Ep 57: Former CTO of Meta Mike Schroepfer on the Path to Powering the AI Revolution

On today’s Unsupervised Learning, Mike Schroepfer (ex-CTO of Meta and founder of Gigascale Capital) reveals why energy is a key bottleneck holding AI progress back. Mike discusses how we can scale energy production to democratize AI globally and explores AI’s role in climate change. He also reflects on a decade as Meta’s CTO and how AI coding is transforming the CTO role. Finally, he offers predictions on the future of AI developer tools, VR, and open-source models.

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

• 0:00Jacob Effron?

  Mike Schroepfer was the CTO at Facebook for nine years before founding the venture capital firm GigaScale, where he invests in companies using tech to fight climate change. I'm Jacob Effron, and today on Unsupervised Learning, we had a conversation predominantly around the intersection of AI and energy. Mike shared some really interesting thoughts on what's required to be able to produce energy at a scale to massively democratize access to AI around the world. We talked about the intersection of AI and climate change, and we also hit on Mike's reflections from his time at Facebook and where he thinks spaces like AI developer tools, VR, and open source models are going. Mike also shared what he thinks the role of a CTO will be like, given the changing world of AI and the improvement of coding models. Without further ado, here's Mike.

• 0:40Jacob Effron?

  Thanks so much for coming on the podcast. Really appreciate it.

• 0:42Speaker 01

  Good. Glad to be here.

• 0:43Jacob Effron?

  I feel like there's so many different angles that we can take on our conversation, and so a bunch of different threads will pull on. But I figured, as someone who's done and led technical teams, building a lot of really cutting-edge products in AI, and now super focused in the climate world, the intersection is a really fascinating place. And one thing that's come up on this podcast a few times is what feels like this tension right now between some of the environmental concerns and ESG commitments that companies have made, and then just the massive energy buildouts required for data centers today. And so curious how you think about that tension, the way companies are navigating this, and ultimately what's required for the US to produce enough energy for some of these AI compute buildouts.

• 1:20Speaker 01

  Yeah. I mean, I think this is contrary to what some people believe. This is the best possible news we could ever hope for in terms of mutual goals here, which is AI is an incredible technology that's going to allow us to do a lot of interesting things. And the demand that AI is putting in the near term on sort of energy from customers who have technical savvy and the money and the desire to get this done quickly, allow us to deploy a lot of new interesting solutions to power the grid. Even without AI, in the United States, we need about 5X our grid by 2050 in order to really solve our goals. I mean, think about converting every gas-powered vehicle to an EV. Think about all the manufacturing we need to do here in the United States for steel, cement, concrete, and others. That all takes a tremendous amount of energy. So we need that energy no matter what.

• 2:05Speaker 01

  This demand from hyperscalers in particular is sort of pulling forward the opportunity to deploy a lot of new technology and sort of get it down, what I care about, get it down these cost curves where the reason why people are going to deploy it in later years is because it's the better, faster, cheaper option, not because it's better for the planet.

• 2:21Jacob Effron?

  Yeah, I thought one of those interesting things in hearing your journey into kind of what you're doing now and, you know, coming at it from the private sector perspective is this idea that like, really the only way to massively scale and get these technologies out there is through, you know, through market forces, basically. And that you'd considered a lot of different ways to potentially have climate impact, that this was like ultimately the way to actually scale it. And so I think it's really interesting that, you know, essentially AI provides an accelerator on a lot of those market forces and the demand here. Have you seen anything particularly like interesting in what folks are doing around, you know, data center build outs?

• 2:50Speaker 01

  I think the first thing it's done is it's like caused people to think about this thing that is so much in the back office. It's like you think about power, it's like, where does my power come from? Nobody cares, right? And it's sort of like, how's my package get to my doorstep? Like, no, it's like, does it get there? Does it get there on time? And so we've sort of brought that conversation to the forefront. But I think what it's also done is, you know, it started to ask questions about like, what is the rate limiting step for humanity? In terms of like, how do I make forward progress? How do I get eight billion people to live in comfort and safety, right? And the answer is technology. It's the only answer I know about. You know, most people who need it don't have air conditioning. Many people don't have clean water. Those are fundamentally power problems. We know how to cool air, we know how to heat air. We know how to take water and make it potable, no matter where it came from.

• 3:35Speaker 01

  The limit on both of those things is effectively the cost of energy going into it. And so if you take energy and you attend the cost of it and deploy it everywhere, you kind of now all of a sudden open up the possibility to bring a lot of people to standard of living that we would all love and take for granted. And we think about AI and you say like, okay, great. I would love to have an AI agent, you know, running full time, a full reasoning agent, 24/7 dedicated to me solving my needs. Like, how much power does that, like, what's a reasonable amount of power to me burning? Is it, you know, a kilowatt, a megawatt, a terawatt? Like, how much energy? And if I multiply that times a billion, eight billion people, what does that look like? And you start to get to some crazy fun numbers, but it's very tractable. So I think when you break that down and say like, okay, well, what do we need to solve some of these problems?

• 4:20Speaker 01

  You know, I think there's, to me, it keeps coming back to energy, energy, energy. How do we just make a whole hell of a lot more energy and make it in a sustainable and cheap way? And I think there's basically a few ways to go. I think we have the thing that's working right now, which is solar. Eighty percent of the new energy on the grid in the United States in 2024 is solar. Utility-grade solar. Most people probably don't know this. And the reason why is it's cheap. It's just like the cheapest way to put new electrons on the grid. So that's good. Solar is awesome. It works 25% of the time. You know, nighttime doesn't work in the winter in the UK. It doesn't really work. So you have issues there and it's sort of a bummer to have a $4 billion data center filled with, you know, Jensen's beautiful chips.

• 5:02Jacob Effron?

