Mike Schroepfer of Gigascale Capital joins Nick to discuss The Former CTO of Meta on DeepSeek, Lessons from Zuckerberg, The Next $10T Market Opportunity, and the Impact of DeepTech on Energy Costs and Climate Change. In this episode we cover:
- Challenges and Innovations at Facebook
- Lessons from Working with Mark Zuckerberg
- Gigascale Capital’s Investment Thesis
- Deep Tech Investments and Market Opportunities
- Regulation and Its Role in Climate and Energy
- Future of Energy and Climate Tech
- Inspiring Founders and Innovations
Guest Links:
The host of The Full Ratchet is Nick Moran of New Stack Ventures, a venture capital firm committed to investing in founders outside of the Bay Area.
Want to keep up to date with The Full Ratchet? Follow us on social.
You can learn more about New Stack Ventures by visiting our LinkedIn and Twitter.
Are you a founder looking for your next investor? Visit our free tool VC-Rank and we’ll send a list of potential investors right to your inbox!
0:18
Mike Schroepfer joins us today from theBay Area. He’s the Founder of Gigascale Capital, an early-stage venture firm investing in deep tech solutions that address climate change. Prior to Gigascale, he was the CTO at Meta, leading the company’s technology strategy and driving innovations in AI, AR/VR, and infrastructure. Prior to Meta, he was the VP of Engineering at Mozilla, where he played a key role in advancing open-source web technologies. Mike, welcome to the show!
0:46
Thanks. Excited to be here. Yeah, it’s pleasure to have you. I’ve been following for a long time, and it’s exciting to get a chance to chat. So give us your two minute backstory and your path to giga scale.
0:57
And I’m a techie, so I graduated Stanford in the 90s. CS degree lived through the.com one. Dot boom actually started my first company in 220 2000 you know, NASDAQ crashed in April of 2000 we were the only company that a software company, Sequoia Capital funded and q4 of 22,000 so that was a that was a journey. It was like a zombie apocalypse at the VC office. As you walk in, all the glass conference rooms were empty. Rooms were empty because all the companies are growing out of money. So we started a company in a downturn, which actually was an amazing time to start a company we could talk more about that. We, couple years later, sold it an all cash deal to Sun Microsystems. Then I joined Mozilla and take Firefox into hundreds of millions of users, and then joined a little social network called Facebook in 2008 which at the time was was had fewer users than MySpace. And you know, the word on the street was social networks didn’t make any money, and I sort of was lucky enough to be there and help it grow. When I joined, it was maybe 100 people on my engineering team after I took it over a couple weeks in, and then when I stepped down a couple years ago, it was 35,000 on my team, and we built data centers and AR VR headsets and the meta Ray Ban stories and AI research lab called fair. I built llama on a bunch of other things. So lots of lots of fun over 16 years, and I still actually have a relationship with meta and help them on on AI issues and others. Five years ago, I got very passionate about the climate crisis as COVID was happening, and had sort of more time on my hands to think, and was thinking of challenges that face us in the future and face our children and others. It was obvious to me that this was, this was happening. You know, I thought it was good news, that it’s man made because of it. If we caused it, we can fix it. And I started, just like digging in, trying to understand what to do. And I started, and continue to do a bunch of philanthropic work, funding science and policy to advance things. It’s a ten trillion problem, and we’re going to need industry and capitalism to help. And I just met a ton of interesting founders along the way that had very credible ideas that are what I love in technology, which is like, here’s this massive tail when we’re running that every day, when I wake up, my the things that I use as inputs in my product are getting cheaper, faster, better, as it makes my life easier to build a product that’s going to be awesome in the marketplace. And that’s that’s the magic of technology I’ve seen over 25 years, but just applied in a more specific domain of things that sort of have some sort of not only better, faster, cheaper, but also come along with a co benefit for the planet. So
3:18
when you joined Facebook meta, was it still in the university phase, or were you open to the general public at the time? No, it
3:25
had opened to the public. I think they launched in 2005 so it’s about a three year old company. So it was, it was open to the public at that point, but still less users than MySpace. Yeah, believe it or not, that is true. I mean, this is bad. People probably don’t even remember Friendster. Friendster sort of famously imploded because they couldn’t, couldn’t scale the technology. The Psych at MySpace was was bigger. And, in fact, Mozilla, where I was coming from, we had more users on Firefox than than Facebook ads. So in some ways I was sort of jumping to something smaller. But I think that’s the important lesson, jumping something smaller, but on a, on a obviously, what has turned out to be, obviously much, much bigger trajectory. I
3:58
mean, you went through so many different iterations of very hard technology challenges that had not previously been solved. You know, whether it’s like news feed, interesting things you did with photos and tagging and, I mean, there was just layers and layers of like, interesting technological challenges that were solved at Facebook. Is there one in particular that you’d say was the most ambitious or one of the most difficult that you took on?
