E443: Why the Next Trillion-Dollar Startup Won’t Be a Software Company

What if AI makes software abundant and shifts venture capital’s biggest opportunities into the physical world? Leo Polovets is a General Partner at Humba Ventures, an early-stage deep tech fund investing across robotics, defense, energy, manufacturing, and other physical industries. Leo explains why AI is commoditizing software, why hardware may capture more value as a result, and why some of the best opportunities are hiding in industries with almost no startup competition. We also discuss how to underwrite deep tech before revenue, what separates exceptional founders, the infrastructure bottlenecks created by AI, and why the biggest risk to deep tech may be its dependence on continued funding.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Building software is becoming more and more commoditized. More people can do it. You can do a lot more with fewer people. You can do it very quickly. Software was pretty scarce before, but now with AI it feels abundant. I think it just means more value will move to other areas. And hardware is like, to me, the next natural one. It's an opportunity for a lot of companies to basically like, build the only new entrant in a big space and then dominate the space. Versus in software where I think the competition is just like a lot more cutthroat and crowded.

• 0:26David Weisburd?

  It seems like capital suddenly combusted out of thin air to launch and fund these very large capital intensive projects. How would you explain that?

• 0:36Speaker 00

  All the like techs, talent tiers, SpaceX, Tesla alumni are starting new companies. There's just like a lot of amazing teams and a lot of signs that you can build these like hundred billion or even trillion dollar companies. I think those investors are really excited.

• 0:47David Weisburd?

  Leo, you co-founded SUSE Ventures 13 years ago. And today you're on the sister fund, Humba. Tell me about what you've changed your mind on as investor recently.

• 1:00Speaker 00

  The thing I've really grown to appreciate, especially over the last few years, is just how important it is to basically play in your own lane. And like to have a lane that's kind of uniquely yours. I think it's true both on the company side and the investor side. At SUSE and at Humba, we've always appreciated things like moats and defensibility. But what I realize is like, even if you have a good moat, if you're in a space that's very crowded, there's a lot of competition, even if you have the best product or the best fund strategy, it's just very hard to stand out. So I've just become more and more of a fan of companies doing unique things, like funds with very unique approaches. I think that's the best and most predictable way to win.

• 1:37David Weisburd?

  What's an example of a fund doing something unique?

• 1:41Speaker 00

  To pick one, just off the top of my head, I like Weekend Fund a lot and Ryan Hoover. I think he's one of the most like kind of experimental managers I've met in terms of he's always running experiments on like newsletters, products, you know, ideas for how to raise the fund, how to deploy the fund, how to engage the LP community. And I think all of those, like maybe not all of them work, but in aggregate, there's a lot of successful experiments there. And so I think that's just like very different than how most VC firms are run. And then I think when he talks to founders, they appreciate that he's really like a product guy running a VC firm versus a VC.

• 2:13David Weisburd?

  Oftentimes, a lot of these unique fund models or even unique businesses are downstream of funds or businesses that could actually get funded. So the harder it is to fund the strategy, it gets that much easier to deploy the capital.

• 2:32Speaker 00

  I agree with that. I think that's true again, like both on the funding company side, like the more, the more off the beaten path you are, the harder it is to convince someone else that this is a good investment because they're used to whatever, you know, pattern they see, which is usually like the norm and not the outlier. But then I think you have a better chance of being the outlier if you can actually get the funding.

• 2:52David Weisburd?

  When we last spoke, you said that quality is downstream of quantity when it comes to venture capital. What did you mean by that?

• 3:00Speaker 00

  It's very hard to pick outlier companies if you just don't take a lot of meetings. And I have friends and mentors that try to really scope down like, oh, I think I figured out what the pattern is. I just meet, you know, one company a week that matches this pattern. But I think you end up missing a lot and not being well calibrated. There's like an anecdote here that I like, which is, I remember reading maybe 10 or 20 years ago about this pottery class experiment where a professor basically told half the class, okay, you're going to be graded on like one piece of pottery you make, so just make it really good. And then the other class, the other half of the class, he basically said, make as many pieces as you can, and you're just graded on quantity. But that second group actually ended up having the better, more creative pieces because like through repetition and learning and iteration. And so I think there's a lot of that to venture. Like there's a lot of stuff that, you know, maybe it fits the pattern, so you should

• 3:45Speaker 00

  meet with them, but there's a lot of stuff that doesn't meet the pattern. You should still meet, you know, a good number of those because maybe it expands how you think, maybe you find companies that, you know, on paper, you weren't sure you'd invest in, but they end up being great investment opportunities. So I think just like quantity of meetings and companies that you engage with is really important.

• 4:04David Weisburd?

  Venture capitalists oftentimes talk about seeing what great looks like and being able to really calibrate yourself to what a truly great founder is. Once you calibrate yourself and once you know what is this truly exceptional founder, is that still the case? Do you still need quantity or is it really getting up to that speed?

• 4:25Speaker 00

  Well, first, I think it always helps because when you're at the very outlier level, like the number of examples you see is very small. Right? So if you're trying to calibrate on founders that could build like a hundred million dollar company, you probably meet a lot of those every year. If it's like a billion dollar company, maybe you meet like five or 10. And if it's a 10 billion dollar company, maybe just, there's one company a year you meet like that. And so if you could like double that number through more meetings, after a few years, you have a lot more data points on like, are there common traits between the founders that could build, you know, a billion dollar company versus 100 million or 10 billion versus 1 billion? So I think that part helps. I also think the calibration is across areas, right? So you can calibrate on founder quality. We can also calibrate on things like traction, for example, right?

• 5:08Speaker 00

  And for example, like if you weren't meeting a lot of AI companies a few years ago, maybe you were still thinking like, oh, you know, if you're doubling or tripling every year, it's great. But if you start meeting more AI companies and you see them going like 5X or 10X or 100X in the first year, it like changes what you think great is. And so I think it's important to just keep recalibrating as markets shift, just to meet more great founders and like see what's common and what's not. So I just think more reps is really good, even if you've been in venture for a long time.

• 5:35David Weisburd?

  Today, it's October 2026. What does it take to be a great company in terms of revenue and multiples?

• 5:43Speaker 00

  It's a hard question for me to answer, honestly, because at Humba, we mostly invest in deep tech companies. And so those, they have a very different revenue ramp than a software company. So on the AI side, we see, you know, it seems like AI companies that my peers invest in or fall in love with, they go from zero to like 10 million or 100 million in a year or two, which, you know, five or 10 years ago would have been pretty unbearable. On the deep tech side, the revenue ramp is very different. Usually it takes two, three, four, five years just to build your product. And maybe you have a little bit of like pilot revenue or government grants to keep you validate what you're doing. But you don't have real revenue until a few years in. But then once you get it, the increments of revenue end up being much faster because you're not selling, you know, 20 or 40,000 dollar pieces of software.

• 6:26Speaker 00

  You might be selling like a $5 million physical thing. Most of the companies I work with, like the, it's hard to like say what a good revenue ramp is because they're all pretty idiosyncratic. And the ones that make it through three, four, five years, a lot of times they're going through like 10s or hundreds of millions of dollars of revenue in the first couple of years after that building phase, but that building phase takes a while.

• 6:47David Weisburd?

  You have certain companies where it's not just, you don't, can't just vibe code it and then you have product market fit and then you're scaling. You, you have to build, it's hardware. Sometimes it's like a missile. There is no minimum viable missile, I guess, when it comes to defense.

• 7:00Speaker 00

  That's right.

• 7:01David Weisburd?

  We have many mutual friends, many of them are top venture capitalists. What do you think differentiates those top venture capitalists from, let's say, your average, down the mill venture capitalists?

