product that can generate meaningful revenue in a beachhead market segment or two, and meaningful, you know, I can't really define in absolute terms here, it's kind of fluffy, but basically, I would say it demonstrates a strong signal that when you do broaden the product and its appeal a little bit more, you'll be able to find product market fit.
So this R&D phase, you know, you take an idea and hopefully, it's based on physics, but apparently that's not even always criteria, demonstrate the working principles, prototype the system, and then scale that prototype to the MVP. And then, when you broaden that product in its appeal, you find product market fit, and then you go on to introduce more products or business lines, and usually you need to go back to the beginning here of the R&D cycle and do that all over again. So I, you know, alluded to Theranos when I said, you know, maybe you don't even need to come up with an idea that is based on physics, but I also have Lawrence Livermore National Lab here. So, you know, they just recently demonstrated a fusion experiment and we'll come back to that in a minute. University spin outs often come out of lab into a startup having already demonstrated a prototype in lab, usually a bench top version of the invention, maybe it's operated in a more relevant environment than just the pure working principles had been demonstrated before. But really, between the prototype and the MVP is where a lot of the meat of being a deep tech startup exists. So usually you're having to scale some key parameters from the prototype to something the market will accept and pay money for.
So at Axion, at my ion engine startup, we spun out of university with about 10 hours of operating lifetime. And before we could even start selling to our very first market segment, we needed about 2,000 hours. So for us, that meat was scaling the operating lifetime. And for other deep tech companies, it might look different, but there's usually a big scaling component there. So back to comparing to a traditional startup. So both really face a lot of execution risks, so that's similar, but, you know, on the bottom here, deep tech startups have this existential product risk, the technology, the end solution is very uncertain, the timelines are uncertain, feasibility is uncertain. And then you'll also notice that I've crossed out some of this prototype phase, and that's because deep tech companies can't really usefully test their prototypes with customers. Trying to get the pulse of the market with a deep tech prototype just doesn't really work like with a wire frame or I don't actually know what other tools software companies use, but those are aimed at putting customers in the mindset of the pain that they experience today and the relief that they'll experience when they use your product. And with deep tech, that's just very difficult to do for a handful of reasons. But circling back to the fusion demonstration at Lawrence Livermore National Lab, if they were to, you know, hit the market with a prototype and try to get people to, I don't know, sign up for pre-orders or demonstrate customer traction somehow with that, the pitch might go something like, "I will deliver you 3.15 megajoules of energy and all I need as an input is 300 megajoules of energy. How does that sound? How much would you pay for that?" And that actually sounds pretty horrible.
So there's, you know, a ton of scaling work, like I mentioned, that has to happen before, before I think you can really even make significant headway with customers. So that's another difference there. And then, you know, just to kind of summarize the main differences as I see them, we've talked about, you know, the timelines, doing R&D, it's trial and error, the timelines are uncertain and essentially unknowable. And, you know, to put some numbers to that, to bring a new drug to market in pharma, I think the average time is almost eight years, and that's an industry that has really, you know, formalized the process for doing this. And, you know, and energy, I think the average time from an invention to a product based on that hitting the market is seven decades. And that is, you know, scary to to say in a talk where I'm trying to encourage more people to become deep tech entrepreneurs, but that one is a little bit hairier because, in order to get energy products to market, often, you're trying to navigate and even influence some pretty complex political and regulatory landscape, so it takes even longer. But my point is, is that it's a long road, and for those early days or years, as we're seeing, you're running an R&D organization, and so you need to structure the team differently, interact with customers differently, but these problems are worth it. So let's march on. Milestones and progress for deep tech ventures look different from traditional startups. In my experience, it takes a funding round or two longer to develop the MVP and get it into market. And it's very possible, given longer timelines and uncertainty, you're doing maybe four funding rounds where you're still doing R&D to scale your prototype to your MVP. Hopefully, you are faster and and luckier and it doesn't take as long. But then, as soon as your MVP is in sight, as in you're now doing real product development versus R&D, you then transition into doing some business development because you'll need to be moving your MVP into the hands of customers and showing traction to continue growing the business. So what's, you know, maybe even more important to highlight from this slide is what you're not doing, I don't have it here, but I'll speak to it.
So during these, you know, first several rounds, you probably shouldn't be doing things like, you know, building up quality and manufacturing necessarily, probably not hiring heavy hitters and business development and finance, trying to keep burn low and keep your options open because of the large amount of uncertainty, while also trying to move fast. So then, after your MVP is in the market and giving you signals and direction for your next more mainstream products, you can begin showing your business model is really viable, you know how or can learn to build a sales organization, scale manufacturing, that you have an appropriate cost structure and so on. After that phase, you leverage the product market that you found into growing the business even further and expanding geographically more products and so on. So this first stage of doing R&D to get the MVP out there, I've stressed that it's long, its period of R&D, and it's important to focus and not unnecessarily build up other parts of the business. But at the same time, you also don't want to bury your head in the sand and assume that, when you emerge with your MVP, you'll have a viable business with a lot of market appeal. So to that end, we should continually be validating key assumptions. So I'll walk through these. Likely, you have a research background in this field. Deep tech founders often come from academia or other research institutions, so from your background and essentially your intuition, you should be able to identify a handful of first applications and use cases you're excited about and that have the potential to grow in the future. With those in hand, since you're likely a scientist, use your first principle's thinking to understand your competition in these applications at a really fundamental level. So I got caught out on this, I assumed that a competing technology in the the space propulsion space, hall thrusters, were sold at like 12 to $15 million a piece because of some inherently really expensive parts and processes in the bill of materials. But after a couple years, SpaceX went on to vertically integrate the system into their line for on the order of $10,000 a piece. So I was orders of magnitude off. And so, I think the takeaway there is don't be lazy when you're understanding your competition. And then, of course, make sure you have a real value proposition. So usually in the form of dollar cost saved or dollar opportunity created for your customers. And I think the key here is to keep yourself from cherry picking the parameters that you use. You can check with industry advisors to make sure you're being objective. And all along the way, from the very earliest days, you really should be talking with customers. I don't know that I would spend too much time with the sales or, you know, acquisition folks right off the bat, beyond some learning and exploration, because you may start to annoy them over the seven decades of, just kidding, it's not gonna take that long. But you do need to understand how to sell into these organizations, the real cost they'd experience in switching to your product over whatever they currently do. And also be checking for genuine enthusiasm for your solution, you know, not just a lukewarm, polite response. And then, given what you learn from real humans, you know, iterate on the use cases, maybe even on your vision to make sure you're still excited by what you've found out. Okay, we'll pause there for today, I think.