4:22
Well, I mean, there’s, as you said, there’s so many. I mean, back, if you go to the very early days when I first got there, and this idea that you could build a site where anyone on the planet could interact with a piece of content, and everyone else would see the updates, you know, and so back in the day, it was the the Bieber phenomenon, Justin Bieber would post, and everyone commented like and it was just a madhouse to, like, make sure the systems didn’t crash, because this idea and computer science terms, multiple writers, multiple readers, you got a bunch of people inputting and a bunch of people reading, so much, much harder problem than the sort of, like, search or news problem, which everyone kind of gets the same results. You can cache it. It’s there’s a bunch of hard things on the back end, but the front end is a lot easier. And so we. Build a whole brand new software suit from front to back to make this work. And we went over multiple iterations. And we do what I actually advise a lot of our startup founders to do, which is, like we started with off the shelf stuff. So it was MySQL and memcache and PHP, which was kind of like the best stuff you could get in open source at the time. We then started, like, opening up the hood and saying, like, okay, what are these things? Can we change? And we made a bunch of really interesting revisions to memcache in particular, and a lot to MySQL. And then we got to the point where, like, we kind of need a different car, like, we can’t just, like, add a turbocharger to this engine, ripped it out and replaced it with new things, a new software stack that allowed us to sort of scale in the caching layer, which is replacing memcache with Memcached, with a different system, and on the software side, that was like a lot of the early days. And then we ran into the hardware problem, which is like running around in the Bay Area. This the financial crisis that happened. And so people weren’t investing in real estate. So people were not in 2008 speculatively building data centers. And we were growing really quickly. And one way to scale the site is rack more servers so you can support more users. And so not only did we have these software challenges, we would like spend a whole bunch of time trying to make sure that we tune the system to get some efficiency out of it so the thing didn’t crash and we couldn’t get more data center space. We’re just like, literally, there was no none available. We got some on the East Coast in Virginia, did a whole bunch of work to make the site work across two coasts. And then we’re like, this is this is crazy. We should just do this ourselves. And so we embarked on a project to start building our own data centers. And like, like, buying a land, designing it. And I think our Prineville, Oregon was our first data center. And 1015, years later, forget the public numbers are but 10s of millions of square feet. Open new data centers all the time, and we did a bunch of innovations on top of it as we like, it’s a chance of looking at something from scratch and saying, like, Well, why? Why do all these data centers have air conditioners in them? You know, are the servers really? They’re not bottles of wine. They’re actually not that delicate. Like, turns out you can run them at 8090, degrees, and it’s fine. And if you do that, then you don’t need a chiller. And we can just like, free air cool, bring air in from outside the building, your costs go way down. Electricity use goes way down, and so anyway, so that. And then we could talk about a shift from a web company to a mobile company, introduction of AI. We can talk about moving from enterprise hardware to consumer hardware. We started shipping VR headsets and smart to people at Best Buy and writ large, but lots of stuff to talk about. Crazy
7:20
Give it. Give us one of the best lessons you learned while working with, you know, Mark at Facebook meta.
7:26
Well, I mean, I think you can see this every day, which is, Mark is a commensurate student. He is just like, always learning like he went 20 years later. And because the situation, we are a web company that a mobile company, then we’re doing consumer hardware and enterprise hardware, and then integrated into one product. It was Instagram plus WhatsApp plus Facebook plus messenger, plus others, and dealing with sort of the AI boom going on now. And so the field changes. There’s no world in which you’re doing just the same thing for 20 years straight. And so the thing, there’s many things that are special about Mark, but I think the thing that is the biggest lesson for the entrepreneurs that we work with is just like a just a voracious appetite to learn at any point in time, and the humility to not be like, Okay, I got it. It’s just like, Okay. I need to learn new thing. We need to learn how to manage the company more efficiently. We need to learn how to do AI, who are all the people in the world who can teach me this? Let me go talk to them like, let me take everything I can, summarize it, build it into a new model that, like, helps me operate the company. And so that, I think is a thing that we see in some of our founders, and it’s just just incredible. Because the test for me is, like, we pick a topic area. I know a lot, I start brain dumping it to the founder. I give them six introductions for people, and then, like, three or four weeks later, they come back and they start teaching me things about that topic, like, then, then I think you’re off to the races, because it just means that it doesn’t matter what the problem is, you’re going to learn everything you need to know about it and and be better than anyone else and more contemporary than anyone else. And so that, combined with the like ridiculous work ethic and a like ridiculous and kind of, I kind of think at least a few of the many amazing things about Mark, unbelievable.
9:03
Well, Shrek, tell us more about the thesis in the investment approach at gigascale. Yeah, we talked earlier on about
9:09
how I sort of started in philanthropy and then went into investing, and that actually ended up being a true gift, because I know in the climate tech world, there was also sort of a mixing of those two. And when I decided to do the investing side. I said, Look to make this durable and to make it a ten trillion economic transition, fundamentally, people have to be self interested in this problem, that people are excited about these companies because they’re making them money, because they’re cheaper, better, faster, because consumers love them. And that has to be the lead. That has to be the thing. And our first one. And it sounds simple, but I just got to start. There’s often this thing. People call it the Green premium. We’re like, we’re investing in the green discount. It’s like, people like it because it’s cheaper also happens to be lower or better in some other way. And so that really is the first and the main part of the thesis is we’re only interested in technological shifts that fundamentally change the economics. That sounds real high. And mighty. It’s happening right in front of her face all the time. In 2024 in the United States, 90% of new electrons on the grid at utility scale is from solar. It’s because solar is the cheapest. It’s not really for any other lithium ion batteries are 98 99% cheaper than they were 30 years ago. These are products that are on these massive cost down curves, which is, I can go, take a year vacation, come back, they’re cheaper. And so they cause massive disruption to the industry. And this is kind of similar to what’s happened in computing over the 25 years I’ve been there where, like, Moore’s Law and equivalent, just like, shows up every year with with new gifts. And so that’s, that’s where we start. It’s like, has to be cheaper. Has to be like, riding some technological innovation that that is a credible story for why you’re sort of cheaper, better, faster. And then obviously we care a lot, as we just discussed, about the qualities and characters of the founding team, is that building a startup is a fundamental, fundamentally irrational outlier behavior. Most of them are going to fail, a few of them are going to be massively successful. And so you’ve got to have individuals that have the sort of capability and fortitude to sort of be on that journey for the long run. Is
11:04
that really what you’re indexing on when you’re looking at founders? You mentioned some of Mark’s characteristics before, and Mark is a one of one, but like, are you looking for these? You know, founders with this ability to learn, this tenacity, this ambition, some of the things that absolutely,
11:19
yeah, absolutely. And they’re out there. They’re hard to find, but, but we like this part of why I was kind of, like, litmus, you know, I did a dozen, 18 different angel investments before I started a firm, because I like to, like, another thing I learned is, like, get data as quickly as you can, so I want to run a bunch of experiments and see, like, all right, well, like, what, what are these people like? Let me track over a period of time. They’re, they’re rare, but they’re, there’s more than one. And the other good news is, like, we don’t know. Don’t need a million new companies, like, it’s going to be 10 to 100 new companies that transform industries, right? And so we’re fundamentally the game of outlier hunting. We’re looking for these unique people with unique technologies out there. It means a lot of frustrating meetings where we do a lot of work to get to know or look at a lot of things, to not find anything. But that’s okay, because when you find the one the company mixed with the founder, attacking the market, it’s really exciting, and it’s really fun to be part of that journey. So
12:12
you mentioned solar in the solar industry is quite mature, investigating the industry. Many years ago, basic photovoltaics, and then they went to thin film. There was a lot of subsidies in the industry and whatnot. Right now, it’s gotten to the point where it’s much more economically viable. How does that factor in your into kind of your framework, as you think about technologies, that may be a green premium at the beginning, and there, there is some curve to that. You can the green disk out in the future.