• 7:13Speaker 00

  That calibration piece is a good one. In that the people that have been in the industry for 10, 20, 30 years, a lot of times they've had a little bit of exposure to, you know, like the Zuckerbergs or Travis's of the world, so they, they can see what like, you know, real outlier performance and leadership looks like. So I think that's, that's a big thing. The other thing you get with experience is you just build like a better and better network for diligence and support and deal flow. So I think that's also great. The other one, and again, this kind of relates to the quantity pieces, like just being open minded, right? So I think a lot of the great investors I see, they're not rigid in their frameworks, right? Like they have some sense of, I think this is what a good founder looks like, or this is what a good opportunity looks like. But if they see something really promising outside of that framework, they will still

• 7:58Speaker 00

  take a bet on it. And sometimes those bets look, you know, poor in retrospect, but a lot of times like the really great ones, like, like they start out like that where it's a little bit less conventional, a little bit less consensus, but then over time it turns out you were right. So I think that willingness to like be creative and take, you know, risks, especially on like kind of the perception risk and how it affects your career if you're wrong. I think that that's a good hallmark of like a really good VC.

• 8:23David Weisburd?

  Give me an example of investment you made that really didn't fit the criteria and ended up doing well.

• 8:30Speaker 00

  One company we invested in a few years ago is called The Lumber Manufacturing. And they build, they basically build like next generation sawmills.

• 8:39Speaker 00

  And it's pretty funny because almost every investor on the cap table thinks it's like their favorite or one of their favorite companies that they've met in the last five years. And most investors off the cap table are more like, this company makes lumber mills? Like, why is that a venture investment? That sounds like some kind of like PE or maybe like, you know, like traditional business play. But I think it's one where like the founders are extremely strong, the market is huge. And then in terms of like picking their own lane, this is a good example of, you know, in 14 years in venture now, like it's the only lumber mill company I've ever seen. And so they're just competing with incumbents from like 30, 50, 100 years ago, but they're bringing in, you know, modern automation and AI and like, you know, teams from like, or like team members from like Anduril and SpaceX and places like that.

• 9:24Speaker 00

  And I think it just gives them a huge advantage in a big market. But it's one that in the early days just kind of looks weird versus, you know, especially promising.

• 9:31David Weisburd?

  This episode is presented by Juniper Square, the operations partner for private markets.

• 9:38David Weisburd?

  The startup ecosystem and the venture capital ecosystem surrounding the startup ecosystem is changing now, it seems like every 90 days in a dramatic fashion. What do you see today that just wasn't the case three to six months ago?

• 9:55Speaker 00

  It's definitely changing slower than that on the deep tech side. On the software side, I think there's just more and more acceptance that, you know, building software is becoming more and more commoditized.

• 10:08Speaker 00

  And that more people can do it, you can do a lot more with fewer people, you can do it very quickly. I think like maybe six months ago or eight months ago, I think there was less consensus there because people were looking at the previous generation of coding models and seeing that they're like good for support, but they can't go build like an amazing product then to end. Now we're seeing more of the, you know, you give it a few prompts and you build an amazing product then to end. So I think it's making everyone rethink moats and defensibility and barriers to entry on the software side. On the deep tech side, again, I think we're starting to see a little bit of signs of that on things like circuit board design or, you know, like robot design programming. But I think it's still, yeah, it's probably still a few years or, you know, at least one year, probably a few years away from where software is.

• 10:52Speaker 00

  So I think, you know, maybe in the next couple of years, we'll have a slumber moment on the hardware side.

• 10:57David Weisburd?

  Why are you so bullish on hardware investing today?

• 11:00Speaker 00

  At a macro level, I think the fact that software was pretty scarce before, but now with AI it feels abundant. I think it just means more value will move to other areas and hardware is like, to me, the next natural one. And so that's sort of like the macro shift of like AI commoditizing a lot of software development. But I think there's also a lot of tailwinds on the structural side in terms of, you know, I think 10, 15 years ago, like hardware was on a bottleneck for a lot of industries. It was more like, hey, we need better software, we need better tools, we could be more productive. Now we're starting to hit a point where there's but there are many more physical bottlenecks. Right? So we want to bring manufacturing in the US, but there's not enough labor for it. So that's where like robotics and automation really helps. Or we want to build more data centers to, you know, give everyone more kind of, you know, genius brain power in their, on their cell phone.

• 11:45Speaker 00

  But you need more data centers and data centers are bottlenecked by, by things like energy generation. And so now you're starting to see all these bottlenecks that are more on the physical side. So I think a lot more opportunities are there now instead of on the, your software side. And that's why I'm excited about it.

• 12:00David Weisburd?

  What specific part of the hardware space are you most excited about and why?

• 12:07Speaker 00

  We like companies that are, you know, doing unique things in terms of startup competition. And the lumber manufacturer that I mentioned before being a good example where they might have a lot of incumbent competition from decades ago or centuries ago, but there's no other startups in the space. And I actually think a ton of hardware businesses are in that realm. Like almost any industry you think of, like there's, there's very little startup competition there. Like mining or, you know, power plants or like, you know, lumber manufacturing, manufacturing. I think all, all of those areas, like they're mostly, you know, owned by incumbents. And even though historically there hasn't been a lot of startup activity because those are hard businesses to build.

• 12:50Speaker 00

  Now I think the technology has gotten better and better. So you could do more with like a small team and less capital. There's more non-dilutive financing. So I think there's an opportunity for a lot of companies to basically like, build the only new entrant in a big space and then dominate the space. Versus in software where I think the, the competition is just like a lot more, you know, cutthroat and crowded. I would say like, I'm generally excited about all these areas like energy and mining and robotics where there's just like so much opportunity. So it's less about like one specific thing and more of just like, there's an abundance of, you know, big markets with like very little technology in them.

• 13:24David Weisburd?

  And you referenced earlier, it doesn't follow the same life cycle. So you're not just going from pre-product market fit to a million revenue to five million revenue. What's the life cycle like for hardware companies and how do you go about evaluating which hardware companies to invest in?

• 13:42Speaker 00

  I think the life cycle tends to be a lot more around a few years for development and iteration of the product. Partly because a lot of times products just, you know, they're complicated and take a while to build. Like, for example, we have a company that's building nuclear reactors. They're about three and a half years old. They're, they're sort of like 90% of the way there. They just showed a couple of months ago that they can sustain like a nuclear reaction. And they'll probably be selling in the next, you know, 12, 18, 24 months. So you kind of think of it as like, you know, about five years from zero to like having their units deployed in the field. And I think that, you know, that's a kind of unusually ambitious and, and complex technical project. That's a company called Antares. But a lot of these, like the lumber mill company, you know, they also take a few years to develop.

• 14:26Speaker 00

  Because again, you're combining multiple different areas, right? So it's like software engineering, hardware engineering, nuclear engineering, like maybe it's chemical engineering. All of these things have to build in parallel. Then you have to integrate them. You have to go through like regulatory bodies and certifications. And so it just takes a few years for most companies that are doing ambitious things on the hardware and physical side. But then once you're there, especially if you're like, you know, the first kind of new technological advance in an area, that's where sales usually go pretty quickly. And a lot of the ACVs, like the, the contract value years, the contract values here are pretty high. So they're often millions, sometimes tens of millions or even more. So it's kind of like, you know, three to five years to actually get your product out. But then every customer you get might be millions or tens of millions of dollars.

• 15:10Speaker 00

  So I think the overall time period to go from, you know, like start to public company is still seven, eight years for a lot of these businesses. But the first five look very different because you're just in R and D mode. You're growing the team, you're hitting technical milestones. Maybe you're getting some pilots or pilot validation. But then the next couple of years, I'm not sure if the screen bears or not, instead of this kind of like, you know, steady growth rate for software, it's more like 000 and then like a vertical line.

• 15:37David Weisburd?

  And in terms of nuclear, obviously there's demand for it. There's some regulatory risk, but the customer, the customer base is pretty much there. You just have to underwrite the scientific base. How does a company like Antares or another nuclear reactor go about fundraising through this mostly binary revenue ramp?

