Mike Schroepfer, Partner at Gigascale Capital, on sustainability in emerging tech

MIKE SCHROEPFER: Show me something. Give me data. Talk to me - a customer. Give me a prototype. The more you can actually touch and feel it, or talk to a customer, the more you are actually burning down the question of like, can we do this and do people want it? And that's the two questions. KEVIN SCOTT: Hi everyone. Welcome to Behind the Tech. I'm Kevin Scott, Chief Technology Officer and EVP of AI at Microsoft. Today, tech is a part of nearly every aspect of our lives. We're in the early days

of an AI revolution promising to transform our lived experiences as much as any technology ever has. On this podcast, we'll talk with the folks behind that technology and explore the motivations, passion, and curiosity driving them to create the tech shaping our world. Let's get started. CHRISTINA WARREN: Hello and welcome to Behind the Tech. I'm co-host Christina Warren, Senior Developer Advocate at GitHub. KEVIN SCOTT: And I'm Kevin Scott. CHRISTINA WARREN: Today we are bringing you an intervi

ew with Mike Schroepfer, who spent many years at Meta, but is now working on some really exciting projects through his investment firm, Gigascale Capital, as well as some super interesting philanthropic work with Additional Ventures and the Carbon to Sea Initiative. KEVIN SCOTT: I have wanted to have Mike on the podcast for a while now because every time he and I chat about what he's doing now, I am incredibly enthused to have someone with the platform-building and technical experience that Mike

has, applying that same bag of tricks to some of these really gnarly problems that we have related to climate change. I hope in our conversation with him we're going to hear about some of those super fun things. But he can go on for hours and hours about both the entrepreneurs and the tech that they're working on that seem like they have a real, and huge potential to help us solve some of these really gnarly climate problems. CHRISTINA WARREN: Which we need. I'm glad we have people like him tak

ing those things on. So I'm really looking forward to hearing this conversation. KEVIN SCOTT: All right, let's get into it. Mike Schroepfer is the founder of and a partner at Gigascale Capital, an early-stage climate tech investment firm. In addition to his work with Gigascale, Mike started and serves on the board of the Carbon to Sea Initiative, is the founder of the philanthropic organization Additional Ventures, and is a senior fellow at Meta focused on artificial intelligence and the develop

ment of technical talent. Mike led Meta engineering teams from 2008-2022 and served as Chief Technology Officer from 2013-2022. He led the development of 8 gigawatts of clean energy infrastructure and technology and teams that enable Meta to scale to billions of people around the world and make breakthroughs in fields like artificial intelligence and virtual reality. Mike holds a bachelor's and master's degree in computer science from Stanford University. Mike, thanks so much for being on the sh

ow today. MIKE SCHROEPFER: Good to be here, Kevin. Thanks for having me. KEVIN SCOTT: You have had one of those legendary Silicon Valley careers. I am going to ask you the same first question that I ask for everybody who's on the show, which is, how did you get interested in technology in the first place? MIKE SCHROEPFER: I don't know that there was a singular origin story. In a funny twist of fate, I grew up in Boca Raton, Florida, which people may know from two very different worlds. One is Se

infeld, who lampooned it with Del Boca Vista. And the other is, obviously that was the birth of the IBM PC. If you read the origin stories of Microsoft, which you know well, you hear the stories of Bill flying into Miami airport and driving up to Boca Raton in the 80s to convince IBM to use DOS for their IBM PC. That was my hometown. I was there at that time as a kid. We had, in the career days at school, someone from IBM would always show up and talk about computers. My neighbor's dad worked at

IBM so they had a PC Junior, I remember because it had the amazing 16-color graphics, I think. We played King's Quest on that. I think a lot of it was just exposure plus video games. My neighbor had the IBM PC. We couldn't afford that but we had a Commodore VIC-20 and then a Commodore 64, which was the bomb. So I had just earlier exposure to that, and that was as a young kid. Then as I got further on in school, science and math was always really fun for me. I think the real "ah-ha" was in high

school, I had a chance to take a physics class. This was just like, there's something about, like wait, with these five equations, I can explain how objects move around in the world, and I can understand how light bends through a small pinhole. It was really fun. I think that probably was the cincher on, when I go to college, I want to do something in science or engineering. But I wasn't exactly sure yet till I got to college what I wanted to do. KEVIN SCOTT: Were your parents in science and tec

hnology? MIKE SCHROEPFER: They were not. They were in the radio business. They ran a really small little AM radio station. I actually had an FCC license to broadcast as soon as I was old enough to do it because I would help them on nights and weekends, run the radio station. So there was some technology, like we had early CDs and things like that to help upgrade the radio station. My dad was very tech-curious, liked gadgets and stuff. But neither of them had an engineering or science background.

KEVIN SCOTT: It's really interesting. A lot of people that are in our age range had a somewhat similar story. They came of age, right around the time the personal computing revolution was starting. They also, were teenagers or kids as video games were on their ascendancy. Those two things got many a computer science career underway. But I'm wondering, did you have role models in your schools? Was there any one person who was, because this is interesting, I don't think I did, I grew up in rural

Central Virginia, I can't point to one person. MIKE SCHROEPFER: No one I knew. It would be people I saw on TV or read about in books or things like that. But I'm trying to think, and that is not to say I did this all myself, it's more like there wasn't a person I was like, I want to be like them when I grew up that I knew in person. So no. KEVIN SCOTT: How did you get from Boca Raton to Palo Alto? How did you choose Stanford? MIKE SCHROEPFER: Well, I knew I wanted to do engineering or I thought