  You want those things running 24/7.

• 5:04Speaker 01

  That's like not an acceptable answer. So you kind of want to run that thing 24/7, right? So you've got this sort of, you know, time balance mismatch. So then you say, okay, well, solar is great. Deploy as much solar as you possibly can everywhere you can to solve it. And then you start looking at other techniques to solve this. And I think, you know, one of the ones that I'm most bullish on is fusion. You know, it's a reaction that works. It's in the center of our sun. Humans have made it happen multiple times. It's incredibly power dense. You know, I have sort of rough calculations that I have to double check. But, you know, if you wanted to 5X the power grid in the United States and power all of it with fusion, one super tanker could fuel the entire United States for a year. You know, that's the sort of order of magnitude or a pickup truck would fuel a major power plant for one year.

• 5:47Speaker 01

  It's just like bonkers how much energy you get out of a teeny, teeny weeny bit of matter. And there's, there's, you know, none of the other concerns with, with waste and other things that you manage. So you kind of give me a 40-acre plot of land, you give me two years to build something, and then I've made a factory that makes energy with a teeny weeny little bit of input in terms of matter. So that, that is sort of magic. There are other really interesting ideas. There's a company out there called Panthalassa and what they have built is an offshore compute platform. And it's just a 200-meter-tall thing that bobs in the ocean and it generates energy, harnesses the power of waves, basically. But in doing so, you basically combine energy generation with cooling because you're in

• 6:28Speaker 01

  the ocean, so you have immersive cooling in a nice big heat sink right there. And you put this all together and you say, hmm, that might be the cheapest inference planet. Platform on the planet. It requires a little bit of creativity to think about how to deploy it. But they're they're out there building these things. And so there's always the opportunity for an interesting new idea to emerge that allows you to build in a way that you couldn't. And if I said, OK, well, if it's 50% or more cheaper, how much more compute capacity could you put online for inference and others?

• 6:56Jacob Effron?

  I mean, given what you said about energy likely being the limiting factor in the future, and I love people have talked about this idea of like UBI being like compute that you give to the world or like a certain amount of energy. And so it certainly resonates. Do you expect kind of more, you know, obviously, like Sam is invested in some of these energy companies, do you expect even more, you know, this intersection between the foundation model companies, hyperscalers and, you know, energy providers?

• 7:19Speaker 01

  Yeah, I mean, you're seeing it already. Most of the hyperscalers have announced purchase agreements for power for either existing or new nuclear power plants, you know. Meta put out an RFP for next generation, you know, fusion plants. There were several off takes, you know, underway in some of these things. And I think, you know, again, back to how do we stoke sort of the markets to get these things done is if you take a long term view on this and anyone is thinking about AI, it's like, look, I got to have data, I got to have computation, I got to have the right algorithms, I have to have energy. And like, that's not a conversation we had five or 10 years ago, but it is a major conversation now is how do we get enough energy at the right prices to allow me to power the training? But more and more, as we're seeing with reasoning models, you know, the compute demand is moving from training to inference time.

• 8:04Speaker 01

  And I think that that gets even more interesting because you then move into a supply demand curve for consumers, which is like at runtime, I can decide how much compute I'm going to burn for this request. And like, the cheaper I can make those requests, the more demand I'm going to get, which is like, I think there's a near infinite supply for inference time reasoning compute, you know, at the right price point. And so that's when you're not just like solving a problem for someone, you're basically enabling growth in the business by giving lots of energy. That's where companies like Pantholosso or Next Generation, Fusion Technologies or others, I think, can really just be a, the thing that makes the difference between one of these hyperscalers growing and not, you know, in 23rd.

• 8:41Jacob Effron?

  Yeah, at each price point, there's basically some massive new set of questions that would get asked or use cases.

• 8:46Speaker 01

  100%. This is the connection between these, you know, so I started in the 90s, you know, the internet came on when I was at Stanford, you know, worked in the dot com boom, worked in the mobile transition, worked in the ad transition. And the sort of the tailwind behind all of this was like the creator and cost to compute. You know, when I was at Stanford, they just stopped requiring you to take assembly programming to graduate with a CS major. You used to have to do it. I took it anyway because it was super fun. But it's like, yeah, like getting every ounce of compute power, you know, cycles out of the compute was no longer useful. Like, we could use high-level languages like C. And of course, now, it's like, who doesn't program in, you know, Python, Rust, or JavaScript, or some other high-level language? That's just like throwing away compute cycles on a daily basis for programmer productivity.

• 9:30Speaker 01

  And of course, now, we're moving to AI systems, writing a bunch of our code, eventually running our systems. Those are even less power-efficient per cycle for the thing they're doing in many cases. And so, you know, the, under this tailwind of sort of compute capacity has just allowed us to do so many things. And I think, you know, what I'm looking for in the next 10 or 20 years is, if you take that same concept and you say, what if we apply that to energy? And what if energy costs go down 10% a year, year over year, for the next, like, 20, 30 years? What happens? And I think the answer is a lot of really amazing things. From AI compute to everyone, to everyone living in air-conditioned comfort, to manufacturing stuff we didn't think we could manufacture, like fuel.

• 10:09Jacob Effron?

  You're kind of arguing it's imperative to work on energy that's really around the fact that we have this technology and it's just not going to be, like, widely distributed and widely available to people unless we figure out things on the energy side. You know, I feel like another part of this is obviously climate change and its impact. And I'm curious, I feel like a lot of people in the AI community, when, you know, faced with this idea of like, hey, you know, there's all this natural gas being put online to power these data centers, in the short term, they're like, well, you know, yes, that's happening in the short term, but like, AGI will solve climate change, and so, like, no big deal. Like, it's just, you know, three, four years. How do you kind of think about, like, that line of thinking and some of the short-term decisions being made?