• 15:58Speaker 00

  What we've seen is for the first few stages of financing, investors are mostly looking at, you know, three areas. One is like team strength, right? So at precedency, that's the founders, but then towards series A and B, it's more like what kind of team are you recruiting? You know, is it like top folks from, you know, SpaceX and Anduril? Or is it like, you know, new grads that like don't know much about hardware and you're trying to train them? Like those are very different team profiles. It's technical milestones, right? So clearly moving at a fast speed towards building something hard. I think that was something I was really impressed with Antares where they went from design to kind of like components to subsystems to the whole system in about three years, which I think is very fast for a reactor.

• 16:39Speaker 00

  And the third one is market validation, but usually that's more like LOIs or like letters of intent. It might be pilot contracts or demos or, you know, like future customers that are investing in your company because they're excited about what you're doing and they've, you know, they've kind of seen what you have planned. So I think that's sort of the form of customer validation that people want to see at, you know, series A or B in a lot of these companies.

• 17:01David Weisburd?

  People used to say that hardware is hard, but also hardware is just really capital intensive. So you're going to be diluted significantly from the early stage to an IPO. Is that still the case today or have the goalposts changed?

• 17:16Speaker 00

  It's a good question and it's one I thought about a lot a couple of years ago when I was raising the, the second Humber Fund. Because it was one of the most common questions we got from LPs, you know, it is, is a deep tech like more capital intensive? Aren't that like the exits, you know, the exit profiles are worse. And I looked at a bunch of pitch book data on companies that had large exits and how much they had raised relative to the exit size. And what I actually found was like, whether you were building in hardware or bio or software, the amount raised was usually like proportionally about the same, like for any of those sectors relative to the exit size. So I don't remember the exact number, but I think it was like, if you had a billion dollar exit, you probably raised roughly like 200 million. And if it was life science, maybe it was 210. If it was deep tech, maybe it was like 190, but it was all like clustered in the same range.

• 18:01Speaker 00

  So I think the capital intensity is pretty similar. Going back to the revenue, like how the revenue shape looks different. I think the financing shape is a little bit different in that the, the physical companies usually need more financing upfront. Because they don't have revenue to offset burn. It like R and D takes longer. You need a bigger team for it. But then I think you need less financing afterwards because a lot of the most capital intensive companies, I think have actually been in software, right? It's like Airbnb or Uber because the product obviously takes some work to build. But then you spend so much money on like go to market and advertising and sales. And I think that is less of a case for deep tech. So in deep tech, you spend more upfront on the R and D side, but then I think the marketing and sales side is a lot more affordable.

• 18:45Speaker 00

  Cause you're not competing with a lot of firms and generally like, if you just, you're like, let's say I made a power plant that got you energy like for 20, you know, percent less. It's not that hard to go sell energy. That's like 20% cheaper to everyone. Like everyone's going to want it. It's just a matter of like, can you build it? So that's the more capital intensive part, but afterward I think it's less capital intensive. And then additionally, there's a lot more options for hardware companies as they're getting going that are outside of venture. There's venture debt, there's equipment financing, the government, defense department are, you know, giving more and more grants and, and contracts. So I think all of that stuff really helps as well.

• 19:20David Weisburd?

  It's an interesting point in terms of software is cheaper to start, much more expensive to scale. Hardware is expensive to start, much cheaper to scale because you, you have these large contracts. And oftentimes they're, they're more profitable because there's less competition. How do you factor that into how you invest? You invest in the earlier stage. Do you have different models for how much dilution you're going to take after a round? Or does it more or less just end up in the same place?

• 19:49Speaker 00

  I think it probably ends up being a little bit more dilutive than software in the long run, because those first rounds where you need capital, they tend to be a little bit more dilutive because you,

• 20:00Speaker 00

  like if you have a software company and around it and you have some revenue and around structure is not what you like, potentially you could just wait and try to grow organically or through like revenue financing or something and see if you can get a better valuation in six or 12 months. It's usually not an option for deep tech companies because they don't have enough revenue to bootstrap. So they don't, because they don't have that fallback, I think the first few rounds tend to be maybe like three or 5% dilution more each. And then again, hopefully, you know, once you have the product live and revenue is growing fast, the subsequent rounds can be very low dilution. But I would say, yeah, so we sort of model like, maybe there's like 10% more dilution over lifetime for a deep tech company versus software. But it's pretty comparable and still, like, still the goal is like, if you're finding five or 10 or $20 billion outcomes, that dilution doesn't matter.

• 20:45Speaker 00

  And also, if you're not finding them, also the dilution doesn't matter.

• 20:49David Weisburd?

  You either are investing in the next generation of great companies or you're not. Is it economically rational to invest into hardware startups or is it just something that's important?

• 20:58Speaker 00

  I definitely think it's rational. I'm biased because, you know, the whole fund is based around it. So I definitely don't want to make an economically rational fund. The exits here are going to be huge because these are large uncontested industries when it comes to kind of tech savvy competitors. I think the businesses often end up being a little bit more binary than software because the chance of failing is higher. Like, very few software companies failed because they couldn't build the thing they wanted to build. I think, you know, a decent number of hardware companies failed because they couldn't build the thing that they wanted to build well enough or, you know, with the funding they had. But because these are big spaces with less competition, I think you end up having a lot more 5, 10, 20 billion dollar outcomes or bigger versus a lot of like half billion, 1 billion, 2 billion dollar outcomes, which is, I think, more common on the software side.

• 21:42David Weisburd?

  Does that affect how you build your portfolios or your portfolio construction?

• 21:47Speaker 00

  A little bit. Because of the binary in us, I think we want, you know, somewhat more shots on goal. Right? Because, you know, if you think about like, let's say for software, like maybe like 10% of companies become unicorns, but or, you know, something like that, but like very few become decacorns. Like you probably a portfolio of like 20 is great because hopefully you hit two unicorns and then you're done. And that's a great fund. On the hardware side, if it's more like, you know, 3% of companies are unicorns, but most of those end up being 5 or 10 or 20 billion. Because once you get to that stage and there's no competition, you just keep growing. I think that's where you want more shots on goal because you want to hit one of those, but you just have to hit one or two of them out of maybe 30 or 40 companies to have like a really great return profile.

• 22:30Speaker 00

  So let's say a lot of venture funds will have like a 20 to 30 company portfolio per fund cycle. For us, we have for about 40.

• 22:40David Weisburd?

  And venture today is becoming more consensus. You have all these winners raising more and more capital. Is that happening in the hardware space?

• 22:46Speaker 00

  It is, but you know, I'd say like it actually feels similar to software in the sense that it's kind of bifurcated. Right? Where everyone talks about all the money in software, but it's really like for AI companies. And then if you're not an AI company, it's hard to raise. If you're like a more consensus AI company, you're just drowning in cash. So I think the hardware side is similar where, you know, if you're like ex SpaceX founding team, you're working on rockets or, you know, or like, you know, turbines or power plants, you get a lot of funding and you get it quickly for a long time before you have to like really show a lot of traction or progress. I think if you're working on something more off the beaten path, usually funding comes slower, but then hopefully, you know, by the time you're like out of the product development stage, you're like so far ahead that where you can catch up to, you can build something really huge.

• 23:32David Weisburd?

  When it comes to deep tech, it's obviously many different industries. What subsectors of deep tech are you most excited about? And what's the most underrated deep tech sector today?