12:41
Yeah, there’s a bunch of things in there. So let me, let me, let me, kind of take in a couple different first, I think a mistake people make is not realizing there’s usually multiple phases in how these technological disruptions happen. And I think of the use of technology, you think about smartphones, which is sort of this transition from the web to smartphones, and who won as a result of that, it started with some of the big players. So like Apple and Samsung are making the bulk of the money on the phone sale itself. You’ve got Google is doing pretty well with with Android. You’ve got chip supplier so Qualcomm, Nvidia. These are all like great investments if you had done them in 2008 but then once they arrive at scale, you have an explosion of new things that you’ve never seen before, Instacart, DoorDash, Uber, these companies wouldn’t exist without the smartphone, and they’re also a huge opportunity. And so when you think about solar, I think solar is a little bit in the like smartphone, or it’s like, solar is here. Like, I don’t think investing at a venture capital level, at a solar cell producer is there may be some outlier technologies, but it’s like less likely, but you have to start asking a question, well, now that it’s here, what does it do? And it makes electricity generation really cheap but really variable. Shows up quarter of the time, rather 90% of the time. And so that’s going to disrupt a bunch of things, because it just like, changes the way power is made. So, like, I’m curious in all the things that that like intermittent but cheap generation sort of disrupts. I think a thing that’s maybe be like left shifted about a decade is batteries. Batteries are here. You have one in your pocket. You might be driving around on one, but they continue to get cheaper on the order of 10% a year, actually 50% in the last year for lithium ion. So you can bank like we can bank another easy 10% win per year for the next five or 10 years. So you just start looking at a bunch of injuries, saying, when’s the crossover point when a battery powered version of that ends up being cheaper than a gas or diesel powered version? And just to give you a very specific example, so we’re not talking and a company just launched that we invested in, called SDS energy, and they’re basically building it’s a bunch of amazing people been working in the space for a long time at Proterra and other places, and they said, Look, commercial vehicles. We’re talking busses, we’re backhoes, things that are big are all diesel right now, and as these battery packs get cheaper, we’re going to basically be able to supply them into these industries and provide devices that are cheaper to operate over time, TCL will be lower. Fits our green discount. Um, through a bunch of innovations and other things. And so that’s so it’s not, we’re not making battery cells. We’re not sending battery chemistry, but it’s like, hey, we can package batteries and attack a market, which is the commercial vehicle market, with an incredible founding team. So that is just putting together the high level trends with like, batteries are getting cheaper, what’s going to get disrupted? Here’s a team boom. Here we go. And that that sort of like in in our sort of nutshell, what we
15:21
batteries is an interesting one to to reference, because I was involved in battery R and D many years ago. And I think an interesting thing about batteries is the technology evolved much more slowly than people talk about Moore’s Law and whatnot. Capacity and efficiency and cost of batteries did not improve and decline, respectively at the rate that we expected. So I’m kind of at what point you get involved in technology. And to be more specific, with Mike, there’s R and D, and I’ve been a part of organizations that are trying to do novel things, and there’s a lot of focus on the R, and then I’ve been a part of organizations where there’s a lot of focus on the D, and it’s a lot of Applied Technology instead of new, novel technology, what would you say you spend more time on at gigascale?
16:05
Yeah, I think we’re definitely more in the development and productization of these things. And there is, you know, usually I’d say there is some technical risk. So there isn’t, like it’s all solved. We’ve got it all figured out, but you’re beyond sort of lab scale or bench scale thing kind of working. I have done both too, at meta. We, you know, working on augmented reality, virtual reality, for a decade plus, built a lot of VR headsets, did a lot of frontline research on those. They started an AI research a decade and a half ago, ish, 2013 decade ago. And so I’ve been on the on the frontier. End. My experience on this, personally is, like, the R part is quite expensive, like, and you just can’t get around that, which is, your rate of failure is high. So it’s you have to have, effectively, a portfolio and, or patience to get there and, and that’s a really important task, but you need some well funded source that can be a government lab, it can be a large company, it can be a university. Those are great places to try a lot of things and have a lot of things fail. I think it’s a tough place for a venture capitalist to be, because the math just is enough companies going to fail on the commercialization side. When you add an 80% failure rate on the technology side, it gets, gets tricky. So I think we’re sort of in that crossover point where, like, we’re pretty sure this technology works, or we see this trend happening, and now we’re, like, five or 10 years out from massive commercial adoption, because, like, that’s kind of what you’re going for. And so what’s really on that sort of transition point where we really are, and where I get get really excited,
17:27
amazing. While I’ve got you, I have to ask you about deep sea. It’s on everyone’s mind at the moment, just to start off. Do you think this is a legitimate threat to the US, LLM model companies? You know, obviously you’ve had some experience with llama. I doubt you can talk about that in detail, but, you know, open, AI, etc, or there’s, there’s many that are saying this is really just some sort of distillation wrapper disguised as a foundational built on some of these others. What’s your take?