• 10:40Speaker 01

  Yeah, I mean, I don't, I don't love the, and then AGI will solve the problem. That's an answer to anything. I think it's a really useful tool for a lot of things, but I think you're really kind of just punting the problem down the hill. And look, I think the answer is the next five years are going to be messy. Because if you want to put a gigawatt of power on the grid in a, you know, in a year or two, want to run that, you know, 70, 80, 90% of the time, you know, combined cycle and Gas turbine is a really effective way to do that. If you don't care about secondary effects of climate change and all the rest of it. And so, so I think a lot of people are making rational decisions, saying we're getting power now, we will buy carbon offsets, we'll do other things later. And so you're going to have some weird inversions where we're going to put some new gas assets on the grid, you know, for AI, which we're doing anyway.

• 11:23Speaker 01

  If you look at the sort of, you know, energy generation in the US, you basically see coal doing this, gas doing this. So coal's going down and being replaced by gas. And then you have to see this S curve on the bottom, which is solar. And that, that's going to catch up to us eventually and be real exciting. So, but I think what you need to do is say like, look, we need to sort of do, like, we can walk and chew gum at the same time. We need to like, plan for some short-term things and like, get some stuff online. But like, let's start putting the markers down on what we want to do over the long run. So, cool, 100 megawatts of gas this year. When I put a gigawatt on, you know, in 27, 28, what is that? Is there some combination, like, can I build some solar, which I can build very fast? Can I start trying to look at battery backup as a way to make that solar last longer?

• 12:05Speaker 01

  Can I be investing in next generation geothermal and next generation fission and fusion? And like, basically put a marker on, can I, can I pull that technology forward a little bit and/or increase the probability that it's actually going to hit? Because if it does, it's a huge enabler for everything I do. Like, all of these hyperscalers should be pumped if, you know, one of these fusion companies makes it, or if next-gen geothermal really breaks out. Because it's just like, all of a sudden, like, great, now I have a whole bunch of new suppliers I can get energy from.

• 12:31Jacob Effron?

  Totally. Especially it's the gating factor right now, or the limiting factor on the solar.

• 12:33Speaker 01

  And so that's what you see. I mean, thankfully, a lot of them are doing this, right? They're all making bets in these different areas.

• 12:37Jacob Effron?

  I'm curious, as someone who's so close to this technology, it feels like there's lots of experiments being run and different things being tried. And what kind of, actually, similar to AI in many ways, like, you know, if we fast-forward three years, we'll know a lot more about where, where things are going. What are like the two, three things that are kind of, you know, that we'll learn in the next two, three years around, you know, mass kind of energy, you know, clean energy production, that you feel like are most top of mind for you?

• 12:59Speaker 01

  You know, this is the funny thing is, I think a lot of these curves are very obvious if you stare at them closely. I mean, I started working in AI in 2013 is when we started the Facebook AI Research Lab. And because that was when the first convolutional neural net won the ImageNet challenge. And it won it by like a landslide compared To any prior advancement. And you sort of looked at that and said like, okay, well, what's going to make that be better? And you say, well, we can add more compute and we can add more data to it. You know, and use the basic same algorithm. You're like, holy crap, like we have a lot of compute and data we can add to that. Like as big as it was at the time, it was teeny compared to everything else. And so the tail ones are really, really in your favor. And I, the, the funny thing about like, when I got passionate about all, all the common stuff is like, we have a bunch of these S curves happening right in front of our faces. Like our, our face is pushed up against the glass.

• 13:43Speaker 01

  I, you know, I mentioned solar, we haven't even talked about batteries. You know, we, we talk about electrifying sort of everything and doing it because it's the cheapest form. You know, batteries, the lithium ion battery came on the market in 1991. It's not that long ago. It's 97%, 98% cheaper than it was when it was introduced. It's still decreasing in cost by well over 10% a year, year over year. And I think when you think about it again, computation, to me, the advancement is like we poured all this human energy into chip making. Like the SML machine is just like, I think one of the most brilliant pieces of, of technology ever. If you really want to geek out, go check this thing out. Right. And then we pour all this work to make a thing that like I can hold in the palm of my hand. Right. That is what makes AI work.

• 14:24Speaker 01

  You think about similar things with the lithium ion, like my car is full of things that look like a double A battery. Right. And just a lot of them. And that's because we can make a lot of them really cheaply in these giant factories. Solar wafers are not that big. I can hold one and hand it to you. Right. So anytime you find these technologies that have this capability to like pour a ton of R and D, scale up manufacturing in them and just like make the same thing over and over again. That is the engine of progress for humanity.

• 14:50Jacob Effron?

  I mean, you mentioned there that obviously when, when you were heading up, you know, a CTO over at Meta, you saw kind of the early ImageNet stuff and it was like apparent to you, hey, this is something that really could scale. I'm curious, obviously, you know, since that adventure and launching FAIR, Meta's done some incredible things. Yeah. Many of which you oversaw. You know, there's kind of this journey, maybe, you know, all the way through open sourcing the Llama models. Like as you think about that path.

• 15:12Speaker 01

  Yeah.

• 15:12Jacob Effron?

  You know, was the plan always to like, hey, we'll invest in this compute and scale up, but you know, kind of do this open source?

• 15:17Speaker 01

  Yeah. If you look at the early days of FAIR and, you know, work that was done. Produced a lot of things like PyTorch is, you know, the dominant sort of framework for for doing AI development these days. But a whole bunch of other models were released from from FAIR. Even things like FIS, which is a nearest neighbor search algorithm that is sort of like down in the weeds of lots of things. But the way I always think about technical systems is in layers. You know, I think that, you know, in the old network days, you used to talk about the multiple layers of the stack, you know. And you think about computers of, you know, I've got a chip, I've got an operating system, I've got applications that run on top of it. And the further you are down in that stack, the more commoditized you want to be. Because it's like, look, is every company in the world going to build their own foundation, like train and build their own foundation model?