• 23:43Speaker 00

  It's a good question. You know, I'd say like maybe actually building on the previous answer, I think, I think the most underrated areas are the ones where like, there's just like, it's more off the beaten path. Right? So again, anything with like ex SpaceX, Xanderall team or like humanoids, or there's a lot of these high areas where like investors just throw in a lot of money into a lot of different companies. I think it's the like lumber manufacturing or nuclear for defense like Antares or with these kinds of companies where they're sort of like one of one or one of two companies. And then because they're one of one or one of two, there's not a lot of comps to look at. Right? There's not a lot of like, you know, not a good framework for them. A lot of times where you can say like,

• 24:26Speaker 00

  hey, here's how like 20 other humanoid companies are valued. So like I could estimate the value of this one. So I think that's like probably the most underestimated area, these places where there's not a lot of startups and you're looking at the first or second one. And you have to decide, is this a venture scale market? Is this a good startup team? Are there going to be better teams going after this ecology unique? I think that's the place where the, you know, there's less funding upfront. And then once it's obvious, there's a lot of funding, but then it's, you know, usually those prices are less attractive to investors.

• 24:52David Weisburd?

  One thing I've learned from talking to hundreds of investors is that great investment firms aren't built on investment returns alone. The firms under are great at the things most people don't see, their operations, their relationship with LPs and the quality of information they use to make decisions. And here's what AI has changed. Every firm now has access to the very same models. So the intelligence isn't the edge anymore. The edge is what you could feed it. A firm with its fund operations and data in one connected record can actually put AI to work. A firm running on disconnected systems simply can't. That's why thousands of GPs run their funds on Juniper Square. Juniper Square puts your fund operations, data and administration together in one connected record.

• 25:33David Weisburd?

  That means less time managing disconnected systems and more time investing, working with LPs and building your firm. This episode is brought to you by Juniper Square, the operations partner for private market GPs. Learn more at junipersquare.com/ how I invest. That's junipersquare.com/ how I invest.

• 25:53David Weisburd?

  Everyone I talked to on the show is chasing the same thing, an edge. And more and more, the edge comes down to your information, not just having it, but being able to trust it when the stakes are highest. AI is doing more of that information gathering for you every day and most tools are very good at sounding right. The summary reads clean, but can you trace it back to filing the transcript, the specific passage that drove the answer? Or are you just trusting the confidence of the output? For investors, that's not a minor concern. And this filing and misweighted source, context that got lost somewhere in the retrieval chain, those aren't edge cases, they're how decisions go wrong. AlphaSense is the AI market intelligence platform built specifically for this.

• 26:32David Weisburd?

  They own the content, over 500 million curated documents from broker research and expert transcripts to filings and earnings calls. And they own the retrieval layer on top of that. So every answer links back to an exact, verifiable source. Because the answer is only as good as what's underneath it. And with AlphaSense, you know exactly what that is. The edge goes to whoever can trust their information and prove it. See it for yourself. Start your free trial at alpha-sense.com/ how I invest. That's alpha-sense.com/ how I invest. That's the joke about venture capital is that's a herd mentality that no one's doing their own diligence. How do you go about building that skill set in a fund where you're diligencing these very

• 27:14David Weisburd?

  unique industries like lumber or deep tech or autonomous robots? How do you gain that skill set to diligence across sectors?

• 27:23Speaker 00

  Especially in the last couple of years, like AI helps a lot. Because, you know, these days it doesn't hallucinate that much. It understands a lot of like, you know, technologies and business types. And I think it was actually like a very hard thing to diligence a few years ago. If you're looking at technology, because you'd see someone that's like, oh, I'm using, you know, whatever cobalt based propulsion for satellites. And so you can kind of search for that keyword on Google and see like, is anyone else doing it? But you can't see like, well, instead of cobalt, is someone using like other chemicals or other types of propulsion? It's like very hard to find, like, what do you even search for to understand like the competitive landscape or the other technologies that are in play? But with like a chat GPT or Claude, it's very easy to do that, right? Because you could be like, you know,

• 28:06Speaker 00

  explain this technology to me, what are the drawbacks, like what else is out there? And it has like a very good semantic understanding of the landscape. So AI helps a lot. Like I said, I think calibration helps a lot in reps, just because the more companies we talk to that are in some, you know, broad area, the better our questions get over time because we kind of see maybe where the companies got stuck, what are the hardest things about a certain type of business? And then expert diligence helps a little bit, but honestly, I would say like with AI, you know, tools getting better and better, like we rely less and less on expert diligence.

• 28:38David Weisburd?

  How much of a signal is the team, especially post-founders, the team that's, that's joining the startup?

• 28:47Speaker 00

  I think it's really important. You know, I'd say for me, it's like akin to, there's some specialized areas in software. Like let's say healthcare or fintech where you don't necessarily need experience in those areas to build a big company there. But it definitely helps a lot because, for example, like navigating the FDA or like the hospital buyer landscape. Like you could learn those things on the fly, but it's just very hard if you're touching them for the first time. So there, like investors love founder market fit where it's like, oh, you've already worked on a few healthcare businesses. So you know how it works. You know, like the sales cycles are longer and here's how you like sell to a hospital or, you know, an insurer. So deep tech's a lot like that where there's a lot of areas where you have, you know, regulatory bodies involved.

• 29:30Speaker 00

  Or, you know, maybe like the manufacturing or like the design manufacturing time are really long. And ideally the founder has experience with those areas, like, or the founding team, not necessarily the single founder across both the technical front and the business front, so that you know what you're getting into, you can plan accordingly, you kind of know, you know, here's how the timeline will look, here are the kind of hires I need to make. So I think founder market fit ends up being more important on the, the physical side. Just because the industry is a lot more complex and the products are a lot more complex.

• 30:00David Weisburd?

  Obviously, every vertical is different. Nuclear is different than robotics versus space. But is there some different flavor of founder archetype? Or some specific patterns across deep tech founders that you find that just isn't something that you need as a software founder?

• 30:19Speaker 00

  Fluency across all areas of a company is important.

• 30:24Speaker 00

  And what I mean by that is like, on the software side, a lot of times we'd see sort of like, you know, a business founder and a tech founder. And a lot of times, like, the business founder might not know a lot about the tech side, like, the nitty gritty of it, because it's not that important to them. And, you know, in terms of like, them being able to sell the product. And then on the flip side, the tech founder often like, focus on the tech, they, you know, like, they're not like, in the weeds on the sales side, and like, understanding like, what customers want, you know, beyond like, maybe what the CEO passes on to them or the head of sales. And the deep tech side, a lot of things are much more intertwined and have like, very large ripple effects across the whole business. And so, like, I'll give you an example, like, if you have a software product and someone's like, can you build a new integration? I think a salesperson might have a good intuition of like, yeah, that's easy to build. We've done a lot of times before.

• 31:10Speaker 00

  We can price it at this much, like, do you want it? On the hardware side, if, you know, let's say you make robot arms and some customers like, hey, can you make the robot arm like, an inch longer? Right? Like, it has all these ripple effects, like, it's changed the software, you have to change, maybe like, the computer vision that the robot's trained on. Now it changes like, how much the robot can lift, because the arm is longer. And so there's all these ripple effects. And so, even on the business side, like, if you're talking to a customer, you have to understand the product, you have to understand it pretty deeply in terms of like, what are the trade offs? If you change something, like, what's possible, what's not possible? So because of that, we've been really drawn to founders where, and no matter what you ask them about, whether it's, you know, the actuators in the robot, or the regulatory requirements, or like, what customers need, they just have like, a very good fluent understanding of that.

• 31:54Speaker 00

  I've never met Elon, but I think that's the sense like, everyone has of Elon, right? Like, you could ask him about any of his businesses at any level of detail, and he could just like, talk at length about, you know, the decisions they made, the trade offs, like, what it applies for the rest of the business. So I think that that's a lot more important on the physical side of the world.

• 32:09David Weisburd?