I wanted to do engineering but I wasn't sure exactly. When you're in high school, it's like, what does engineering mean? I liked computers and so it was basically just Stanford was good at everything. They weren't as well known but they - like, all the engineering disciplines, were good. They're actually good at most of the humanities too. And they didn't make you choose before you got there. It's like oh, this would be a great place for me to really understand which of these different disciplin

es. I looked at a couple of all engineering schools and I was just like I just, I don't know. What if I decide I want to do philosophy or something else? That was really the drive. I also just when I visited it, I did the little like they have after you get accepted you do the overnight visit where you get to stay on campus. And I did it there and at Princeton which was the other place I was considering and they got me right. They got me in a dorm with a bunch of nerds and I was like, oh, this i

s awesome. They're all nerds. They have Macs, we're talking about computers. And it's sunny out. This is amazing. Clearly if I had this opportunity I'd be a dummy not to take it. KEVIN SCOTT: What was it like once you got to Stanford? When was this? MIKE SCHROEPFER: It would be '93, 1993. KEVIN SCOTT: Got you. That was a super interesting time at Stanford. Just a bunch of stuff going on in Silicon Valley and a bunch of interesting people at Stanford. What was that like? MIKE SCHROEPFER: I just t

hink we, I remember my freshman dorm. It's like one person down the hallway from me had an Apple Newton. Do you remember the Newton? Pen-based computers? KEVIN SCOTT: I loved that device. MIKE SCHROEPFER: "Whoa, you have a Newton?!" And it was just, for whatever reason, the computers were backordered but I brought my computer with me so I had the first computer there. We didn't have internet in the dorms. We had dial up. There was a little computer room in the basement that actually had Internet

among other things, It had a NeXTcube in it. If you remember the old NeXT, because I think NeXT basically donated a bunch of computers to Stanford to try to get people to actually buy them. It was just like, wow this is so cool. There's just so much stuff to do here. I went through - and you'll hear this theme throughout my life - how do I run experiments to figure out, like, I want to do engineering? What does that mean? Let me just start taking what's the closest proxy to this. So I took the

Intro to Computer Science class, the Intro to EE class at the same quarter. Because I'm like maybe I want to do electrical engineering and build computers and chips, or maybe I want to do the CS thing. I took them in parallel and - both amazing classes, CS 106-A, and E-80, I think it was in EE - and I was like, by the end of E-80 were like - we built like a counter or just we spent all this time because it's hard to build circuits, and every week in computer science we were building something n

ew. It's like, "oh, here's a basically Google Maps directions thing. Here's a blackjack game. Here's..." I was like "this is a very high leverage tools. This is amazing. I sit in front of my computer and then three days later this new thing pops out the other end that is a fully useful piece of software." I basically I was hooked at that point and that's what's kept me in the industry for 25 years. It's just that feeling of like "whoa, nothing to...we made this thing." That's a lot more fun than

any other hobby I can think of. KEVIN SCOTT: I'm going to want to get back to this in a minute. But I think when you were at Meta and I suspect even with what you're doing right now, certainly everything benefits from being able to operate at software speed, but it also at a certain level of scale like you actually do have to think about the hardware again because you just can't pick off the shelf like a quantity or form factor compute or networking or storage that meets the needs of the thing

that you're trying to build. I totally understand - I made the same decision myself - but then found myself much later in my career caring a lot about hardware again. MIKE SCHROEPFER: Me too. We can talk because again it was just such a fortunate time because the Internet, the World Wide Web, we already had Gopher and a bunch of other things, and Telnet and what was it, Usenet/NNTP, when I got there, but then the web obviously came out while I was at Stanford and so that was just amazing. But on

the hardware side I agree with you. When I joined Facebook now Meta in 2008 those first couple of years -- everyone thinks of Meta as a software company. I was like, those first couple of years, our biggest problem was the site was just growing so quickly that we had all these emergency fire drills, like we're literally going to run out of capacity and everything's going to crash. So, a combination of software optimization and how do we uncork the hardware growth rate thing? As you well know as

well as I do it's, you can't just order a new datacenter and have it next week. You're like, "oh crap, we gotta..." And I remember in these conversations it's like "OK we're going to build a data center." We have a 2-4 year lead time between the go and it's operational. How much capacity do we need? I was like, I don't know, it's so hard to predict that far in advance. We literally have to order steel now and buy some land. You can't tell me this in nine months. You need to tell me now how much

we are building. Then we built lots of datacenters and then we made another move into consumer hardware. I joke I made two transitions - I went from software to hardware and then when we got into consumer I was like, oh, we've been building datacenters, I know this hardware thing. This is easy. And it's like, oh, no. It's a totally different game when, if the consumer returns the thing you've now lost a ton of money. Because you just ate everything. And so making sure that they like it and the

re's customer support and your price - there's a whole different ballgame in consumer hardware. So it's like: software, enterprise hardware, consumer hardware in terms of difficulty I would say. That has been really fun and has actually informed a lot of, in my climate tech investing we're investing mostly in hardware and a lot of it is enterprise but some of it's consumer, selling products that your average person is going to buy. Anyway, we can talk a lot about that. It is definitely a shift a

nd a lot of fun. KEVIN SCOTT: There's an interesting philosophical thing here. I think one of the things that has made you so successful and the companies that you've worked for so successful is there is this underlying assumption that no matter what the problem is or the thing that you're going to go tackle next, that you're going to be able to figure it out. But sometimes you can let yourself be overconfident. Managing that balance between competence and over-confidence. You do have to be fear

less. If you are fearful you will never get anything done and it's one of the things I'm most impressed by what you're doing right now because some people look at this bundle of climate change issues and think that the problem is completely intractable and yet you are just, with great enthusiasm, you are trying to leverage the best of this technological entrepreneurial mindset to try to fix this. But have you thought about this tension? How do you temper your optimism with pragmatism? MIKE SCHRO

EPFER: I often describe myself as a grounded optimist or a practical optimist. I think the extreme of optimism is naivete. It's just, it actually doesn't work. I think so much of what you say is true. I think there's this phrase "stop energy," which is you're like, "oh, Schrep, I have this idea!" You're like, "oh, here's all the reasons you can't do that or that's not going to work." Stop energy is so easy. It's so easy to be a critic and say "I've thought of 30 reasons this doesn't work." It's

so much harder to basically say "Well let's think about this. Why would this work?" Let's talk about the physical constraints that means that this is not possible. Do those exist? Are you positioned well in the market? Are you leveraging your strengths? All of those things. It's a much more constructive conversation. The good news, and getting back to this hardware software thing, is the fun thing about hardware is you can model a lot of things out on paper. If I told you I wanted to build a chi

p with this much memory bandwidth and this many transistors you can do some back of the envelope and you're like "you can't fab that right now." You can try to convince me all you want, but TSMC can make it. Or in AR/VR worlds, you have these headsets and a big challenge is you need to get the field of view to be big. I have a big screen in front of me. Well, you can model: I've got to get a photon from back here from some light engine. It's got to bend around and hit my eye. There's only so man