• 16:02Speaker 01

  Probably not, right? Just just like most people don't make their own chips, most people don't make their operating systems like, hey, let's all use Linux, it's pretty good, we'll contribute to the parts that we want to change to it. And that's like an aggregation of, again, human ingenuity into one artifact, rather than doing lots of copies of the same thing, which is inefficient. And so our our vision for AI from the very beginning was like, look, this is a foundational technology. It's going to be input into a lot of things. You're going to use it to make funny videos, you're going to use it to help with health diagnostics, you're going to use it to sort of run the power grid, to discover new materials, all sorts of things. Each one of those applications is going to require a tremendous amount of domain-specific knowledge, right? That a whole company, a whole industry could build themselves around. But underneath that, there's something that's generalizable.

• 16:46Speaker 01

  And that's the, you know, what's been called now the foundation model. And in the beginning, it was like the the tools, like PyTorch, was that was the first thing is like, OK, well, the way we train these models, we can at least share the work there. Let's like start there. You know, CUDA is another part of this and and other things underneath it. So, you know, my thought was always like, let's get that technology. And then from a very selfish, you know, from a standpoint of what's good for Meta, you know, my thesis was always, we know, we know how to take this and put this into products in our domain. Build great social products, great ad products, great, great consumer products. And so my goal as CTO is always, how do I make sure that our company has access to the very best technology it needs to build the products we want? And the important word there is access, not not, we made it, right? I don't really care whether we made it or someone else made it, I care that we can get it.

• 17:31Speaker 01

  If it's open source and broadly used, by definition, we have access to it. And if we're helping to develop that open source thing, then not only do we have access to it, we're helping to drive it. It's like the best of all possible worlds. Get a lot of collaboration across the world, and we have sort of obvious access to it at a $0 cost, right? And that is the best possible outcome for the company.

• 17:46Jacob Effron?

  Totally. I mean, every time you're releasing these models, people are making all sorts of inference optimizations and finding ways to, you know, run them faster.

• 17:52Speaker 01

  Yeah. And it also just accelerates. I think it accelerates progress. I think that like the definition of American innovation is sort of decentralized innovation, but collaboration at the right points. And I think open source is one of those. I think scientific publishing is one of those. And like, the more you can like do that, the more people. I start by building on the work of you, rather than replicating everything myself. And then we can advance to the end quicker. And to me, that was like, let's get to the end quicker. Let's make AI more powerful, more useful, and then we'll figure out how to put it in our products and turn that into returns for the shareholders.

• 18:23Jacob Effron?

  When you mentioned PyTorch, obviously, you know, a somewhat piece of technology that you guys developed and use incredibly widely. As you think about like the landscape for developers today and, you know, what folks are using, if you were still kind of, you know, running that organization, like, what do you think are the gaps today in kind of the developer tooling, how we mess around with these models?

• 18:40Speaker 01

  Well, I mean, it's, you know, classic transition of developer tools where you start to sort of move up the stack, right? You know, we talked about assembly code to C to, you know, Python. You know, PyTorch is great, but sort of model architectures is not the place generally where people are innovating as much these days. Like, we are doing, some people are, but for most people, you know, transformers or others are there. But you start to talk about like, all of the systems around it. How do I collect my data set, train it for pre-training? How do I do post-training and RLHF and RL? So like, it's more of a system design problem than it used to be. And then how do I manage? Okay, this is great. I've got this like 25,000 node cluster, which means that at any one point in time, some of those nodes are definitely down. Like, how do I manage that?

• 19:22Speaker 01

  Restart, do checkpoints, all of this sort of systems management. So, we kind of knew this was going to happen, but it, you know, it has moved from a sort of sitting at my desk doing work. You know, I bring this analogy into physics. This is like, you know, it's hard to do physics these days without a, like a, you know, a superglider or something. It's not like every physicist has one. And under their desk, right? You have to have a big machine to do it. And like, we moved from, I used to have GPUs under my desk and like, do a bunch of work here. To like, I need a cluster. And I need a bunch of software and other stuff to manage it. So I think it's, you know, that's where it's sort of really moved is, is the entire system design from training to post-training to inference and how we manage all of that.

• 20:00Jacob Effron?

  And you kind of were referring earlier to, obviously, you know, a key part of the stack here is hardware itself. And I feel like, you know, over the last years, there've been some interesting movements from, you know, hyperscalers and building their own hardware. And I know Meta has obviously done some of this too. Like, how do you think about, or how did you think about the role that should play when you were there? And like, you know, do you, you know, in the same way, not everyone's going to train their own models, like, you know, does it make, how many people should we have training their own hardware?

• 20:21Speaker 01

  Yeah. I mean, it's a great question. I mean, it's a thing we, I often talk about with it, because I do a lot of deep tech investing. So we talk to people. And I think, understanding your supply chain and which parts of it you can outsource and which parts of it you should own is a really critical question for any company. I think it's the cool thing about technology companies is like, you have the choice, right? You don't have to buy everything, you can make it yourself if you need to. And exercising that choice at the right moments is really, really critical. You know, when I joined Meta in Facebook at the time, 2008, we were leasing all of our data. Other people build the buildings, right? And we bought, other people built the servers. And we bought the servers and put them in the buildings and put the software on top. You know, as we started to scale up, A, like, they weren't building fast enough. They actually, the data centers we were releasing were terribly inefficient. And so we built our own. And so we built a ground-up data center, and then we started building the servers in

• 21:06Speaker 01

  that data center. Importantly, we didn't build every single piece of equipment in that data center in the Gen 1. We started with, you know, the web servers, which are the most frequent at the time, the largest end, the largest number of servers in there. And then we sort of worked our way through the rest of the stack to get there. These days, it'd probably be hard to find something in a meta data center that wasn't designed by Meta, you know, in terms of the boxes themselves. You know, at the time, Intel made great chips, so we didn't need to make CPUs. We could just buy them from Intel, because it sort of generally suited our needs. So I think it's really important at all times to say like, all right, of all the stuff I buy, like, am I getting what I need out of it? And I think, you know, Nvidia makes just unbelievably great tech. Jensen is a hero, is a friend. He is, you know, a story of compound R and D investment.