  So, within deep tech, you're looking for CEO that has technical knowledge. Are there cases where they may not have technical knowledge and they might have the right team around them? Or is this kind of the singular architect that does the best in space? If you run a business, you probably start your morning with way too many tabs open. You're checking revenue, ad performance, email, inventory, spreadsheets. And before you've made a single decision, you've already spent your morning just figuring out what happened yesterday. That's what I like about Perplexity Computer. Instead of jumping between every dashboard yourself, Perplexity Computer could connect to the tools you already use and build a command center for your business. It could pull together things like revenue, ad spend, margins, inventory, and email

• 32:51David Weisburd?

  performance, then show you what changed, where you might be losing money, and what you should look at next. But what makes Perplexity Computer really interesting is that it can actually work on projects. It could help rewrite an email flow, build an ad campaign, or draft outbound outreach. And unlike most AI tools, you're not locked in into one model. Perplexity Computer orchestrates leading AI models behind the scenes and automatically chooses the right one for each step. For a limited time only, you could try Perplexity Computer for free. Head to pplx.ai/invest. That's pplx.ai/invest to try Perplexity Computer today. Most AI tools are still answering questions while Perplexity Computer is completing full projects. One thing that I've learned from talking to hundreds of investors, founders, and

• 33:35David Weisburd?

  business leaders is that planning for the future isn't just about growing assets, it's also about protecting the people who matter most. Life comes with a lot of important milestones. Maybe you're buying a home, starting a family, planning for college, or simply thinking more intentionally about your financial future. As those responsibilities grow, it's worth making sure you protect those you love. Not sure what kind of life insurance you may need? Concerned about how much it may cost? Think life insurance is too complicated? Your local State Farm agent can help answer those questions. They can help you explore your options for life insurance and choose coverage that you can adjust as your family's needs change over time. Financial protection for your loved ones when it matters most. Learn more about life insurance at statefarm.com. Like a good neighbor, State Farm is there. Support for today's episode comes from Square, the all-in-one way for business owners to

• 34:20David Weisburd?

  take payments, book appointments, manage staff, and keep everything running in one place. Whether you're selling lattes, cutting hair, running a boutique, or managing a service business, Square helps you run your business without running yourself into the ground. I was actually thinking about this the other day when I stopped by a local cafe here. They use Square and everything just works. Checkout is fast, receipts are instant, and sometimes I even get loyalty rewards automatically. There's something about businesses that use Square. They just feel more put together. The experience is smoother for them, and it's smoother for me as a customer. Square makes it easy to sell wherever your customers are, in-store, online, on your phone, or even at pop-ups, and everything stays synced in real time. You could track sales, manage inventory, book appointments, and see reports instantly, whether you're in your shop or on the go.

• 35:05David Weisburd?

  And when you make a sale, you don't have to wait days to get paid. Square gives you fast access to your earnings through Square Checking. They also have built-in tools like loyalty and marketing so your best customers keep coming back. And right now, you could get up to $200 off Square hardware when you sign up at square.com/go/HowIInvest. That's S-Q-U-A-R-E.com/go/HowIInvest. With Square, you get all the tools to run your business with none of the contracts or complexity. Run your business smarter with Square. Get started today.

• 35:36Speaker 00

  I would say it's not, uh, it's not required, it's just like, I think it's a very strong signal when a founder has that kind of depth.

• 35:46Speaker 00

  A lot of times, we do see founders spanning multiple areas, but they will have like, experts around them that are, you know, especially deep in those areas. Right? If you look at Antares, for example, I think Jordan, the CEO, has a really unique background. It was one of the things that got us really excited when we met. Because you think about Antares, they're building nuclear reactors, mostly focusing on like, defense and space applications. And so if you zoom backwards, you're like, or you zoom out, you're like, well, what are good traits for a founder like that? Like, well, one, ideally, they understand, you know, nuclear engineering a little bit, and like, know what they're getting into. Ideally, they have some government or defense experience, because you're going to sell into that sector, and it's not an easy sector to sell into. And then also, a lot of like, nuclear folks are maybe coming more from like, older industries or like, older nuclear companies.

• 36:29Speaker 00

  So ideally, there's some like, startup mentality to the founder, where they can move fast, they understand, like, you know, you could do a lot with a small team and a little bit of capital. I think that's important, too, for a business like this. So Jordan's background is, he'd co-founded a company as a CTO that had raised tens of millions of dollars. So I think he'd kind of seen startups and startups seed the speed and execution. He had minored in physics in college and been thinking about nuclear for a while, so he was very technical. And then he actually worked in the White House for a couple of years as well, so he understood how the government worked and how procurement worked. And so it was like, very unusual to see like, all three of those things in a single founder. If we saw them across a team of two or three, that would still be good, but I think, you know, I think the fact that Jordan combines all of these things into one person, and

• 37:10Speaker 00

  he understands all of them himself, I think that's part of what's made the company, you know, reach the success they have and the progress they have in just like, three, four years.

• 37:19David Weisburd?

  This is a pattern that I keep seeing time and time again. I've had probably half a dozen people that have worked with Elon, to your point. He has this really top-down understanding, and he probably has it more extremely than everybody. He has it down on the product level. But even Steve Jobs, who is not technical in terms of hardware, I had Steven Rosenblatt, who worked with him for many years after being acquired by Apple through Qualcomm. And he talks about how Steve Jobs had this granular understanding of everything going on in the company, even down to the legal structure, to the advertising structure. He knew exactly all the ads, he signed off on some of those ads.

• 38:00David Weisburd?

  So there's something to this, I guess, like, founder mode of running a company that doesn't just fit on to the physicists running the sales, but also just, it's pervasive across the startup ecosystem among top founders.

• 38:14Speaker 00

  That's a great example. I think part of it is like, on the software side, usually they're sort of like, a software team and a go-to-market team. And people kind of talk about like, well, the more the team grows, the more you have these like, dependencies and connections where, you know, if you have a team of 10 people, for example, it's like, when you do something, you have to go update nine other people, it just creates a lot of, you know, slowdown and friction and, you know, potential miscommunication. I think on a deep tech company, it's often magnified, right? Because you have like, you know, the sales kind of go-to-market side, and that's still like, you know, one side. But the technical side, instead of a software team, now it's like, well, we have like, a nuclear physicist and a mechanical engineer, and electrical engineer, and a software engineer. And each of those people has to communicate with the other ones on what's going on on

• 38:58Speaker 00

  their side or their bottleneck, so they need changes from somebody else to, you know, to enable them to do something. But now, you have all these nodes and they're all like, cross-connected, and I think, again, there's a lot of opportunities for like, miscommunication, inefficiency, friction. Again, going back to the example of Jordan, like, if you have like, one person understands all of those things, even if it's, you know, maybe not to like, 100% depth, but it's like, 80% or 90% depth, like, you just cut down on so much of the, you know, that inefficiency and miscommunication, right? Because it's not like, he talks to a nuclear engineer and then he tries to pass it on to a software engineer, and he's not really sure what's going on. And things like, because he understands all of these things, it just speeds up the company's progress a lot to have, you know, to reduce all of those like, like, edges between all the nodes.

• 39:42David Weisburd?

  What other traits would you say that top 0.1% founders possess?

• 39:47Speaker 00

  Bias, fraction, execution are really important on the physical side because there's so many things that take a long time, right? Like if you, if you're scaling, like if you're building a software business, like you build it, you put on AWS and it like kind of scales automatically for you. On, on the deep tech side, there's a lot of interdependencies and a lot of them have a lot like long time constraints. And so like to give you an example, like let's say you need, you know, two years for regulatory approval. If you like, don't realize that and don't start working on it, you know, today, and maybe you start working on it when your product's done in two years, now you have to wait two years to launch because like you didn't start it in time. But there's like regulatory, there's permitting, sometimes there's like, you have to get, buy some piece of equipment, but it's got like a seven month wait time.