17:53
I think a couple things all at once. One is, I think that it’s both exciting and overblown, and I think that there’s a lot of details that I think we don’t have full data on, that we’re getting more it’s like some basic questions, like, how much money was spent on the infrastructure? Numbers range from 5 million to 500 million. And depending on the answer that question, you may be more or less excited about it. How much did they sort of distill off other models? You know, where does the data set? These are all like, important kind of questions to answer before you can get to a fully formed here’s my opinion on the thing. I will say, zooming out a little bit, I’ve been a huge, huge, massive proponent. You know, I’ve worked at Mozilla on open source. At meta for a long time, we open sourced lots of things we worked on. I’ve been a huge proponent of open source and open weight models, primarily because they’re an accelerant to progress. So I would say that, like, wherever it came from, or whatever they did. The idea that there are multiple models out there that you could poke at, look at the way it’s examined, read the papers, and how they did it like it just means, in the next three to six months, we’re going to see a tremendous follow on progress from these whether it’s from open AI or others. And so you’re just going to see an acceleration of the capabilities of AI models, which is super exciting because, to our general thesis of technologies, as they mature, they are there. I think of them like a toolbox where it’s like, oh, I didn’t have a screwdriver before. Now I have a screwdriver. Cool. There’s like, things I can do I couldn’t do, and that, that, to me, has been the exciting part of technology, in terms of giving us options as humanity, to go solve and attack problems. And AI is doing that for us in a variety of ways. So I’m excited to see things progress and move more quickly. I
19:24
mean, there’s an argument that we’re going through this huge platform shift to AI at the moment, and sort of the application layer explosion is upon us. Is there any risk that China overtakes the US and Silicon Valley and kind of being the lead in that. I
19:41
think anyone in the United States who isn’t worried about the the competitiveness of China isn’t paying attention. It is an incredible force of nature. They plan in very long time intervals. They invest a lot in infrastructure. They like particular targets of industries to go take down. And like, go work. On step by Okay, we got to build we got to build airplanes. We need engines. We’re going to go do all of this. So I think looking at that and waving it away is, is like is, is a fool’s errand. I think it’s competition, and it just means that we got, we got to be on our game. I think the US, and this is part of why I’ve been a fan of open source ideas. The US model is fundamentally model of distributed innovation. We let the markets figure things out. We have universities. We have government labs. We like kind of let 1000 flowers bloom, and believe in the power of sort of competition to make the best come out. And I think we just got to kind of push really hard on that. I think there’s a place for companies to work really hard at the US. Government can do more to fund clusters development, natural lab, national labs, others. But I think that we have to be thinking of where we’re moving in the future, which is part of also why I’m excited about when you think about if we can crack, for example, fusion. Fusion is just a state change in our ability to produce anywhere we want, on demand. Us is currently in the lead. All the best companies are in the US or some in Canada, some in the UK, China is investing heavily in this, to try to replicate what’s happening. I think, I think we need to be thinking about this of, how do we invest in us, from an infrastructure standpoint, from an AI standpoint, from an energy standpoint, to make sure that we are sort of continue to be on the forefront of these things, because we have very, very good global competition, and
21:18
back to energy in a second. But, but last question, it sounds like you’re more in favor of the open source approach to AI, why? Because
21:25
I, you know, I think that the lesson is sort of when you have foundational technologies, whether it be operating web browser and the CS, the original protocols of the internet, TCPIP and IP interoperability, like these, are things you’re going to build a lot of stuff on top of, like cool i’ve got a model. I need to apply it now to all I need to apply it to weather simulation. I need to apply it to legal I need to apply these other things, which are a whole bunch of application specific work that needs to happen. But like, should each of those companies be building their own special foundation model? No, they should probably be using, like, a shared model underneath that. Should that thing be owned by one company and be the only place you can buy it from. Like, I think it’d be better if there was some competition in the world. One way of having competition is, like, you’ve got this open source model, maybe some commercial providers as well. Microsoft still makes money selling operating systems. A lot of people use Linux. I think this is great, like, so I think that there’s a way to really stoke both competition plus distributed sort of innovation, in a way that acceleration, that helps us get what we need when we need it.
22:26
Trip. What do you think has the most positive impact on GDP and quality of life? Advances in AI or clean, cheap energy? Well,
22:34
I mean, I think that my, my reading of history is like energy, the plot of like energy use per capita and like, like, quality of life is a fairly, fairly clean plot and so, and if you just think about it intuitively, in your daily life, the number of activities you do where you burn energy for convenience, it’s, it’s incredible. And if you just distill it down to like, starting with keeping me, like, healthy and happy, it’s like, am I living in an environment where it’s not too hot, not too cold. Can I drink clean water? Like those things are fundamental energy problems at this point, and so I think clean energy is the unlock for humanity. As I started looking at a lot of sort of climate tech companies, there’s a lot of problems to solve where we kind of know how to solve it, but the problem is energy costs, and it’s not super obvious, like an easy one. Let’s talk about sustainable aviation. It’s nice to be able to fly from the US to Europe or the US to Asia. We need something that’s super energy dense to do that, batteries are on no track to get there. So you need a liquid fuel of some sort, cool, cool. We can, like, dig it out of the ground and burn it. We can also make it by taking CO two from the atmosphere and creating we have the technology to do it. It’s just like three to four times more expensive per gallon, and nobody’s going to pay that. And if you kind of back out to where all those costs go, if I could take my energy cost down, 510, X, all of a sudden, now ESAF is competitive with everything else. I think of energy costs as like so upstream of so many other problems. Even when you talk about it, you talk about advances, its data, its algorithms, its compute power. Compute power is limited by like, availability and power of the chips and energy and like, as these clusters get bigger and bigger and bigger, like energy actually becomes a larger and larger component of this over time and so and I think especially as you move into like, if you think of a world where every human on Earth has a reasoning model working on my behalf. So we got 8 billion inference models running, like, on a daily basis. And you start thinking about like, should I have a kilowatt at my disposal? You know, times 8 billion? Like, that doesn’t seem totally given that I drive around in two and a half tons of metal every day. And like, you start thinking about that, and you’re like, ooh, we definitely want ways to produce that that are cleaner, better than what we’re doing today, and cheaper too. And I think demand for compute will be basically gated primarily by energy costs over time. And so I think there’s a lot of interesting technological innovation opportunities there. So it’s a hard like, it’s a little bit of a like, a yes, and they’re both really useful, but I think that part of why I got passionate about what I’m doing is just like. Having seen multiple technology trends back in the internet, 1.0 days to the mobile to AI era, it like is obvious to me that like, unless we crack some ways to, in particular, produce and distribute energy at dramatically lower costs, that will end up being our rate limiting in you know, if it’s not obvious now, it’ll be obvious in five years. I
25:17
think I just saw news that boom supersonic jet took its inaugural flight. Do you know what fuel system it’s using? I
25:24
mean, they’re using jet fuel. Are they using? You know, I was actually going to go see it, but the timing didn’t work out. So I’m excited. I would love to ride on supersonic to East Asia on E jet fuel, like, and it’s a cost problem. Talk to us
25:38
a bit about energy types, like we’ve talked about solar, you know, obviously there’s nuclear, you know, there’s gas based, you know, fuel technologies. Like, is this part of the framing as you think about thesis and areas where there’s the most opportunity, and how do you kind of stack rank, you know? Yeah, I think there’s,
25:55
there’s this. I’ve had lots of conversations with my friends and tech who, like, always want to have the, like, the one answer, and so I was like, why isn’t it all just solar and like and like? That’s a really tempting place to be, but like, the more you start digging out all the details, and you’re like, Okay, I’m gonna, like, heat England in the winter with solar. Like, there’s literally no sun for months at a time. Like, I need really, really okay, batteries. I’m gonna need a whole hell of a lot of them. Like, this starts to not make any sense. And so I think the reality is, I think of it this way. I think solar is like the cheapest so deploy as much solar as you can, everywhere you possibly can. Next thing you can do is, like, anytime I could time shift that production, which is batteries, like you’re in really good shape. So we’re investors in form energy, which try to do 100 hour iron hour battery. A lot of other people are working on ways to do storage. I think anytime you can unlock storage, you just like, open the deployment window of solar dramatically. And so I think those are just like, and those are going to happen. We’re going to put a lot of solar and a lot of storage on the grid no matter what. Then I think you start looking at like, Okay, are there ways to complement with places like the UK, where it’s sunny all the time, or times at night, or it’s not there. And that’s where I think you look at geothermal, which is really interesting. You look at a next generation of fission, you look at fusion, then you look at, sort of biomass, and bio energy as a combination energy source and and carbon capture. And so I think it’s all of these things, and I think we’re going to kind of need them all. And so we’re, we are looking, and have looked at all of these spaces, because at 1020, years from now, it’s going to be a grid of a mix of these things, and anytime we can move one of them forward in cost competitiveness, I think it unlocks a whole bunch of a bunch of new opportunities.