• 21:50Speaker 01

  Like, the thing he does well is he just reinvests it and builds a deeper and deeper mode. But I think a lot of these companies, when the, you know, the cost of the Of the GPUs ends up being a significant line item in the CapEx. You start asking yourself questions like, can we do it better? Is there something I can do to specialize the way my chips are used? To do it cheaper, better, faster? And I think it's a question a lot of folks are asking. And I think there are different answers for different companies about whether that makes sense.

• 22:17Jacob Effron?

  I feel like maybe in the past, there would have been more time to breathe and make that decision. Whereas it feels like the pace of how fast everything is changing in this wave is, I don't know. You've lived through a bunch of these waves, but it does feel like the pace is so at breakneck speed that like you're constantly wondering, well, in six months, if I make the decision to build now and in six months, something news out, will I regret that?

• 22:35Speaker 01

  And the bummer about chips is like the best way to make gains, like it's really hard to beat Nvidia with a general purpose chip because like they have a bigger R and D budget than you do probably, you know, they have a great team. So if your goal is like, I'm going to make a better GPU, general, general purpose flops machine, like I have low, low hopes that you're going to do anything there. The only real joy is to basically specialize and say like, all right, I know this specific algorithm, I'm going to implement this in hardware. And typically if you take a software and you move it into hardware, it's about a 10x. You should expect performance per watt or price advantage over that. But you've got to guess the algorithm right. So do I, do I like literally embed a transformer architecture in there? And then in two years, when my chip actually comes out, is there some variant of that that's different or some other part of the process that's a bigger?

• 23:20Speaker 01

  So it's a real delicate balance because if you can nail it, you can do a really good job. If you miss it, your chip can be worthless because it's like literally can't run the algorithm I want it to run. And so I think you're just seeing a lot of people trying to figure out like where is the, where are the places for leverage here?

• 23:33Jacob Effron?

  And even, you know, even in the commitments folks had to make these data centers, right? Like years ahead of knowing where, I mean, obviously it seems like all science point scaling laws, but a lot can come out in the next year or two that might change.

• 23:42Speaker 01

  Sure.

• 23:42Jacob Effron?

  Those two years in the future commitments folks are making.

• 23:44Speaker 01

  I think that's an easier bet because I think that you can repurpose that capacity for lots of other things. You know, the idea that you'll need the compute for something is like not that hard. And this is why Nvidia continues to do really well because if you buy GPUs, like let's say training isn't as important, they're not as efficient for inference, but they're, they can be used for inference. So it's like, it's not like you have to throw it away. And so I think that's a relatively safe bet. And I think more people have been frustrated. You know, with wanting to do something and then realizing, I mean, this is the hardest, the hardest part about my job. I remember early on when it was like we were building our first data centers. I was like, well, how much capacity do you need? I was like, I was like, I don't know. It's like you're asking me like, well, how many people are going to be on my product in 18 months? I like, there is no way for me to know the answer to that. And they're like, well, we have to order steel. And like, you can't

• 24:28Speaker 01

  just like, order that. I can't go to Home Depot the day before to build you 150,000 square foot facility. Like, I kind of need to know now. And so it's just like, ugh, that impedance mismatch of like, I got no idea, but we're working on things that you have to decide two years in advance, which is anything in the physical world, you know, is a real, is a real gut buster for all of my friends at all the hyperscalers.

• 24:48Jacob Effron?

  Yeah. What do you learn about doing that? I mean, obviously, you had to make those calls many times.

• 24:52Speaker 01

  Don't screw it up. No, it's a, you know, it's painful to underpredict and it's painful to overpredict is, is, is, you know, but I think I've, I've regretted more underpredicting than overpredicting because it's, it's, um, it's just a bummer to see something happening and not have the capacity you need to go after and get it. And if you have extra and you need to deal with, that's kind of like a financial problem. And if you, but if you don't have the, the capacity, it's like a technical and a product problem. And I think that that's a much more frustrating thing. You know, the other thing is, I think there's lots of tricks. This is, you know, when we look at the hard tech companies, you know, if you think of this as a classic, you know, back to like chip design pipeline, you know, I've got a multi-stage pipeline where I've got to like, go find a piece of real estate, I got to buy it, I got to build it, I got to do all these things. It's like, there are

• 25:36Speaker 01

  tricks to deploy to sort of like, optimize that pipeline to, to give you some ability to either reduce latency or hide it, or give you some more flexibility at different stages, you know, on it. And so I think we've learned a lot of tricks there that I probably can't share in detail, you know, about how to do that more efficiently. But I think it's, again, like anything, if you optimize that problem, you can, you can do it really well.

• 25:54Jacob Effron?

  You kind of oversaw so many interesting things. And one of them, I just wanted to touch on, because I thought it would be fun, is, you know, you obviously had Oculus and kind of the whole VR world. And obviously, that's always been a big, you know, vision of, of kind of the meta platform. You know, I imagine there's, you know, a lot of cool things you can do now with, with what these general models can do. And I'm curious if you could just riff for a bit on like, how you'd think about like, the opportunity ahead of that now?