• 40:32Speaker 00

  Right? And so if you didn't realize that early on, like again, you just spend all your time like waiting for something. And so I think founders that manage that kind of, you know, very complex timeline well, like there's all these things we depend on. Some of them are fast, some of them are slow, some are expensive, some are not managing all of that, like the operational execution ability is like, I think very, very important. The other part is like, because you're working with, you know, older industries, I think being able to, you know, kind of translate startups to like, you know, what a customer might want that doesn't work with startups, I think is also especially important.

• 41:06David Weisburd?

  What are you personally most excited about in the deep tech space?

• 41:11Speaker 00

  I've been pretty excited about the data center build out in terms of, you know, data centers have deep pockets. All they care about is speed. Because they're just bottlenecked by so many things like permitting and design and energy generation. I think that's actually a really great opportunity for startups, especially on the energy side. Because there's this kind of window in time where you can go compete against like Siemens or some, you know, big like power plant company. Because they might have something that's like very efficient, but it's custom and it takes five years to build. And if you make something that's 80% as efficient, but you can get it out in six months, like data centers will line up to buy it from you because they just care about time to, you know, time to go in live. So I think that's an area I've just been really excited about that, you know, we have this huge build outs for like one of the biggest build outs we've had since the industrial revolution.

• 41:55Speaker 00

  But it's bottlenecked by a lot of things that the incumbents still do well because they're just, they're not fast enough. So I think that's a great startup opportunity.

• 42:01David Weisburd?

  What are some companies in that space that you're very excited about?

• 42:06Speaker 00

  Maybe to pick a couple that we've invested in. So we invested in a company called Critical Energy. They make geothermal power plants. So the drilling for geothermal has gotten better and better with companies like Furbo over the last few years. But then you still need a power plant to take like the heat from the bottom of like from down in the earth and actually convert it into power. And that part would take a couple of years because you built a custom power plant. Critical Energy is trying to get on the order of like, you know, a month or two. So I think that that would be like a pretty game changing thing where you could be a data center, you could have like, you know, like fast, renewable, clean energy within a couple of months, sort of a couple of years. We invested in another company called ARN Industries that's doing data center design.

• 42:49Speaker 00

  So this is another place where, let's say you want to go build like 100 megawatt data center. What you have to do is you have to engage with an EPC firm, which is like, I think it stands for engineering, procurement, and construction. It's almost like a general contractor for large mega projects, but they have like a huge staff of dozens of people doing like the wiring design, the HVAC design, like how to soundproof your data center. So does that, you know, affect like the local residents too much? And it takes them, you know, a year, they charge you tens of millions of dollars for it. And our own industries is kind of rethinking like, well, what if you can use AI and software to bring that year down to, you know, a few weeks or maybe a couple of months?

• 43:27Speaker 00

  And again, that's one of those things where data centers just care about like, how fast can they get in the ground and building? And so this is one of those things where, you know, maybe like, nobody thinks about data center design, but it's like a big part of the process and it takes a lot of time. And so if you can accelerate that, like we can, you know, build like AI infrastructure a lot faster.

• 43:44David Weisburd?

  What do you see as the bottlenecks for AI from a physical infrastructure side over the next couple of years? And how do you see that evolving?

• 43:51Speaker 00

  I think energy is probably the biggest one, just in terms of, you know, the time to get energy to, especially if you're building somewhere where like, there's not a, you know, easy access to the energy grid. Or the energy grid doesn't have enough capacity to service the data center. So I think that's a big one. I think design's a big one. We've seen that as these data centers get big, like maintenance becomes pretty challenging. So some of that is like part of your rack crash. Like what's the problem? Is it, is it a chip? Is it some wire? Is it the software? Like, you know, I think it's hard to diagnose. I think we need better tools there. I think also like more automation of actually like laboring in the data center for maintenance is really helpful. Because a lot of times you have these racks, you have like tons and tons of wires, you have to like move one to another one. It's pretty error prone.

• 44:36Speaker 00

  Right? So if you have like a robot do it, that can be like, on the one hand faster, but more importantly, maybe like more accurate. Right? So you don't cause problems. The cooling is like always a challenge. So there's always like innovation to be done there, especially as like chips need more and more energy. So I think it's just like kind of across the stack, like almost any area there, there's some opportunity for like better technological innovation.

• 44:59David Weisburd?

  Tell me about the semiconductor space. For over a decade, there was really no innovation. Nvidia, now you have Etch, you have Cerebras. You have reportedly Anthropic building their own chips and probably OpenAI as well. How do you look at that part of the ecosystem and what excites you about it?

• 45:21Speaker 00

  What's exciting is there's, this is true across a lot of areas of deep tech where there's some structural shift in the market. Sometimes it's AI like driven, sometimes it may be like tension with China or our desire to like manufacture more things at home. But there's a lot of these like big tailwinds or big points of time opportunities. Data center energy demands another one. On the chip side, I think the big opportunity is you have LLM models and they run much faster on GPUs than CPUs. But GPUs were not built to run LLMs. They just happen to be very good at them because of the things GPUs are good at. But these companies like Etch, they basically make more trade offs in the direction of like, hey, if we really hyper specialize a chip just for LLMs, what can we, like, what kind of other performance gains or energy gains can we get out of that?

• 46:05Speaker 00

  And so I think that's generally like the exciting opportunity of, you know, if you specialize chips more, this use case of LLM inference is now like huge enough to, you know, to warrant like a specialized chip where it's not going to be able to do all of the things an Nvidia GPU can do, but the things it can do, it's going to be 10 or 100 times faster. And I think that's the bet that like Etch and its investors are making. But I think we're going to see another categories too, right? Like you're going to have things like you have humanoids and they have specific needs for power consumption and throughput and the kinds of data they can process quickly. Like maybe it's video data specifically. And there'll be companies that build chips specialized for that where they're not, they're not, they're going to be too specialized to work in other instances usefully.

• 46:50Speaker 00

  But for that one instance, like let's say it's humanoids, or maybe it's like, you know, autonomous vehicles. If you make a specialized chip and the market's big enough, you can have like a really interesting opportunity where you can beat out some of these big incumbents like an Nvidia or AMD. But it's still a huge market because like it's a new market that's basically like being created right now.

• 47:08David Weisburd?

  That's really skating to the park. LLM, the inference, which is basically the prompts you put into the AI. It's so much of the token usage versus actually training the actual model that now you have this massive market for a semiconductor chip like Etch that could go after that. I know you've spent a lot of time on humanoids. Today it seems like there's the Tesla, Optimus, and then also Figure with Brett Adcox's company. What do you see as the future of the humanoid market and how is that evolving as an industry?

• 47:41Speaker 00

  I think humanoids are like a really hard like category to invest in. So like, we haven't done investments in this space. And I think, you know, we might like, I'll never say never, but like, I feel like we're pretty unlikely to. Because it's very well funded. You've got, like you said, you've got Figure, you've got Elon working on Tesla. You've got probably another couple dozen companies that are raised hundreds of millions or low billions each. So it just feels like a, you know, pretty competitive space. There's a lot of funding. I think the chance that, and you know, a new startup raising five or $10 million can incredibly go after and beat everyone else is like, it's not zero, but it's low. But also they, you know, over 10 or 20 years, there's a good chance that like humanoids might be like one of the, you know,

• 48:21Speaker 00

  new like trillion dollar categories, right? Because I think what they're not good at so far is like efficiency. Meaning, you know, if you have a specific task, right? Like, let's say you want to like, you know, put like a bolt on a nut over and over and over. Or you want to like, you know, pack boxes on a pallet. I think things like robot arms that are specialized and trained for just that one task are going to be way more efficient. But if you have, if you have a, you know, a market like, or are you a target where you have a lot of different problems to work on? That's where a humanoid and its flexibility are just like invaluable. And so for example, like, let's say you want like a robot housekeeper. You wouldn't want one robot arm at your, your sink to do dishes, another one, you know, to like the, like odd wheels to do vacuuming and like another one to make your bed.