y ways to get that photon to bend around and optics is really well understood in physics. You can do index of refraction. You can model given the material we're using, the maximum field of view you can get is x. Is that right? I can't tell you how many startup pitches I went through where they're like we're going to build it with this, using this thing. We'd sit there and be like, you can't. It doesn't - you've violated laws of physics at this point. I think, you're like, "are you violating laws

of physics?" As the starting point. But there are plenty of products that don't violate the laws of physics but you just you don't understand your customer. I think a lot of businesses try to go way too far out of their comfort zone. When Intel tried to make microscopes, that didn't work out well. It's very far from chips. You always want to take one step, like "I know part of this business, I'm just moving a little bit to the left" as opposed to "I'm switching industries entirely." There's a l

ot of questions that you try to ask yourself. Then the other thing I'd say is, is the constant question I ask is "what's the cheapest, fastest, easiest experiment to learn more?" Can I mock it up in cardboard? Can I go ask 30 people whether they want this thing? Can I build one in my garage? This is the place where I think smart people really get themselves in trouble. You try to spend too much time imagining and it's just like: stop imagining, go do it. Show me something. Give me data and talk

to me - a customer, give me a prototype. The more you can actually touch and feel it or talk to a customer the more you are actually burning down the question of can we do this and do people want it? That's the two questions. KEVIN SCOTT: Yeah, it's really interesting. I think there's a related thing with smart people where sometimes it's very enjoyable to wallow in complexity, to take a very hard thing and even to make it harder or like there's this joy you can get from spending cycles there. B

ut like those overly complicated things are almost never really useful. MIKE SCHROEPFER: A hundred percent. I think - I often describe when I'm working with people, this took me a long time to figure out, is I know there's complexifiers and there's simplifiers. KEVIN SCOTT: Yeah. MIKE SCHROEPFER: There's someone you give a big hard problem, they go like, here's 30 pages, but you really only need to understand three things. Like here's the three biggest things that matter here. KEVIN SCOTT: Yeah.

MIKE SCHROEPFER: If you want to get into details, I got it. But here's the thing. There's other people that come up with a, here's 26 pages of detail. I've covered every base on this thing and you're like, that's not actually helpful. That's actually much worse. I find simplifiers are a secret weapon of a lot of organizations. It's what we sought in our PMs at Meta. That's what I look for in founders I back. And it repeatedly has been successful for me in finding people who take a big complex g

narly thing and say, "but these are the only things that matter." KEVIN SCOTT: Yeah, I mean, I feel like you're giving the listeners sage advice here. So like you compound these things and they get very interesting. Folks who have high learning rate, who know how to experiment quickly, who are simplifiers, like you just stack these together and really the union of those things are just superpowers. MIKE SCHROEPFER: Yeah, 100 percent. KEVIN SCOTT: Let's go back to Stanford. You're there in '93, y

ou decide to take computer science over, over EE. What was the most interesting thing that you learned at Stanford? Like whether a class or something you learned from a friend or an internship? MIKE SCHROEPFER: Yeah. I mean, I think it gets back to your question on like, how do you have no fear about having to solve a problem? Which is like, I think the best of computer science is like learning how to organize and decompose complicated things. You go back to like, what is a data structure? What

is an API? How do I layer things? How do I like...okay, I can program in Python. I don't need to know what the machine is doing, but then I can start to learn C. I can learn assembly language. I can look at the transistors on the chip. There's this very clean layering that allows me to abstract out complexity. I think that that is probably by far and above the best skill because it gets back to the simplifier/complexifier thing. It's like in everything in life, you have to be able to figure out

how to do that. If you're doing a climate tech investment, Someone shows up with this magic box. This magic box in which air, water, and electricity comes in the side and jet fuel comes out the other side. Like, okay, I'm going to pretend for a second. I don't even know how this box works. I just trust you it does. If it costs you $1,000 a gallon to make that jet fuel, you do not have a business, right? KEVIN SCOTT: Yeah. MIKE SCHROEPFER: I don't even care how the box works, but if you come to m

e and say "I can make it for five bucks a gallon." You're like, Oh, I have a bunch of questions about how that box works. And let's tear apart whether that box works, whether it scales, and it's going to be reliable. And so like being able to move at multiple levels of detail and move up and down is by far and above the best. I think CS also in a weird way teaches you - like I had this feeling even in the '90s - it was like, wow, this field is moving so fast. Like languages were showing up all t

he time. Chips were moving and so it was like the idea that you just like learn your thing and then spend the next 30 years applying my thing was not the lesson I got. The lesson I got is like you need to learn how to learn stuff. Because the thing you've just learned is going to be obsolete in like a year. If you want to have any longevity in this field, like you got to be constantly -- you've got to know the new language, new framework, the new, this, new that. And so that kind of sense of lik

e, I just, it's a meta process. I've got to get good at learning. I've got to get good at decomposing problems and understanding what's important. I feel like those skills, I was like, I think everyone can do that regardless of their talent and their background and their profession I mean, and I think those have served me incredibly well through throughout my career. KEVIN SCOTT: I have another question I want to ask you, but I think the other thing that people underappreciate is practicing lear

ning makes it easier to learn. I feel like it's one of the unfair advantages that I got as a computer scientist because yeah, like I've always assumed the half-life of the technical stuff that's going into my head is very short and so like I'm always. But like it lets you learn a whole bunch of other stuff. I'm guessing in your case, like it let you be fearless about going into climate investing. Like even though like you just: Okay, well I know how to learn, like I can go learn this stuff obvio

usly with humility, because it's very complicated. But I see it even with my children. Once they figure out that learning is fun and useful then they are in the learning loop and it just snowballs into something like very interesting. MIKE SCHROEPFER: Yeah. Exactly right. KEVIN SCOTT: The other question I have for you about Stanford is, was there a course that brought everything together into like, oh my god, like I didn't realize I had this superpower to build something very complicated before?