27:32
You know, back many, many years ago, my Dan or her days, I ran a aerospace M A strategy, and we know it was based on power and we looked at power generation, distribution, conversion. And do you think about energy in that context as well? I mean, you talked about, oh, storage as well. I forgot batteries. Do you think about those categories and which ones you want to participate in versus which provide less interesting sort of opportunity? Yeah, for
28:00
sure. I mean, there it’s, I’m sure you’ve read the book the grid. It’s sort of like, by some arguments, the electrical grid is the largest interconnected machine humanity has ever made. It’s kind of amazing. And there are ways I think of it like this, which for intermittent sources, wind and solar in particular, you can either time shift it with batteries, or you can location shift it with a grid. So I can move it from one location to the other. And we’re going to do both, and we’re going to need both, and there’s no version of this that doesn’t include 5x or more upgrade and grid capacity. There’s a bunch of exciting again, you start saying, like, what are the technological innovations? Like, oh, we have, like, we know a lot more about materials than we did 50 years ago. It turns out, like, we can build something better than a steel cable, you know, that can carry more power. We can dynamically line rate it so we can say what temperature is it actually and like we’ll rate it based on that, rather than the averages for the year. Bunch of things to basically get a lot more power through the same sort of superconductors is another one. So grid interconnects, super awesome. Batteries is a time shifting device, super awesome. I think one of the things that’s exciting about these point source power production capabilities. And I put biomass in this category, geothermal, fission and fusion is you can drop them really close to the load. If I am building a massive data center, I can put my vision fusion geothermal biomass plant next door, and I don’t have to send the electron so far away. I can still grid connect for backup and a bunch of other things. But the dominant load source can go straight to where it is, and that’s a different way of sort of solving all of these problems. I have to build a giant high voltage power line across the United States. So I think, I think it’s easy to try to be reductionist on all these things, but it’s a little bit more like we need all of these things. So anytime there’s an interesting opportunity, not to mention infrastructure, we’re, like, buying a lot more transformers than we used to. There’s a lot more stuff going from x to electric and so power, how do I need? Gal, especially with automotive, right? Like a lot of things, like, there’s just a lot of opportunities for optimization. And when you make these big changes, you know, I’m fundamentally interested in big changes. You know? Why I think EVs are an argument I keep making to people with EVs is, like, the powertrain. About three to 4x more efficient a gas engine, 20, 25% efficient. Electric power chain, 90s, mid 90s efficient. So, like, I gotta screw up a lot of things ahead of that to, like, burn down my 3x advantage. And, like, that’s kind of the way I think about technologies is, like, fundamental we look at things that, like, start out with a five to 10x and, like, a bunch of things can go wrong. I’m still gonna beat you so and that gets really, really exciting for me. I think batteries, electric power trains are there. I think there are other technologies coming, electrolyzers, I think are getting there soon. Like, there’s a bunch of stuff there that I’m really excited about that, just like, start with such a massive advantage that that it’s hard to catch,
30:39
not even to mention, like, all the point of failures that exists with an ice based car in the aftermarket. I mean, most, most of the money is made in the aftermarket, right? Because so many things break. And think about all the efficiency lost all the time, all the energy that goes replacing transmissions and
30:58
and then I like these places where you basically integrate a whole bunch of innovation into one point. So if you just think about, like, clearly, the pinnacle of humanity is chips, I think. And if you think about, if you’ve ever looked at what an ASML machine is, and the sort of just unbelievable, like throwing little, microscopic dots of tin and hitting them with every point, oh, to make up UV light in order to etch these it’s just like, unbelievable, the compound investment in R D that has gone into the like, this machine that makes something that I can hold my hand right and then power something else. And so it’s just this, like, amazing concentration of energy of humanity. And so I think, like, solar has that. I think batteries, like everyone keep looking for, like the great big battery breakthrough. And what they kind of missed was, like, an accumulation of 1000 half a percent wins. Just like, oh, we dope it this way, a little bit differently. It’s better. Oh, if we package it this, okay, the thermal management, if we manage this, okay, the charge rates, okay. Like, you just, like, look at over the last 10 or 15 years, there hasn’t been a fundamental chemistry shift. There’s been just an accumulation of a crap ton of into this, then those things get me really excited, because, like, that’s how we just, like, will continue to make progress that this, to your point, isn’t going to be this, like, oh my gosh, here’s this new thing, but it’s just going to, like, accelerate us out of, out of orbit.
32:12
So we’ve been talking around a lot of supply and demand, dynamics of energy. Jeff Bezos had this recent interview on New York Times, which was interesting. He where he said that the best way out of the government debt crisis is raising GDP, and if we can continue to innovate and substantially raise GDP, then the debt becomes a smaller relative issue. Do you feel like there’s a parallel here? In a similar argument to be made with energy, where if we increase our productive capacity of clean energy exponentially, and do so at low cost, that this will be the single biggest lever to combat climate issues.