• 49:06Speaker 00

  You just want like one robot that can do everything. And I think that's where humanoids will shine because they're very general purpose. And so I think in scenarios like that, where you have not one or two tasks where you want to make them really efficient. Where you have like 50 or 100 or 10,000 tasks that are very flexible and you just need fluidity. I think that's where humans do well, or humanoids will do well. And I think that's where like, there'll be some company that, you know, puts humanoids in the home that does really well, you know, in the next three, five, 10 years. And whatever company dominates that space, I think that could easily be like, you know, a trillion dollar company. So it's sort of like a space that I'm reluctant to invest in because it's already so well funded and competitive. But I also think there will probably be like one or two huge winners in that space in the next 10, 20 years.

• 49:50David Weisburd?

  How do you look at, there's these trillion dollar markets, you mentioned humanoids, but even AI and AI chips, and there's these derivative markets that form as a result of these new markets. What are some of these derivative markets that are coming from these new industries that you're most excited about?

• 50:08Speaker 00

  It's a tough question because like, there are a lot of spaces where I think there's a few big companies that will have a lot of demand for a lot of these downstream components. But then I also think increasingly, a lot of these big companies are trying to like verticalize as much as they can, right? So you have like SpaceX, for example, you know, could be a big customer for almost any kind of space infrastructure, or it's like, maybe they, maybe they can like buy the propulsion systems from somebody else, or their navigation systems from somewhere else, or software. But then I'm just like trying to do as much as they can in-house. And I'm, you know, it's been hard to find areas where like they would be a customer versus like a lot of space startups, they're basically like selling to anyone that's not SpaceX, right?

• 50:51Speaker 00

  And I think humanoids might be similar, right? Where there's a lot of interest in better motors, better actuators, you know, better like vision systems and sensors. And it's not clear to me where some of those humanoid companies would stop development and like buy from a third party versus, you know, maybe they buy from a third party in the early days because they can't do everything in year one or two or three. But let's say like Figure or Optimus takes off and they have, you know, a million robots in the field or 10 million robots in the field in 10 years. Are they still going to buy their actuators from a third party or are they going to like develop their own? I think maybe they'll develop their own. So to be honest, like I've kind of struggled with a lot of the businesses that focus more on components or infrastructure for like the big players. Just I think the big players are doing more and more verticalization.

• 51:36Speaker 00

  So, you know, outside of things like chips where I think it's just so hard to build that most companies just don't want to touch that, right? Or I think mostly we'd focus more on like the companies building the whole thing versus the company that's building like a component for the big companies.

• 51:52David Weisburd?

  I'm curious because in software now, the cost of starting has gone down so much. You have these seat strapped businesses. Henry Shi, previous guest, keeps track of these companies that have a couple of employees and have hundreds of millions in revenue. On the hardware side, obviously today it's still very expensive to start these startups, but are there certain parts of the hardware industry where you see one to two person teams being able to create some components in order to be a supplier to some of these large companies and maybe get aquihired? Or is it just too expensive to start these companies?

• 52:28Speaker 00

  It's a good question. I think there's some categories that are easier to start in, depending on what your product is. I would say manufacturing, especially if you're doing manufacturing as a service is one area that comes to mind, right? So, you know, there's companies like Sen Cut Sen that, you know, basically like buy a lot of existing equipment. This is my understanding of them, I'm not an investor. They have some software on top, they have like efficiency gains on top, but like, I think they've largely bootstrapped, you know, I think they raised some money recently, but they were able to grow off of revenue because you buy some equipment, it's expensive, but maybe you can get it financed. And then as you, and then you start telling people like, well, I can make this kind of part for you.

• 53:11Speaker 00

  And as you buy more and more equipment, you can make more and more types of parts, but you sell those parts at a profit, right? And you could start selling them basically on, you know, as soon as you have the equipment in place. And so I think both for, for companies that buy existing equipment, like existing CNC machines or mills or things like that, or even companies that develop their own, like let's say you're developing your own 3D printer, like you can start selling parts as a service and there I think the revenue can scale, you know, pretty quickly and pretty well. Probably not with a one or two person company, but maybe with like a 10 or 20 person company instead of hundreds. The other thing we've seen is like, there's categories where you might have, there's a hardware component, but the hardware is mostly off the shelf, right? For example, you, you have some kind of camera, maybe it's mounted on a truck or like a ship

• 53:55Speaker 00

  or something and you're capturing some data and you get value from that data and you basically charge for the value you get from that data. And so there, you could often start with a pretty small team where you're buying hardware off the shelf. And as long as you can get customers to like put your camera on their, on their system, you could start selling them the insights pretty quickly. So that's another place where I think you could do pretty well with a smaller team. But a lot of the big complex projects, I think it's, it's hard to get anywhere until you're like, choose to be funded and like have a bit of a larger team.

• 54:23David Weisburd?

  Goes back to what you were talking about, which is it takes so long to invest in so many rounds of funding before you even have something that you could sell, before somebody will buy the complete product that you've developed.

• 54:35Speaker 00

  Yeah. Yeah, like, like to take the nuclear reactor example, there's not really an MVP for a nuclear reactor that's like not building the nuclear reactor, right? And that just takes a certain number of millions of dollars and like, you know, teams of like a wide ranging, like, you know, set of expertise and time, and you just can't fast track that.

• 54:53David Weisburd?

  It seems like capital suddenly combusted out of thin air to launch and fund these very large capital intensive projects. How would you explain that?

• 55:05Speaker 00

  I think there's a couple of components here. So one is, I do think people recognize that a lot of these areas are like just became bottlenecks, right? And, you know, like for example, like chips for data centers or, you know, energy generation or like all of these things where, you know, for example, like even outside of data centers, like I think the energy grid's been struggling a lot for the last few years with disasters and resilience. It's like people just see the need a lot more acutely versus, you know, I think like 10 or 15 years ago, it felt like, oh, this would be a nice thing to work on, but we can make a lot of money investing in software and there's a lot of software to build, like let's just do that because it's easier. So I do think there's a lot of tailwinds that just make the problems in the physical world more acute and more pressing.

• 55:48Speaker 00

  And I think the connection to that is like, when you have a pressing problem, people will pay for it, right? So like, that's usually a good business and investment opportunity. So that's one piece. I think the other piece frankly is like a little bit of FOMO because 15 years ago, it was hard to see like, you know, a big list of deep tech companies that had done well in some of these areas. Like they're kind of these one offs, like you have Nvidia, you have Apple, but like, you know, there were once in a while. Now in the last 10 years, you got like SpaceX and Anduril and like, I would say Palantir kind of more in the defense side, which is kind of deep techy and Tesla. And so like you just have all these examples where all the investors in the last 10, 15 years that were avoiding these categories 10 years ago are now seeing like, well, there's a lot of activity here.

• 56:30Speaker 00

  All the like ex-Palantir, SpaceX, Tesla alumni are starting new companies, they're all really good, like we don't want to miss the next wave. So I think there's like a bit of like FOMO there. And I think it's justified FOMO, right? Like I think it'll be a great wave, but I think the fact that there's just like a lot of amazing teams and a lot of signs that you can build these like hundred billion or even trillion dollar companies on the deep tech side in the last 10 years, I think those, those get later stage investors really excited.

• 56:55David Weisburd?

  Jeremy Giffen, founder of Octave, has I think the most interesting take on this, which is as funds have gone larger and especially as software has become more capital efficient needs to start, VCs are now looking for capital intensive industries in order to plow their capital into. So a lot of these net new hardware companies and net new companies that take a lot of money to start are downstream of the venture fund size. What do you think about that?