Because some schools do it Mr. Miyagi style, like your wax on, wax off. And then one day you're fighting in a karate tournament and you didn't even know you were going to be able to do this thing. And it's like maybe a compiler course or an operating system course or robotics course or something. MIKE SCHROEPFER: I mean, it's honestly, this is a crap answer, but it's like I loved so many of the classes in the curriculum. It's like hard to pick one, so I'm going to pick several for different rea

sons. I think the thing that probably stands out is the introductory computer science curriculum at Stanford, which is CS106A, and B. Or if you knew how to program when you got there, CS106X is phenomenal and it's phenomenal because it does this layered approach where they give you progressive revealing of more details. They also make you do real projects. But the real secret sauce is they had undergrads as basically TAs. Then it's this thing called the section leader program. The undergrads, gr

ading the assignments and grading the tests. Once you get through CS106 A and B, you can become what's called a section leader, which I did very quickly. This was just a, because it's one thing to learn something it's another thing to teach it to other people. Holy cow. They're asking questions. You're like, oh gosh, I guess I didn't understand that. Then I eventually became a TA when I was a grad student and then had to make assignments and that's a whole another ball of wax. Wow, I gotta, from

scratch-- Like how do I make an assignment that teaches people about data structures? Like that whole process of leading, teaching, working with others. It was like, again, I think the fusion of the thing like because I haven't, I'm an engineer, I'm a technologist by background, but like my career ended up being a lot more about people, actually, people plus technology. KEVIN SCOTT: Yeah. MIKE SCHROEPFER: Then that was the very beginning of this is magic because of all of these people. You get

all these really great motivated undergrads and they're excited to teach it to other undergrads. It's just magic, that was probably formative. And then there's a bunch of just the like because it was just fun to like, you're just peeling layers of the onion back like, oh, that's how that works. Oh, that's networking, compilers, computer graphics, operating systems. Like even I took, they'd stopped doing assembly, it was no longer required but I took it anyway. And we made this like multitasker o

n, it was a 68K assembly, and that was really fun too. There was a contest of how few instructions you could make one thing happen and I spent a lot of time trying to win that contest. I don't know there's a lot there that was just fun. KEVIN SCOTT: So you wrap up your time at Stanford and then it's off to industry. How did you choose? Because again, I'm sure there were so many options like all these startups, we have all these big companies, like there was so much going on in the industry. Stan

ford students have this crazy privilege of being very highly sought after. I'm guessing you had the opportunity to do a gazillion different things like how did you pick? MIKE SCHROEPFER: Again back to experiments. I definitely treated my internships, my summers, as like a chance to run high-frequency experiments. I was like, let me try lots of different things, just trying to dial in what it is that I want. First summer was honestly whatever I can get as a freshman to sophomore, it's like, it's

really hard to get internships. I was like whoever will give me a job in tech, I'll do it. I'm working at Motorola back in Florida. Amusingly in their factory where they made beepers. But I was on the software team helping them with compilers and stuff. Then the next summer I was like, I want startup experience. I went to Austin to work for a company called Trilogy Software that was about 100 people. Enterprise software, boring software but an exciting startup. Then as I got further in my studie

s, I hit my first love in computer science, which was computer graphics. I was just like, computer graphics is so cool. You just like, again, can make beautiful things. We're like, ray tracing was a thing. And how do you make it look beautiful? My next job was at a computer graphics startup in Los Angeles, working on movies. Then the summer after that, I was like, that was really fun. Let me try doing computer graphics at a big company. So I actually interned at Apple on their QuickDraw 3D team.

This was the 90s of Apple, which is not a good time at Apple. Actually, my whole project got canceled while I was there that summer. That's a very funny story that we could talk about. Then I was really into computer graphics. The big industry conference is called SIGGRAPH, so I went to SIGGRAPH that summer and this is going to date me. But because we were still in the early days of the Internet, there was literally a resume board where you put your resume up and you would tack it up there and

then people - And so I was like I'll put my resume up for when I graduate because I was going to go to stay for an extra year and do my masters. When I graduate next summer I can get a job and I got a call from a recruiter who was working on this small company that was doing computer graphics for movies. It was one of the main people in Industrial Light Magic. They were using Macs to do real-time special effects. I was like well, that sounds interesting. I went and I talked to them and I thought

I was going to be interviewing for jobs in the next summer. And they're like, no, we want you to start right now, we're a startup, we can't wait a year. I decided to defer my master's degree because I was like, I was engineer number two this little startup working on software for digital effects. Which felt like just the most amazing thing to me. That's how I started. KEVIN SCOTT: Well talk about that a little bit. I think both of those things are fascinating experiences for someone in college

to have. Like both your getting your project canceled while you're an intern, and like being employee number two at a startup, like they're just fascinating learning experiences for different things and I'm sure valuable stuff for everything that you did after. MIKE SCHROEPFER: My recollection is this was '96 or '97, at Apple. Apple was going through tough times. They were cutting projects here or there. About halfway through the summer, my manager who was amazing, calls me in his office and he'

s like, hey, I want to talk to you. My recollection of this conversation is something along the lines of, hey, so this whole project is going to get canceled and everyone's going to get reassigned or laid off. But don't worry, like your internship basically is funded through the summer. it was like, I need you to come into the office and do some stuff, but none of the code - we're going to throw away the project, so none of the code you've written is going to be used for anything. You would thin

k as a 21 or whatever year it's like, come in for a couple of hours and have the rest of the time off would be great. KEVIN SCOTT: That sounds awful. MIKE SCHROEPFER: I was miserable. Probably the saddest i had been in any work context that I can remember. Because I was like none of this matters. I really wanted to go, I want to build stuff that people are using. I was just told that it's impossible at this point. That was a, it's kind of a linchpin in my mind of both - like it really matters wh

ere you are, the macro really matters, economic conditions, how good the company is doing. And like wow, I don't like sitting around, I like building stuff that people use. That was one. Then the startup was a lot of fun because it was, I joke, I constantly got handed tasks that I wasn't qualified for because there wasn't anyone else to do it or I was stupid enough to volunteer. It was just a chance to try a lot of different things and write a lot of code. I think it also created this moment tha

t is like seared in my brain, which is...Okay, it was used for a terrible movie. I'm just going to caveat that I don't think this is a great movie. I'm going to offend a lot of Star Wars fans. Our software was used on the Phantom Menace and they were making it at the time. And Skywalker - we were in Sausalito, which is up in Marin County, just north of the Golden Gate Bridge. ILM at the time was up in that corner of like Lucas Ranch just whatever, where they were doing a bunch of the specs work.