32:49
Yeah. I mean, I’m definitely a like, Let’s build this. And it’s been my experience in tech over 25 years. It is the thing that shows up to zero sum problems and removes the constraints. You know, so many things in life are like, do we fund this, or do we fund that? And you show up, you’re like, Look, I just made a 50% cheaper. You can go ahead and fund both, and that. That is technology. I mean, if you just, like, plot progression of things like chip prices over time, solar prices over time, battery price, health care, a bunch of things like, they get they exceed inflation, and a bunch of other things are, like, getting cheaper on a regular basis. So like, you definitely want to push your chips into the category of like, this thing’s gonna get cheaper over time. That’s gonna then give us a bunch more options as humanity is what to do. And I 100% agree that if we can grow the economy, we can grow productivity, we can grow innovation. Invest in AI, invest in new energy, invest in fusion, invest in a whole bunch of other things that’s gonna, like, fundamentally, give us a whole lot of options as to what to do, as opposed to, like, try to, like, cut our it’s like any startup or any company, it’s like, you never, can’t cut your way to revenue growth. You need to, like, invest and innovate to get there. And that’s always a much better place to be than to, like, manage a shrinking thing. So
33:58
you’ve spent a lot of your career dealing with bits, you’re spending more time dealing with atoms. Conventional thought says that hardware startups or startups that have deep tech components are more difficult to build and scale and have capital efficiency issues. Why do you think these present good investment opportunities?
34:16
Yeah, it’s a great I think there’s like, it’s like, a classic, you know, anything you hear that’s common wisdom. There’s like, part of it that’s really right, and part really right and part of it that’s really wrong. So the part of it that’s really right is, like, yeah, it is harder to pivot a hardware company, you know, Estes, who’s making battery packs for commercial vehicles, they’re like, not going to turn into a fertilizer company. Like, it just like, it’s it’s different. If that turns out to be the wrong market, then we’re in deep trouble. Now I’m pretty sure that’s the right market, so we’re okay. So I think this is a case where, like, you measure a couple times and then cut. And anyone who’s moved from hardware to software will tell you this. It’s the first lesson is like, yep, mistakes are more expensive, so you’re going to try real hard to make less of them and or get to them faster. And I think that the other lesson that SpaceX dot everyone is get to those mistakes faster, because then you. What you can do. So you got to be more careful. On the other side of that is like, would you have liked to been an early investor in Nvidia? Yeah, I think so. Like, would you like to be an early investor in Tesla? Should you have gotten in on SpaceX early like, these are all hardware, deep tech, hard companies, hard tech companies. So I think, I think that like, prize on the other end can be effectively unlimited. And even if you look at Microsoft, Google, meta, Amazon, they all have very large hardware components to what they’re doing at this point, they’re not just like bits companies anymore. So do they all build data centers, they all ship consumer hardware. They all like, and it’s not by accident. So so I like, it’s just like, hard for me to, like, pretend like we’re gonna solve an atoms problem with bits, like, we do some software investing. I think software is needed in certain places. But I think the opportunity for, like, I got, like, a new way of making electrons that is fundamentally cheaper, and I can put it anywhere in the world. It’s like, I think that’s gonna be worth a lot of money. I think that’s a safe. I think betting on commercial vehicles switching to batteries pretty confident in that one. Like, so, so yeah, I think there’s a lot of lot more sort of ways to make mistakes. And so this is part of what I did. Is I was like, Look, I have a deep tech background, but we brought in with chemical engineers, mechanical engineers. I have people all throughout and part of the process, we do a lot more diligence, particularly on the technology, and particularly on the cost and commercial side of the technology. It’s like, where are you now? Where will we be in 10 years? And unfortunately, a lot of companies fail those tests, and it’s just like, you’re just not going to be fundamentally cost competitive. And so as much as I want you to win, we’re not. This is not for us. But then you you do find the ones are like, Huh? Not only you like, good now, but you’re going to be like, overwhelmingly competitive in five or 10 years based on some very simple commodities. Don’t get cheaper. No magic occurs like, and I think when you do that, you find those, you’re like, Ooh, yeah, this is this. Is it? This is exciting. And so that’s what we do.
36:47
We’ve done a very limited amount of deep tech investing, and we’ve gotten really lucky there. But, you know, a lot of our assessment based on gut and based on some principles that we know a lot of these companies are going to have deep remotes. They’re going to have stickier there’s going to be less competition. Their pricing power is gonna be higher, et cetera, et cetera. But how do you think about the underwriting process on like, a hard tech company versus a software company and the key factors that kind of change? Yeah, I
37:15
mean, the economics of the venture business aren’t fundamentally different. A common thing I’ll say, is, like, we need at least a 30x return on any single investment, I need a reasonable shot at a 30x return. You hope for more. You take less, but, like, if you don’t, if I don’t get at least that, then, then we’re wasting our time. And so you just, you have to do all the basic the same basics. Like, all right, what market are you going after? Is that market growing or shrinking? Like, what is technical advantage here? Like, how are you going to get there? And and so I think the fundamentals aren’t are aren’t that different? I do think that, like, it’s very easy to get yourself excited about a story of a technology that doesn’t actually work, or doesn’t actually a key problem for a customer. And that’s, that’s where I think deep tech test investing is different. The number of times we looked at something and I brought in three experts, and, like, they all say different things. He was like, there’s definitely gonna work, definitely not gonna work. Just like, and then you’ve got to watch. It’s like, and then you’ve got to kind of figure out how to get in there and and I think the most extreme version of this is, obviously, I’ve mentioned fusion a couple. I’ve done two fusion investments, one in Commonwealth fusion systems, one in excimer energy. Commonwealth is using a technique called a tokamak, magnetic containment. Excimer is using a technique called inertial containment. And that’s what the National Ignition Facility did, where you take a thing and you shoot it with a laser. And there’s a ton of other great companies working on fusion. There’s a ton of great ideas out there on things to do. And well, we looked at these two, we said, these are the two most scientifically de risked. You know, in terms of, I can look at a thing that has built, that has used basically this technique, and then they have a twist. And Commonwealth fusions case, they’re like, hey, the big thing here is, the stronger we make the magnets, the better containment you get, the closer it is together. Everything goes up. Efficiency goes way up, exponential. And so they have set the world’s record for the world’s most powerful magnets, and have done a whole bunch of really interesting work there. So I’m very bullish on like, okay, there’s something interesting. In excimer’s case, they’re like, Okay, NIF basically shot this thing with a laser. You kind of do the math on it. Basically, as you increase power in, you make the lasers more powerful. Your power goes up. Output