• 57:26Speaker 00

  I think that's a good take. I think I largely agree with it because, I mean, it's something I've been thinking about in the AI side, like you mentioned, you have, you know, maybe small teams that have, you know, millions in revenue. But I also think like, it'll probably get more dramatic than that, right? Because, you know, today's coding models are essentially like the worst they'll ever be. And so if they're 10x better in five years or 50x better, like what does that mean for building great software? And like, I think you'll have, you know, one or two person teams of like $30 million of revenue and that's going to be an amazing venture scale outcome if they ever want to sell it. But like, why do you need venture at that point, right? Like you don't need seed funding, you don't need growth funding, you're probably pulling in like tens of millions of profit very quickly.

• 58:10Speaker 00

  And so I think like, to your point, I think you just need to start deploying in areas where that, you know, do need venture funding more. One, because maybe there's like, it takes more time, so there's just dependent venture funding. Two is maybe like, if the outcomes can be really big, you can also justify like, you know, big rounds at much higher prices. And so I think, I think deep tech's very well placed for that.

• 58:31David Weisburd?

  We spent some time before this podcast talking about, and you're, you're a true believer in hard tech and you, you drink your own Kool-Aid. But if you had to steel man the deep tech thesis, what would have to happen for deep tech to fail as investment thesis?

• 58:48Speaker 00

  I think the ecosystem is still somewhat fragile in terms of, there's a lot of companies that are raised a lot. But I think it's still in a place where if you had like a few high profile failures in a row, that would, I think that would like sober up the market a lot. You know, I think if you had like, like, for example, I would say like, cloud software feels big enough that like if a huge cloud company failed, like it wouldn't affect funding for cloud software. But we saw, for example, like 10, 15 years ago at SUSE, like e-commerce was like a small enough segment that whenever there was like a big explosion, e-comm funding would just dry up for a couple of years. And then there'd be like a big e-comm exit and everyone would get excited again.

• 59:33Speaker 00

  It was a small enough market that like single huge failures or, you know, implosions would like have a big ripple effect. So I think deep tech is like, kind of on the verge of outgrowing that, but still hasn't. So I think if you have like two or three companies that raised a billion dollars and failed, I think that would like really, you know, turn off the, you know, a lot of the downstream funding that we see. And then related to that, I think the bigger, this is basically at the core of the steel man argument is like, I think it's still very funding dependent, right? Because with software, you can bootstrap, you can hit higher revenue quickly. With deep tech funding, most of the businesses we fund won't be able to survive standalone until like, you know, series B or C, right? And so if the series A investors decide like, we're just not doing deep tech for five

• 1:00:16Speaker 00

  years, like it's going to, you know, really, really crush deep tech returns and potential for a few years.

• 1:00:23David Weisburd?

  Your first point is such a great point, which is one of the reasons why you've seen deep tech funded today is because deep tech is being funded today. You talk about it being based on Thoma over since SpaceX, Anduril, Palantir. But there's a lot of truth to that in that if you imagine a world where Anduril could not have gone out and raised the amount of capital at a pretty substantial valuation to start, we wouldn't have an Anduril today. So a lot of these become self perpetuating on the way up. They also become self perpetuating on the way down. In other words, if everybody cools on deep tech today and they stop funding, those companies will die. And then everyone could say, see, I told you deep tech was overvalued or was not investable. So you have these recursive loops

• 1:01:09David Weisburd?

  in different industries based on what ends up perpetuating and what ends up succeeding. It becomes somewhat self fulfilling.

• 1:01:17Speaker 00

  Yeah. And I think it maybe like the commerce example, it might be subsectors like in robots. If like figure goes under, I think it's going to give a lot of people pause and humanoids for like, you know, a few years. But so it might not have ripple effects across the whole domain. But I think, you know, deep tech is, it's a collection of a lot of different areas, but there's like a few that are really big, like robotics being one, chips being one like. So, so I think, you know, there's like a few of those areas where one big failure in that area could like really affect that area for a couple of years.

• 1:01:48David Weisburd?

  Yeah, this is something Elon was very conscious about that if he had failed in SpaceX, he would really take back the rocket industry by another 20 years just because it was such a loud and audacious attempt that it would have this ripple effect psychologically that would take another generation of investors before it was investable again.

• 1:02:06Speaker 00

  Yeah. And sometimes it's like, you know, that would be like a failure, like not do lack of trying, but just more like it didn't work out. There's also like, like Theranos is a good example, right? Where I think, you know, the idea of like testing blood off of a very small sample is like very interesting. But I think Theranos dried up that market for like, you know, five years plus because of the way they operated. I think there's some risk there for like a lot of the subdomains where if you look five years from now and there's like, you know, 10 defense Decacorns, I don't think it'll matter if one goes under. But I think when there's like two or three, if one goes under, that would be a big deal.

• 1:02:40David Weisburd?

  If you could go back 13 years ago, right before you started SUSE Ventures, and you could give yourself one timeless piece of advice, just one piece of advice, what would that be?

• 1:02:48Speaker 00

  So there's like a tactical one and like a philosophical one. So I'd say the tactical one is like, don't pass because of price. Cause I think kind of our biggest misses were ones where we liked everything except the price. And then we passed because of the price. And then years later, it's like, you know, a multi-billion or one times $10 billion company.

• 1:03:06David Weisburd?

  How did those play out?

• 1:03:07Speaker 00

  I'm happy to throw out a couple of examples. Like we, we were interested in investing in Checker in the first round. We were interested in investing in FAIR in the first round. But I think both of those essentially like skipped the seed and went to A, right? And so when we were a seed investor, we're like, hey, we'd love to invest. Here's like a seed price that we're pretty comfortable with. And then, you know, the price ended up being like three times that or something like that. And we're just like, oh, it's like, you know, it's really more of a series A. It's too late for us. We're like, we liked everything else about the businesses. And now I haven't kept track, but I think FAIR is like a $12 billion company and Checker is like five or something like that. So even a two or three times entry price, those have been a great, uh, great investments. So a lot of times it's like someone skips straight to A or like maybe there's it's a seed round,

• 1:03:51Speaker 00

  but it got priced twice as high as we thought it would be. Usually if it's like 10 or 20 or 30%, we can bridge that gap, but like double feels pricey. So, so that's sort of like the, you know, but like, if, if the concern was like, well, we're not sure about the valuation and we're not sure about the team or the product or the market, that's usually like a good reason to just pass. But I think we found with those companies like Checker and FAIR where it's like, we love everything except the price, like we should have just done it. So that would be more of like a tactical lesson. I think the philosophical one is just to, basically the exit is the thing that really matters. And there's a lot of companies where maybe it feels like they're going slow. And so like, you know, it's been a couple of years, it feels like they're running out of money. So we start thinking internally

• 1:04:34Speaker 00

  like, well, it's going to be a zero. Or there's other companies where in the short term, they're raised really quickly. So they go like seed to series A to series B in 12 months. We're like, this is going to be amazing. It's going to be a fun returner. And then sometimes like the companies that go slow explode and do really well. And sometimes the companies that raised a lot implode and, you know, because like maybe the Unity Comics work or like some competitor did better. And so I think I'm just going to appreciate that the interim markups are like an interesting way to measure progress, but like they're very far from perfect in terms of predicting how well your fund's doing. Cause sometimes, you know, the big positions go to zero. Sometimes the small positions become big, but just like much, much later. So yeah, I think now I kind of, you know, I'm not like, I try not to take too many lessons until something's actually like exited or there's secondary opportunities.

• 1:05:18Speaker 00

  Because before then, it's all just feels like, you know, kind of big.

• 1:05:23David Weisburd?

  DPI is king.

• 1:05:26Speaker 00

  Yeah.

• 1:05:27David Weisburd?

  Well, Leo, been absolute masterclass. Thanks so much for jumping on.

• 1:05:31Speaker 00

  Yeah, this is a blast and I love your questions and a big kind of the show. So I'm really excited to be on.

• 1:05:36David Weisburd?

  Thank you, Leo.