The founder of the company came up to me, he's like, hey, the folks over at Lucas Ranch are like using the thing you've been working on and they need some help. Can you go over there and help them? Here's me as like a 22-year-old, like driving my Honda Accord or whatever it is, over to Lucas Valley Ranch, like signing this NDA, which like I always joke was like somewhere in there was embedded like if you reveal anything, we have Jedis who will come get you to sign this thing and then like open

the door. There's like figurines from the movie on a giant table in front of you and a bunch of people in front of their computers. They were working on the one of the palace scenes with Padme. They're using literally the code I'd written, which was this system around motion tracking. How do you find that object in the scene and then apply effects to it. It was just one of these like holy crap. Like they're using my thing for this. This is amazing. It was just like, again, back to like, there's

nothing better than this. There's no like hobby I can think of that's more fun than we're building this thing, and then someone else is using it to do something awesome. That more than anything was a formative experience. I think the next lesson I learned. The unfortunate thing is, this was an amazing company with great people. Our software was like revolutionary. We were in a limited market. We like sold it to everyone doing special effects, which was 10, 20 thousand people. We had this like no

w what moment? You we struggled to try a bunch of different things, and again, back to the macro it's like the macro conditions really matter. It's like agreat team, great company, great this, but like we're selling a product that there aren't enough customers for and that eventually make things hard. And the company eventually sold and I decided to go off and do something else. KEVIN SCOTT: We don't have we don't have infinite amounts of time, like you've done so much interesting stuff. I'm goi

ng to fast-forward past like a bunch of things, including like what you probably are most well-known for, which is your 15 years at Meta. But maybe one thing there, 15 years in Silicon Valley time is an awful long time to be at one place and you still are there as a Senior Fellow. What is it about Facebook/Meta that held your attention for as long as it has? MIKE SCHROEPFER: I think it basically boils down to three things. One is it was never the same thing, the time I was there. When I joined i

t was Facebook.com. Fewer users than MySpace. Web only. Hundredish engineers. 15 years later, it's billions of people using our product, it's Facebook, WhatsApp, Instagram, Oculus VR systems. It's an AI Research Lab and it's a much bigger organization. You can imagine all the steps along the journey there. There was a lot, as we talked about earlier, how do I build a datacenter? How do I build four of them in parallel? How do I build 20 of them in parallel? How do we build a research lab? How do

we sell consumer hardware? Like all of these things were like, this is like my graduate school. I get to learn a bunch of things. It was constantly, is constantly changing. Even now, why am I still there? We built PyTorch, which is the leading open-source framework with a great cooperation with Microsoft, huge partners in that. Then Llama is the leading open-source model out there. I think it's been downloaded like 100 million times or something and there's 19,000 forks of it. It's like, hey, i

f I could spend a little bit of my time and help build a technology, AI, launch language models that can be leveraged by a lot of people because it's open source like that seems like a worthy use of my time. When it's different than what we were doing five years ago. So, it has constantly changed is Number one. Number two is I love the people. Mark is just unbelievable and there's a lot of lessons from him, but it's not just Mark. There's tons of brilliant people there. Then the third is this im

pact. Great, I work on this AI model, but it's used by so many people like that is a great way for me to again, get back to that. I do some work, we build something. A lot of people do awesome stuff with it. That's like a throughline through my whole career. That's why I'm still spending time. KEVIN SCOTT: I think there's this other interesting thing too about your work. Even if we - you were CTO at Mozilla before you went to Facebook. And so much of what you have done has been platform building

, like you're building like apparatus for other people to build on top of whether it's like Facebook's internal infrastructure or Facebook itself. Web infrastructure like, you know, I hadn't heard this anecdote about your computer graphics startup experience before. But like even that's like a piece of infrastructure that other people are using to build things. What is it that has attracted you to making tools and infrastructure, and systems and platforms? MIKE SCHROEPFER: Just one small clarifi

cation. I was VP of Engineering at Mozilla. We had two amazing CTOs while I was there. I would hate for them to think I was them because they're amazing. It's a really good question. I think there's just something about it that intuitively, like, I love the idea of leverage. Technology is leverage. I always say that technology is one of those few things that removes constraints. So many problems in life you know, Economics 101, you take in high school where it's like all right: you have a $100 c

ity budget. You can either fund the libraries or the police or fire department, but you can't fund all three fully. A lot of people live in a world every day where our problems are tradeoffs. I can do this or I can do that. Technology is one of the only things that's like, Oh hey, it's now half the price. Like these lithium ion batteries are actually 99 percent the price they were when they were introduced on the market in 1991, 99 percent cheaper, like 1 percent of the price. They're still gett

ing cheaper. You're just like, I show up with this thing. It's just like better. That is awesome. I think that there's, for whatever reason, I think I have a decent intuition. Platforms are hard because people like to think concretely like, what exactly am I going to use this for? What's the product, Who's the customer? When you're building a platform, you have to have a little bit of a leap of faith or an ability to believe, okay, here's what this is going to be used for. For whatever reason I

think I'm just pretty good at spotting - Like I think this is the need and I think if we build this, people will use it. If you look at my time at Meta, it's React, it's PyTorch. Those are the things that I'm probably most proud of because they are things used by millions of people around the world that fit a need better than anything else out there. It's just I don't know what it is. I think I just have like a natural attraction to these lever points that just provide tremendous value for peopl

e. KEVIN SCOTT: Maybe this is the perfect segue to what you're doing now. In 2023, you co-founded Gigascale with Victoria and Evaline to invest in, and I'm guessing, accelerate, the development of technologies that will help with climate change. I've got to tell you, man, every time I talk to you about what you're doing. I am so much more hopeful and optimistic about things than I was like five seconds before I was talking to you. I mean, I guess why focus your attention on this? Which may be an

obvious question, and what makes you think you can have leverage there? MIKE SCHROEPFER: Let's have some real humility on this. I think it starts with why, and I think, I bet you a bunch of people watching this are maybe thinking about this or struggling about this. I think it's an important point and it took me a while to figure out exactly what happened. But it was honestly during COVID, it was 2020. The whole world was shut down. No cars on Alma street here in Palo Alto, which is usually rea

lly busy. I work lots of hours. I had a weirdly, all of a sudden, extra time because nof ot driving the kids around activities, everything's on Zoom, so we just had a couple of extra hours in the day. It also just gave me this moment to reflect as the world was in this crisis then I felt like: What is my role in this world given everything we just talked about, like tremendous opportunity and fortune, what am I going to do with it? We'd already been doing a bunch of, as you do, philanthropic wor