• 26:14Speaker 01

  Yeah. I mean, I think, you know, that The vision I'd have is like the operator from The Matrix, you know, is your AI model, which is like what you can do with, you know, VR in the future is I can be immersed in any virtual world I want. The real rate limiter for that is creation of content. It's like it takes, it's a lot of work to create really great content. And so if you could imagine a generative AI system in a 3D world that allowed me to create anything I want sort of near instantaneously, that is just a sort of a mind bending activity. So and it's sort of, you know, like The Matrix calling me, hey, operator, I want this, I want that. And it's like just appears and I'm in that world. Now, we're, we're, we're not, it's not tomorrow, this is not next year, but like that is definitely within the grasp of us in the next, you know, next X years, I don't know what X is. And so that, that is obviously super exciting.

• 26:59Speaker 01

  And then I think, you know, the other thing that's really exciting is the idea of a contextual AI walking with me everywhere I go. Right? So if I have a pair of smart glasses, um, you know, that has the ability to sort of help me, whether that be anything from live translation. So I'm traveling somewhere and I want to speak to someone in their native language. I want to interpret a menu, um, wants to help give me context, you know, it's effectively a tour guide at the same time of, you know, what I'm looking at, um, or connecting me with friends and family to bring them into that experience or share memories or remember them from that. Like that, that is going to be pretty transformational. And I think we haven't had, you know, there's so many interesting things happening in AI, but there's still like a thing I do on the side. It's like this thing I go text or I go to my computer and like the idea that it's sort

• 27:44Speaker 01

  of a thing with me all the time is something we've really yet to experience. All the people have tried this in a bunch of ways. And I think that's something you're going to see in the coming years and that'll be

• 27:53Jacob Effron?

  It feels like folks are still trying to figure out the form factor, but obviously, you know, uh, your former employer with the glasses and lots of people, you know, messing around with different kind of hardware you can go around with. I guess I asked it similarly on the climate side, but on the AI side, I mean, everyone's asking like, you know, where's, where's this world going in the next two, three years? What are kind of the key questions on your mind right now, uh, that, that you're kind of most paying attention to that we'll, we'll know in the next year or two?

• 28:11Speaker 01

  Yeah, I mean, uh, the obvious thing that's been happening recently is sort of the emergent, you know, so for a long time, the, the long time. A year or two ago, the big question was,

• 28:20Jacob Effron?

  Dog ears in AI world, I would say.

• 28:21Speaker 01

  You know, how, you know, the debate of like scaling versus algorithms, like how far can you go by scaling an LLM? And, you know, I think the answer is we're hitting diminishing returns. And just like, just a straight up pre-training, lots of data, you know, add a straight up language model. And so then the question is like, what's next? And I think that surprising thing has been, you know, treating the LLM as sort of an input into a reasoning model via post-training RLs or things like that, that allow you to use the LLM, but sort of get better results on the outside. So I think that that's a really curious, like, how much legs does that have? So is that a crank we can turn for two years, five years, 10 years? Or does it take?

• 29:01Jacob Effron?

  How many domains are easy to verify?

• 29:02Speaker 01

  Exactly. And then, or are we just like missing some other component? Obviously, a big one is memory. You know, a million token context window from Gemini 2 is pretty amazing. But, you know, you'd really want some sort of associative long-term memory, you know, and that's kind of what humans have. And we're kind of missing that, and you know, still in LLMs. And then I think as soon as you move into domains where they're hard to verify, I think you struggle, you know, a lot. So math is awesome because you can verify where they got the answer right. Coding is awesome because you at least like does it compile? And there's lots of data there. You know, as soon as you move into video and other things like that, it's just like, it's just harder to know, you know, where to ground the model. And so that's, that's where I think we'll, we'll struggle a little bit.

• 29:42Jacob Effron?

  I guess, you know, switching gears to obviously your climate focus today. You know, what applications of AI to climate have you seen that, you know, you think are most interesting?

• 29:50Speaker 01

  I think there's a bunch. I mean, if you think about problems that are, the way I like to think about it is a problem where it's like a really critical problem that AI can solve. And one, one is actually exploration. And this can come in a variety of forms, but you know, geothermal exploration, there's a company called Xanscar that basically said, look, you know, there's places where if you drill in just the right spot, like hot water pops out or steam pops out of the ground. It's like magic source of energy. And so the question is like, well, can we find more of those places? You know, and instead of just like drilling in a bunch of places, can we use data to analyze it? So they're, they're trying to use AI to basically find, you know, the best spots for, for, for geothermal. I think there's a lot of other examples of this. If I'm, if I'm looking for. For copper deposits, if I'm looking for hydrogen, like I'm looking for something and rather than like digging stuff up and trying to find it, can I use data to find it?

• 30:35Speaker 01

  That's a, it's a huge, really interesting area that I think we're going to see a lot more in. A little bit more pedestrian, but useful is, you know, weather, weather prediction and other things like that. Insurance, you know, other things, can I predict risk better for, for different things? And then I'd say probably more exciting is, you know, we need either invention of new materials or optimization of new materials. And so whether this is for carbon capture, whether this is for, you know, catalysts for different kinds of reactions. And again, you're in this like, I'm searching a multi-dimensional space for a thing that has a certain set of properties. Can I find that more quickly or produce a bunch of examples that I can then go iterate and test? So materials discovery and other things like that.

• 31:16Speaker 01

  And even outside of climate, they, you know, Reg just started a company on, on sort of, you know, cancer, cancer cure discovery via AI. And I think I'm optimistic on that, see in terms of finding pharmaceutical targets for, for other things. It's like this idea of like, I've got this massive space of variables, can I down select it to 20 or 30 or a thousand that may actually be, and then go test them all? Like that, that is a problem space that I, I think AI is already helping a lot with.

• 31:41Jacob Effron?