goes up exponentially. And that’s the difference between a cool experiment and a power plant is like, we need to get enough power out for power in. And so they said, Okay, well, let’s like, NIF is about a two megajoule laser. Can we 5x that? Can we make a much more powerful laser? And if we can do that, then whole bunch of things become really easy. And so they’ve done all the work and said, Okay, what’s the most powerful lasers we’ve ever built? How are they built? They kind of, like, went through all of this stuff, and they’re, they’re in Denver now, building one. And if they build it, it’ll be transformative in that, in that field, whole bunch of technical risks in both approach, whole bunch of ways this thing can go wrong. But from a like this technique, like, they’ve got a reasonable shot, like, and there isn’t something I if I knew it now, I would say no to it. It’s just like, we just get behind them and push and hope they get there. So that’s kind of my my thesis. There’s a bunch of other ideas that are that are harder, harder to see, but I think the biggest is just like coming down is this dreaded word for entrepreneurs, which is the tea the techno economic. Analysis, we sort of live and die by this thing, which is like, you can’t do everything in a spreadsheet, but you can do enough to basically say, What do I have to believe? To believe that your product is going to be fundamentally, massively cost, competitive with everything else out there? And the easier that belief is, the easier the investment is. And that’s kind of fundamentally what we what we underwrite on that in a market that’s massive and ideally growing
40:24
so shrep, lots of recent discussion about regulation and its role in climate and energy. Bill Gurley gave his 2851 miles talk about how part of the reason why Silicon Valley was successful was due to its distance from Washington. I’m curious to hear your take on the role of regulation and energy. Is it a tax and an overreach that only emboldens and protects the incumbents? Should we have more regulation promoting clean energy? Where is the line and how does one distinguish regulation that’s a net benefit from that which is actually preventing innovation?
40:59
Yeah. Look, I fundamentally regulation is designed, and I’m thinking there’s there’s two, there’s competition regulation, but there’s safety regulation, which is probably the we think about, like, I think of how we cars and planes and other things to be safe, and I think that’s an important thing, and something we should we should have, I think for anyone who’s built anything new, like, when I’m trying to accelerate the probability to innovation. The first thing you want to do is try to remove any constraints that slow you down. And so the number of things I have to checklist, anytime I do, I have to do this, then that, then this, this other thing, it’s like, it’s going to be harder to innovate. So I think it’s just a reality of life. This is one of these places where you just have trade offs, which is like, the more constraints you put on something, the harder it is to, like, make breakthrough progress, but you obviously want constraints for safety, and so you have to find, like, a perfect balance. I’m not a like, we need none. We need too much. You know, it’s like, it’s very context dependent on what you’re doing. And I’ll talk about it in a like, in a very specific way, like, in a smart way, for example, where I think regulation, this is a smart thing, which is the NRC has and Congress have sort of made some early rulings on how to treat fusion. You know, obviously safety is a big concern when you when you think about fission, because they risk either the byproducts, the waste you have to deal with, or catastrophic event and explosion of stuff in the fusion world, if you actually look at how these things are built, they don’t have that. They don’t do that. They like they’re fundamentally they power down if you walk away. And so there just isn’t the same risk level. And so the NRC and others have said, like, Hey, this is actually a little bit more like a particle accelerator or an x ray facility at a hospital, which is like, there is radiation involved. So if you have it radiated surface, you want to keep humans away from those things. Like, that’s really important from a health and safety standpoint, but from a facility standpoint, the like, risks of this building is not really much bigger than, like, a data center, and so really just getting into the details and getting smart and saying, like, what are the actual risks? And how do we allow for innovation without taking unneeded risks, but but not overly burdened in the beginning? So I think that’s an example of things actually working. Is like we have license to innovate on the fusion side because it is
43:06
safe. So what did I miss on climate or energy? The audience should understand. I
43:11
mean, I guess one way to think about it, I’ve sort of said this, but I’ll just say it in a slightly different way, is like my experience in innovation has been like a slow march of iterative improvements that are hard to spot until all of a sudden something happens. And what I mean by that, it’s like the chat GPT moment in AI, like I started investing in building it matter of 2013 we were very bullish. I spent a lot of time between 2013 and chat GPT running around telling everyone how AI was gonna be a big deal. I spent a lot less time after chat GPT explaining that it was going to be a big deal, because it just, like, kind of became obvious to everyone, because you can, like, touch it and poke it and feel it, yeah. And so I think that like, that has been my experience. And like smartphones and the web and like all these things is just like, all this stuff’s happening, it’s kind of like, wonky and hard to understand, and then it like, gets packaged in a way that people like, Oh, that’s really cool. And I feel like there’s a lot of technologies that are like that, that are happening kind of in plain sight, in energy. We’ve talked about solar, we’ve talked about batteries, talked about fusion, next generation fission. We didn’t really get into geothermal, but there’s a lot of nerdy thing electrochemistry, which is like, how do we manufacture stuff with electrons? Basically, instead of thermal chemistry, all of these things are going down these, like, 10% a year cost downs. And so they have, they’re like, we’re like, I’m waiting for the moment where everyone’s like, Oh, that was totally obvious from the beginning. And it’s, I was big on AI back in the day. It’s like, what everyone’s once that happens. And so, so that’s the, you know, that’s the exciting thing. Is it’s like, gonna be frustrating because it’s a little hard to hard to spot. But like it seems deadly obvious to me that we’re in the middle of a bunch of really fun and exciting things. And I think that they’re like, treating this as, like, my joke is, I’m a yes and guy, it’s like, Can I have this and that, like, don’t, don’t make me make trade offs. And I think it’s a bunch of technologies that are like, yes, and it’s like, is energy gonna get cheaper for consumer? Yeah, is like. Our air gonna get cleaner? Yeah, it’s like, water cleaner, yeah? Like, yes. And we can have all of those things we do not need to pollute our air and waterways and food systems in order to, like, live in air conditioned comfort. Like, that’s the future we’re gunning for. And like, we have it is in our in our reach, and that should be really exciting. I
45:18
mean, it is amazing the time period in history that we’re living in right now, and our access to information and knowledge and automation and in to think about the future where energy is going to be a fraction of the cost and ubiquity and inference and AI is going to be a fraction of the cost. Like, it’s really remarkable. A lot of people get hung up on the fear of it, and, like, regulation kind of gets involved with, like, fear on a lot of these things. But I don’t know I’m maybe I’m just an optimist, but I’m very excited the next decade or two, and what that means for both energy, climate and the influence of AI?