k in a variety of areas. But I was like, man, climate, it's a platform problem. It's going to impact tens or hundreds of millions of people, and the people most impacted are the least equipped to deal with the impacts. It's like here I am with a bunch of resources. Isn't just an obligation for me to go off and do this? That's where it started, and I started thinking it was going to be philanthropy. I was like, great, I'm going to direct more of my time and attention to philanthropy, spend a lot

of nights and weekends. Again, I got to have a learner's mind. It's like, I don't know anything about this. I'm just going to learn really quickly about all of it. That was a huge advantage because they just didn't have all of the baggage from before. That's where we started doing a bunch of philanthropic things. Spinning out a non-profit last year that's working on a form of ocean carbon capture and funding a bunch of early-stage science there. Then it's again, back to high-frequency experiment

s. It was like, okay, I love doing that and I bumped into some entrepreneurs, like you're doing some really interesting stuff, and as I just went through it, it's like look, we need to be spending trillions of dollars a year to rebuild our physical infrastructure. How we make energy, how we consume, and how we transport things, how we make food, how we live in buildings. That is not a problem that a government or philanthropy can tackle with direct investment. Like you need the markets to basica

lly, markets can do that. Markets can spend trillions of dollars a year. We spend about that in oil and gas right now. Markets do that when there's money to be made. I think that was like, okay, great, where's the money to be made? And how do we take technological disruption back to my, huh, now this thing is now half the price it used to be. I can now disrupt an incumbent and they don't know it yet. I was like that's where I get excited. It's like cool. We have new technology. We have these cur

ves of batteries. We've got genome sequencing. We're working on a vaccine for, for cattle to reduce their methane emissions. We've got solar cells, all of these things on a massive cost down curve. Then you start asking your questions like, where are the disruptions there and who are the people that are going to do it? That's really the foundation of it, and then taking all of this experience of building hardware, teams, technology, and using it to help, when I find a great entrepreneur who's yo

u know, you've got Sarah Lamaison, who's building an electrochemical cell to make ethylene. When I'm making the pipes in your building, instead of doing all this dirty, nasty stuff with fossil fuels and emitting all these things, I have this nice little cell that just has electricity coming in. Then I make the chemical. And that's really cool. But the cooler part is we think we can do it cheaper. The pitch to the customer is like, yeah, our thing's cheaper. Oh, and it's also really good for the

environment. Like that, I just get so excited about. And I just want to spend my nights and weekends helping Sarah crush it because I think we have tons of things like that. That is leveraging a bunch of technological curves that are happening very quickly. KEVIN SCOTT: I think this thing that you said about capital investment in markets is very interesting. I wonder a lot about how you make sure that you have the right incentives set up inside of the market, where the market is playing the righ

t game. Because look, you can spend trillions of dollars a year, and if you are not spending it efficiently, like getting it into the hands of the people who are most likely to make the big disruptive breakthroughs to encourage the right set of things which is not inventing stuff, it's like inventing stuff and then deploying it at scale and making the unit economics of everything work. I'm a big believer in markets and I think markets can help, maybe it's the best top-level mechanism for making

sure that the capital gets allocated efficiently. But is that a thing that you worry about? MIKE SCHROEPFER: Yeah, and I think we're at different levels of maturity at different technological stacks here. I think you've got to attack the problem from both sides. Meaning I can disrupt the market because my technology innovation is now half the price of the incumbents'. My low-carbon thing is just cheaper than the high carbon thing because of technological advancement. Yep. That's, that's the easy

button in, right? Then there are other places where you need to bootstrap it a bit. Like maybe I'm not cheaper yet, but when I 10x my scale, I am cheaper. That's where you see government incentives, whether it be an EV tax credit or getting charging capability setup, or carbon emission taxes for, whether it's transport or aircraft or concrete or cement, steel. I think there are places where the governments can accelerate those things and get them there. I think that the useful thing about - I a

m doing both philanthropy and investing, and I very clearly separate them. And philanthropy is like great, we're going to fund early-stage science. We're going to fund policy work. The output of that is a public good. It's something that everyone can see and use. There's no money to be made here. Then in the investing hat, I'm like, I'm only going to invest in this company because I think you have a business, right? You have a technology that at scale is fundamentally better or cheaper than the

other alternatives out there. What's surprising to me is like, there was a question in my mind is like, is there a lot of those or not? The answer is there's a lot because there's so much inefficiency in the current market. I talked about Dioxcycle, which is making ethylene cheaper than high carbon alternatives. I'll give two other really quick ones. Arbor Energy, a bunch of SpaceX engineers, know how to make rocket engines and gas turbines. They're like huh, all this like forestry waste we're p

ulling out of the California forest because we want to try to prevent forest fires, like all that stuff either piles up and sits and rots, which makes methane, which is bad. Or you put it into an old school biofuel plant, which is like a big, huge wood fire. Terrible particulate pollution. They're like, "we spent all this time doing these high pressure, high oxygen burn things. If we can build a high pressure, high oxygen burn, we can basically take that same fuel source, and out of it is water

and a fairly pure stream of CO2. Water, we can actually do something with. That CO2, we can inject in the ground or use as an input to another chemical process. And by the way, we're making energy." It's like, you can think of it as like energy-producing carbon capture. We capture carbon and we make energy. That's cool. Then you've got another direction with like a company we just invested in, Arch. You say, Look, we've got heat pumps. I'm trying to heat and cool my home. Air conditioning, heati

ng, this magic heat pump technology is just like strictly more efficient than everything else out there. For most consumers, if you do the math, you're like it's a little bit more money up front. But your bills go down every month. You ask yourself, like, why isn't everyone installing these? You start going out and you talk to the installers, and they can't answer simple questions. The consumer is like, okay, this thing is $5,000, this thing is $10,000. You say it's cheaper. How long does it tak

e to pay back? And they go [shrug] So of course everyone goes, "I'll buy the cheap one." Now you have a software solution that can do all this for you. Say, Oh, your payback period is 37 months. Do you want to do this? Yeah. Like hell yeah I want to do that. It's a massive inefficiency in the market you're attacking with technology and like, and at the other end of it, a whole bunch of people are going to make a bunch of money. Installers are installing more heat pumps, consumers save money, lik