  Well, I mean, this AI material sciences work is so interesting. I think what would be fascinating to see is like the milestones that end up mattering to like the end companies that would bring this stuff to market. Like I feel like in the bio world, you know, traditionally the, the problem has been, there's been all these amazing, you know, target identifications or things that look interesting. And then turns out you still have to run, you know, five plus years of trials. And I feel like I'm actually in the material sciences space, it feels like the equivalent there is people are like, oh, that's an interesting material. Now can you manufacture it at scale on a way that's like interesting?

• 32:07Speaker 01

  A hundred percent.

• 32:07Jacob Effron?

  That feels like it might end up limiting, you know, even if you can find really cool materials, uh, how many, as a company, how many can you actually bring to? Ideally, you guys have a sourcing partner for that, but like, how many can you actually bring to mass manufacturing to show? So if you're in the bio world, there's been all these platform companies that ultimately only are able to bring one or two assets to market.

• 32:24Speaker 01

  Yeah.

• 32:24Jacob Effron?

  I'm kind of curious to that, you, but you must see these companies all the time.

• 32:27Speaker 01

  Yeah. And actually like this is, I am bullish on AI in the long run, and I think it's way overhyped for these applications in the short run for this exact reason. And the way I often describe is like, give me a pie chart of all of the activities involved in bringing this new material to market in terms of time and cost. And like, how much of that pie tart does your AI solution actually address? And in many cases, it's like sub 10%. It's like, oh yeah, we have this discovery part here, but then I got to figure out, actually the hard part is scaling manufacturing. And like this does nothing for that. And like, so I have to do all of that work and then I have to find customers and get out there. And so, you know, getting a couple of new discoveries doesn't actually help you. So I think that the solutions that are interesting are the ones that are sort of more direct end to end. That's why, for example, this geothermal discovery one is interesting is because like exploration of new assets is an actual activity that people care about.

• 33:12Speaker 01

  It's like, I make that much faster. And it's like, oh, okay. Well, that's, that's really interesting. That's really impactful to my bottom line. So finding those connections where you can sort of directly do it. In a, in a more consumer-y lens, um, you know, one of the companies we, we backed is actually a tech founder, Selena Tabacuala, started Evite, worked at Survey Monkey. She's a product engineering background and she's attacking sort of home efficiency and comfort for consumers. It's like, hey, my house is kind of drafty and my power bill's too big. What do I do? It's like, well, you can have like a person show up in a truck and like drive around and like look at your house, or we can have AI do this. Like you hook a little thermal camera to your phone, you just take a bunch of videos around the house, you send it in, their AI agent processes and sends you back reports. It's like, okay, here's the top five things you can do.

• 33:54Speaker 01

  Like add insulation here, replace your fridge, replace your water heater, payback time is seven months. Right? That's pretty impactful, right? And, and it's something that, again, most consumers won't wait to have the person show up and pay the 500 bucks for it, but for 50, 100 bucks, they would do something that saves them money, you know, with a payback period of six to 12 months. So that is a sort of direct application of, you wouldn't think of that as AI, but it really is AI because like that's how we're figuring this out without having a human have to do all the work.

• 34:21Jacob Effron?

  A hundred percent. I guess in the future that will be integrated into our personal agents going around with us and just automatically.

• 34:25Speaker 01

  Yeah, as I walk through my house, it's like, by the way, I gotta insulate your front door. It's like, all right, I'll get on that.

• 34:29Jacob Effron?

  On the personal side, I'm curious, like, how do you use AI in your day to day life today?

• 34:34Speaker 01

  Well, um, it's, um, it's an interesting question because I, I tried on a regular basis for some of the investing work we do. Um, and I think that there's, I mean, this morning, there was a great Hacker News post about, uh, Undergrad who basically found a new sorting algorithm, sorry, a new hashing algorithm to improve the performance of super nerdy stuff, hash tables. And so this is a great, like, one of my favorite uses of LLMs is like, take this complicated paper and can you explain it to me correctly? And like, it did a pretty good job of summarizing, but like, when I asked it to produce examples, they were kind of incorrect. And so I found that interesting. But so I think distilling down large amounts of information, I found

• 35:16Speaker 01

  deep research actually surprisingly good. Like, in my world, which is like, a lot of what we do is thesis investing. It's like, okay, geothermal, like, what are the right approaches to geothermal? Or what do we think about biochar? And so I've asked deep research to basically produce reports on all of these things. And like, I would say it's like, it's equivalent to like, a pretty smart, non-domain-specific human. If I like, take someone, a business school student who knows nothing about the field and says, spend a week, research and then summarize all the different reactor designs for me, and like, give me a guess. Like, it's about how good, it was like, in 15 minutes. It's pretty damn good. So I think I generally use it as a like, accelerant for summarization and other things with a lens that, you know, it's, I want to double-check all the data because it's still unfortunately not always accurate.

• 35:58Jacob Effron?

  Yeah, it's a question I always like to ask guests on the podcast. Like, you have kids. I'm curious, like, how you think about how AI will make their lives different, and if that's impacted the way, you know, you parent or think about the skills you want to impart at all?

• 36:09Speaker 01

  Yeah, it's funny. I mean, I think that the more technology advances, the more I think the fundamentals are important. And what I mean by fundamentals is just like, humans are still the best at looking at lots of data. Like, the more obvious the answer is, the more, the more a machine or an AI is going to figure it out. And the more that, like, you're making a gut call on imperfect data, the more humans are sort of interesting in that. And so sort of, literally, critical thinking skills, basic math, writing, reading, like, all of the skills involved in sort of consuming and understanding, thinking sort of at a higher level. You know, both of our kids code. I don't think that, you know, I don't know if they're going to be programmers full-time, but like, understanding how things work. And the thing, you know,

• 36:53Speaker 01

  The two skills that I think have served me the best in my career, when I sort of look back on like, what were the form of things is like, I was on the debate team in high school. Had a lot of fun doing it.