45:55
Yeah, I think there’s a lot of good that can come. And that’s sort of, like, I think of it as, like, Let’s push ourselves in that direction. Let’s, like, accelerate the technologies that are going to do good for humanity and that that’s an awesome way to spend our time and an awesome future to think
46:09
about. Shred. If we can feature anyone on the show, who do you think we should interview and what topic would you like to hear them speak about? I
46:15
have a long list of people. I mean, obviously I have huge, huge favorites in every one of our portfolio founders, or portfolio companies founders. But, you know, I you know, so I think you can go early, late, others. I think that I’ll just throw out a couple of examples of people I think are inspirational So, and obviously on the sort of later stage in Bob mumgard runs Commonwealth fusion, I think is a prototype. Pickle is a plasma physicist at MIT. He’s now the CEO of a large company, but he is an operator. Now, he is not a plasma physicist playing CEO. He’s a CEO. And so I think that’s a really interesting transition in terms of what does that career arc look like? And obviously the work they’re doing is absolutely sort of on the other end. You’ve got some storied entrepreneurs who have moved from one industry into another. You know, Selena to Bucha. Walla started evite in the nineties.com. Days worked at SurveyMonkey, started Glaxo, a health startup, and a recent company is called home boost. And she basically said, let me take my product user design skills building a consumer product and help the average person, like, improve the comfort and energy efficiency of their home. So home energy of audit in a box. Instead of having a person show up in a truck with a ladder, you just like, get this little box in the mail. You run around your house for an hour and house for an hour. Ai tells you, like, oh, by the way, do these three things and you’ll make your money back in six months, and your house is more comfortable, like, and that’s fundamentally consumer product thing. So you’re applying consumer skills to others. So I think talking to people who like crossover domains, because I think climate tech, the world I’m in can be sometimes people are intimidated by it. It’s like, if I’m not a mechanical engineer or a plasma physicist, what can I do? It’s like, well, no, these are all companies. And whether you’re good at products or finance or people or whatever it is like, there’s a there’s an angle in and find your angle and get there. I could give you tons of others, but those are two. Love it.
47:55
Uh, shrep, is there a book, article or video that you would recommend to listeners?
47:59
Oh, there’s a ton. So how the world really works is, that’s probably my favorite Vaclav smell. It’s, it’s, it’s a read. There’s a lot, but it sort of does a really good job of, like, breaking down a bunch of things like you wouldn’t think of. It’s like, how much diesel is used to produce this ear of corn I’m about to eat? That’s a bigger number than you think, because water, but for that matter, yeah, water too. It’s like, you think of like, it is this nature homey thing. It’s like, no fertilizer, tractors transport. And it’s not always intuitive in terms of, like, local versus remote. Like, it actually, from an energy balance standpoint, it’s often better to have food shipped pretty far away, because you can build, grow it in a much more favorable client. And turns out, transport is a very efficient thing. Generally speaking, fertilizer is not huge energy suck. So there’s just, like, a bunch of common like, common sense, things that don’t, don’t track and how the world actually works. And I think that that, like tearing it apart is really, really quite interesting. Shrep,
48:58
do you have any habits, tactics or behaviors that are a force multiplier.
49:02
I mean, I think the idea atomic habits is another good one. Actually. I think my lesson is, is that like, habits Trump willpower. Like, that’s a really easy lesson. And like, I think all of the research on this is very clear that like, make the things you want to do easy and the things you don’t want to do hard, and like, build the habits that you get into and mindsets you get into, I think, are been in very stressful jobs over very long periods of time. And I remember giving advice to people that basically boils down to, like, make sure you sleep, make sure you get good exercise, make sure your like, nutrition is okay. You have some stress management activity, something you do, could be knitting, could be could be watching movies, whatever it is, or something you do where you get to, like, really zone out, and then, then you can be basically indefinite. You can, like, tolerate indefinite stress, because you’ve got sort of the coping combined with, like, attitudes for productivity, which is just like, how do I set myself up in a way where I can get get a lot of work done? But I think of it less as a specific do this. It’s like a bad, you know, you know. Do this one thing. It’s never one thing. The thing is, a system is like, yes, consistently optimized, I think constantly optimized, systems always beat this one thing sort of nonsense, which is just like, hey, on a like, weekly, monthly basis, look at your calendar. Look at how you’re spending your time. Is it mapped to how you want to spend your time, if not adjust and like, do that again and again and again for the rest of your life. And that’s the everything worthy of doing, unfortunately, is a never ending journey. But like, and that’s the same way of operating, I think, of operating company culture. A bunch of things like these are never ending problems that you can’t just be like, Okay, we wrote our values down on the whiteboard. We’re good. Like, no, you gotta like, work on your company culture on a literally weekly basis. Yeah, so, but that’s okay. It makes life, life fun, so habits and like, really optimize over time. And
50:44
then finally, here shrep, what’s the best way for listeners to connect with you and follow along with Gigi scale. You
50:49
can find us in all the platforms, so I’m on LinkedIn threads X occasionally, but you know, you can also find us at our website. We have lots of contact information there, and I’m hopefully showing up regularly to things, to talk to people about the exciting advances in technology and climate tech. Perfect. Well, thanks
51:03
so much for the time today and your contributions to the world of AI and social and, of course, energy too. So yeah, really appreciate it, Mike and looking forward to talking again soon.
51:12
It was fun. Thank you.
51:19
All right, that’ll wrap up today’s interview. If you enjoyed the episode or a previous one, let the guests know about it. Share your thoughts on social or shoot them an email. Let them know what particularly resonated with you. I can’t tell you how much I appreciate that some of the smartest folks in venture are willing to take the time and share their insights with us. If you feel the same, a compliment goes a long way. Okay, that’s a wrap for today. Until next time, remember to over, prepare, choose carefully and invest confidently. Thanks so much for listening.