e it's just good. Yeah, that's what gets me excited. It's like every time I look at this as like, oh, that's pretty good opportunity right there. And we haven't even talked about the grid, which is a whole other thing which we could spend another hour talking about KEVIN SCOTT: Well, let's talk about the grid actually because I'm sure on multiple dimensions, the grid is a source of concern for both of us. You know, If you look forward a handful of years, the amount of energy that the world will

need is just much higher than it is right now. You can even imagine worlds where if you had abundant, cheap, sustainable sources of energy, like you really could change a whole bunch of things about the world. Like you wouldn't have water scarcity anymore, like drinking water scarcity. But like the grid is like this sort of interesting, interesting thing. Like given the current momentum, like you're only going to have the electric power industry like build this amount of capacity in a particular

way. Like very few electric power companies, like the operators, have R&D functions. Like there's nobody really doing a ton of R&D about changing, like how it is that you're actually doing generation. Like talk about that. I mean, you must have thought about this way more deeply than I have. MIKE SCHROEPFER: Yeah. This is both a deep concern and a massive opportunity. I think that the short way to think about it is like electrical demand in the United States has been relatively flat over the la

st two decades-ish or decade or so. That's good news because it basically means it's actually decoupled from GDP. We've been using about the same amount of energy and growing our GDP mostly because of technological efficiency, LED light bulbs, things like that, which is awesome. But it also means we've lost the muscle on building lots of generation and distribution capacity. If you look at other countries like China, they're like growing their grid capacity at a, at a massive rate, much faster t

han the US is. You start with like, are there physical laws and limits? It's like no, humanity can build this stuff really quickly. We just for a variety of historical reasons plus choices, plus regulations aren't currently. It's like good news, not violating laws of physics. Then you say, increasing demand, that's one side. We're building a lot more factories in the US now. We have datacenters working on AI, like that's creating increased demand. There's also this new set of supply that's very

different than the old set, which is like wind and solar. If you look at ERCOT, Texas's grid, you know on noon, about 50 percent of their powers, wind and solar. Wind and solar is amazing. It is the cheapest form of energy generation we have ever had as humanity, but it doesn't work 24/7. Everyone knows that. So now I have this new variable production and this increasing in demand, which are both new to the grid, which creates disruption. Disruption equals startup opportunity. There are so many

different ways to attack this problem and they're hard, but I'm excited about it. On one level, grid storage is huge. If I've got these - installing a commercial grid solar array is the cheapest form of energy generation and the most likely to be delivered on time, on budget of any project we know how to build, because they're so simple. Like people are building these at massive scales. We're in the - we're getting close to a trillion dollars of investment in a year. If I can then show up and sa

y, Oh, hey, I have this thing, I can put it right next to your solar array that allows you to dispatch energy 24/7. And by the way, people will pay you more money. They do this in Texas, you know, at 8 P.M. when it's peak power load. Like, go ahead and charge the battery when energy is cheap and then discharge it when energy is really expensive. Like that's a really good business. Now there's a huge incentive for me to build grid storage in a way that 10 years ago, nobody cared. Nobody cares abo

ut grid storage and batteries. But like so now we have companies like Form Energy, it's like huh, instead of using lithium ion batteries, I'm going to use this iron oxide battery. It's really cheap to make. It's bigger, it's heavier, which doesn't really work for a car. But who cares? It's on a big concrete pad. It's super cheap. Like that's going to be a great business. You have a bunch of other people saying, Oh, power lines are like rated statically, which means you rate them on how much powe

r you think they can hold. But like, depending on how cool it is outside, you can probably put 30 percent more power safely through that. Maybe let's just put sensors on the power line and say like, it's cool to go 30 percent more power, it's fine. Like magically now I have more power. There's other companies saying like superconductors now work. We can 5x the amount of power through this line via superconductor. There's so many different ways to attack that problem then I think the disruption c

auses the opening for these business models to exist. If I had superconducting power lines 10 years ago, nobody would care about a grid battery and nobody would care. There are now trillion dollar businesses, business opportunities waiting for people. There's a lot of other, and then we didn't even talk about fusion, which is, I'm really excited about. We are going to do multiple bets on fusion. But when you talk about AI datacenters and you're building your latest greatest new datacenter, If I

tell you, I can put a 250 megawatt facility right next your data center on 20 acres that requires no inputs or outputs. Literally you don't even need a fuel line or anything, like one tanker truck will fuel this thing for a year. You're like, yeah I'll take that. There's a lot to do that's really hard. I don't want to overstate my role, but like, part of my job is like, get behind and push. Like can we get these things more and faster and more people thinking about this? That's the way forward.

KEVIN SCOTT: Yeah, I think that is an excellent place to end here with one last question. I ask everybody who comes on the podcast what it is that they do in their spare time for fun. You've said several times now that what you do is like better than any hobby that you could possibly have. But I'm going to ask anyway. Like, what do you do when you're not doing climate investing and Senior Technical Fellowing at Meta? MIKE SCHROEPFER: My favorite three things. Number one by far is anything my kid

s will do with me. Spending time with the kids is Number one. I'm like, I'm a decent skier and a mediocre surfer. I love to go skiing and I love to go surfing. I'll never be great at them, but they are a lot of fun. Those are my three things: family, and then a couple of, like being in the outdoors doing silly things. KEVIN SCOTT: Awesome. Well, it was, it was amazing chatting with you. And as always, I'm feeling more hopeful about our future hearing your high degree of enthusiasm for all of the

climate tech stuff that you're investing in. Whatever I can ever do to help you out, like, I just want you to feel free to deputize me. I'm just glad you're doing what you're doing. Thank you very much. MIKE SCHROEPFER: Keep pushing AI. We need it. Like it's going to be a really helpful infrastructure platform for a lot of stuff here, Kevin. I think keep telling the stories of technologists. I think we have this shot at producing a much better future for everyone, and that is worth getting up e

very morning and putting the shoes on and going to work for. KEVIN SCOTT: I absolutely agree with you. All right, man. Thank you. MIKE SCHROEPFER: Nice to see you. CHRISTINA WARREN: What a great conversation with Mike Schroepfer. So as you were mentioning while you were talking with him, and even at the top of the show, Kevin, what's so striking to me is that what Mike is doing now, he was taking on challenging problems at Meta, don't get me wrong, but what he's doing now is really challenging,

but also really fascinating. And I'm really glad that someone like him, with his experience, is able to bring that lens towards the investment approach and the mentorship approach to solve these very big and very important challenges. KEVIN SCOTT: Mike and I have known each other for a really long time and I have always been impressed, not just with Mike's technical expertise, which is tremendous, but the way that he tackles problems is so great. He dives into things. He's got a learner's mindse

t. He wants to learn as much as he possibly can. Like he's just a really great first principles thinker. But he's led so many complicated things and done things with such big groups of people that he understands just the complexity of, yeah, just how to get a large number of very bright people aligned on a mission and how to go tackle really tough problems. And that's exactly what we've got here. And if anything, I think the bag of problems that he's tackling are the toughest in the world. Becau

se you have a problem that is very big and very imminent. It will not wait on any of us. It's coming whether we like it or not. And you have, even worse than we have some cases in software where you're trying to go fix something where you've got decade's worth of prior investments and you've got to solve these problems of, what can I reform versus what do I have to just tear down to the ground and rebuild from scratch in order to solve the problem the way it needs to be solved. So here we've got

centuries in some cases of things that have sort of been institutionalized and lots of investment that we've made in them that you have to understand how the new fits into that. Then you have a complicated regulatory environment that sits on top of all of that. And so it's just great to have someone like Mike trying to push really hard to empower entrepreneurs to go help us find some solutions to these tough problems. CHRISTINA WARREN: I couldn't agree more and I feel like he is really uniquely

positioned to do that because of his experience. And I think it's notable. His investment firm is called Gigascale, and that's certainly something that he's done in his career building platforms as you two were discussing. And that's such an interesting thing to me too. You've also worked a lot on building platforms and you know what it's like to make things have to scale. And I feel like with these problems that Mike is investing in now scale is really what's necessary. What's your experience?

I guess Mike alluded to this talking about needing to understand the amount of momentum you need for something. But what's, what's your experience with, I guess figuring out when you need to take something from maybe being, all right, this is the size. But if we want this to actually be impactful and actually work, we need to go. We can't just build one datacenter. We need to build 40 of them at the same time. Can you talk to me a little bit about that? KEVIN SCOTT: Yeah. For sure. It's no acci

dent that Mike and I are friends and that we've always gotten along because he and I share many of the same guiding principles for how you go tackle problems. I think the very best way that you can get change made at scale is to use systems that are already in place to assist you. So like the thing that Mike talked about, a bunch of which is just letting the markets work, like finding things that are genuinely and authentically superior and sustainable to the unsustainable things that are alread

y in the marketplace and the incumbents who are making them. I think it just does wonders every time we apply it anywhere. So it makes the incumbents better and it presents an option to the incumbent that even if the incumbent's being especially intransigent, like everybody is just going to adopt the alternative. And so like that, using that market mechanism in that competition is, I think, a really, really effective thing. He mentioned another thing which is really interesting, which is this he

at pump anecdote that he gave where there's a new heat pump technology that might be twice as expensive as the existing technology. And if you just knew that two x upfront costs meant that your long-term operating costs are going to be much less over the lifetime of the system. Like it's easy to choose the 10K upfront. The market-making thing is just making sure that the consumer understands what the tradeoff is. And maybe you also need a policy thing up front, which is, how do you subsidize or

allow the consumer to make that upfront investment? We just take for granted a bunch of things about our financial markets right now, but we don't give loans out to people just for grins like you, you loan people money because it lets them invest upfront in things that are long-term efficient. If that weren't the case, the financial industry would be loan sharking and like none of it would make any sense. So letting the market efficiently allocate capital where it can and where it's not changing

the rules of the marketplace so that efficient allocation of long-term, efficient allocation of capital can happen, is super important, I think. CHRISTINA WARREN: No, I totally agree. And I also think what you were saying and what Mike was talking about too making sure that people can understand, as you said, that tradeoff, understand how that market allocation is going to work and why it might be worth the initial outlay is great too, but no, I think those are fantastic points and I'm really g

lad that Mike is taking his expertise from building platforms and applying them towards these other problems too. KEVIN SCOTT: Yeah. Look, in addition to the climate investing he's doing, I think part of our conversation was just super interesting because Mike again has had one of the legendary Silicon Valley careers. Like not everybody gets to go from Boca Raton, Florida to being CTO, for as many years as Mike was, of like one of the iconic big platform companies in technology. And I think Mike

shared a whole bunch of super interesting stuff from how he even approached his career, that like even if you're not going to go be the CTO of a big tech company is going to serve you incredibly well as you progress through life and try to figure out how to do valuable and rewarding things. CHRISTINA WARREN: No, I totally agree. And his run - he was at Facebook for more than 15 years - but his run as CTO aligns with one of the greatest growth stories of any tech company that we've seen. And I t

hink as you said, even if you're not a CTO, just looking at that approach and looking at the decisions that he made. The tools, you mentioned PyTorch and React, that can touch so many people has been incredibly influential and I think it's incredibly inspiring. KEVIN SCOTT: And the techniques for how he manages his career, like run experiments with your career, like what can you learn? Like how can I go line my activity up with, I'm going to go learn a thing that's going to help me make better d

ecisions about my career in the future. Like thinking about leverage, thinking about markets, like making sure that you're landing in places where your effort and activity for macroeconomic reasons, is going to be aligned with the potential for high impact. CHRISTINA WARREN: And taking risks too right? Like deferring your masters to go do something for working at a startup and having that experience sometimes taking those bets, which is also what he's doing now as investor. KEVIN SCOTT: Yeah, li

ke it's a Warren Buffet quote, but it's a good one. The best investment anybody can make is in themselves. And so just really thinking about how am I getting better or like how do I improve? How am I learning more? Like how am I putting myself in situations where my growth is increasing or has the option to increase. I think those are all good questions for all of us to be asking ourselves all the time. CHRISTINA WARREN: I totally agree. I totally agree. Everybody keep learning. All right. Well,

that is all of our time we've got for today. Huge thanks to Mike Schroepfer for joining us. If you have anything that you'd like to share with us, please e-mail us anytime at behindthetech@microsoft.com, you can follow Behind the Tech on your favorite podcast platform or you can check out our full video episodes on YouTube. Thanks so much for listening. KEVIN SCOTT: See you next time.