• 37:02Jacob Effron?

  A real Renaissance person.

• 37:03Speaker 01

  It was in the nationals for debate. But it just like, from the very beginning, like getting up in front of a crowd and talking is like just not a big deal to me. It doesn't mean I'm great at it, I'm just like not nervous about it. But it also requires you to think about like, I gotta convince you about, I'm trying to hear, trying to convince you and your audience so like, we're in the beginnings of the next big revolution, which is energy. And then I think the second is what computer science and engineering at Stanford really taught me is like, like we learned, I think we learned in C and then I learned C++. But I learned really quickly, like the language wasn't important. What it was is like the foundations, like how does a computer work and what are different ways to process data? And then how do I break problems down and understand them?

• 37:45Speaker 01

  And then how do I have enough of an ego that like, look, if I think really hard about a problem, I can learn anything. Right? And so it's like here, you know, I'm talking about fusion, I'm not a plasma physicist, right? You know, I know a reasonable amount about it, but like, I can learn as much as I want to invest my time and energy in. And if I could wish for anything, my kids, it's like that skill. It's just like, just apply your mind to it, you can learn it. Don't do any of the, I'm not an ex-person nonsense. When I hear people say that, it drives me, I'm not a techie, I'm not a scientist, whatever. It's like, everyone I think has the capacity to learn something if you really apply your mind to it. And just like start peeling back the layers and like understanding it one level deeper each time. And that is a skill that I wish more people have.

• 38:22Jacob Effron?

  I hope with some of these AI tools, like deep research becomes, you know.

• 38:25Speaker 01

  It becomes a useful, you know, it's like a really fast tutor. It's like, hey, I don't understand that. Like what the hell does that mean? You know, explain that to me. And like, that's the, that's an amazing sort of sidekick to have. It's just like, okay, I'm not going to judge you for like not understanding that. Here's the answer.

• 38:38Jacob Effron?

  I always like when I'm like explaining it in a simpler way and then it gets like real simple.

• 38:41Speaker 01

  Yeah, explain to me like I'm a sixth grader. Like, it's like, okay, now that makes sense. Well, why didn't you just start there?

• 38:47Jacob Effron?

  Totally. I mean, I guess you mentioned, you obviously, this kind of, I think coding is one of the areas where these models have made the most progress. Like you ran these massive tech organizations. As you think about like the CTOs of the future in five years, running these really large teams, like, what stays the same and what is like going to be just so fundamentally different about those orgs and like that role?

• 39:05Speaker 01

  Yeah. I mean.

• 39:08Speaker 01

  I think it'll be more similar than people think it is. And if you think about, you know, think about the quality of the tools, the size of the code bases, you know, in the 90s, when it was like, can we compile the whole code base? It was like, it was a challenge that was like common in the 90s, like, just like, nobody talks about it anymore. And, you know, the teams have gotten massive, the code bases have gotten sort of unbelievably large. And so we just got this like, you know, a joke, it's like, when you're kind of digging with shovels, and then like a backhoe showed up, it's like, oh, that's a lot faster, let's like, do that. And so I think it's just an extension of that. It's like, you know, AI to me is, you know, AI is to coding the way like JavaScript is to C, is to assembly language.

• 39:52Speaker 01

  It's like, okay, cool, like, I'm expressing my thoughts at a higher and higher level of abstraction, and like, AI is the ultimate, it's like, write me a piece of code that sorts the following array in the following ways. Like, that's like, obviously faster than writing that, like, calling a sort function built in the standard library, which is obviously faster than writing the damn thing from scratch in assembly language. So, so, but I think at the end of the day, like, getting back to our other point is, is like, the actual important thing is like, what problems are we trying to solve? What's important to go after? And how do we organize this group of smart humans to go after that problem? That is the sort of like, universal skill of like, cool, I've got all this cool technology, but like, do we go really deep on this like sorting algorithm? Or are we solving this problem for our customers? Like, what, what's the important problem to solve? And this is, when I work with people, one of the things I look for is just like, there

• 40:37Speaker 01

  are people who can run this like priority queue in their head, and they're consistently resorting, and they've like, always got the most important problem at the top. And there's other people that are brilliant, awesome, can do awesome things, but they just like, get distracted, and like, they're off like crushing a 1% problem, right? And just like, ah, it doesn't matter, come back to the 99%. Like, doesn't matter if your thing's 10, 10 times as good. And so, you know, just getting that skill in an individual and in an organization, it's like, are we focused on the, the highest level, which most important thing? And like, the more you can do that, then you can use all these power tools to like, build an amazing product and an amazing company.

• 41:09Jacob Effron?

  Do you think the size of the organization required to do that changes? Like, obviously, you know, people talk about, oh, they'll be the first 10-person, you know, billion-dollar company or something.

• 41:17Speaker 01

  I mean, I think you're seeing companies come up faster with smaller teams do it. So yes, I think team sizes will go down over time.

• 41:22Jacob Effron?

  Yeah, we always like to end our interviews with a quickfire round, where I got your take on some overly broad questions that we cram into the end.

• 41:30Speaker 01

  An overly broad question and a quickfire comment.

• 41:33Jacob Effron?

  So maybe to start, you know, do you think model progress this year will be more or lesser the same than last year?

• 41:40Speaker 01

  I think it'll be more.

• 41:41Jacob Effron?

  What's the go-to thing you try whenever a new model comes out?

• 41:44Speaker 01

  Summarize a very complicated research paper for me.

• 41:47Jacob Effron?

  What's your weirdest prediction on the implications for all this AI progress for the future and how it'll be different?

• 41:53Speaker 01

  I think most people will have an AI friend.

• 41:57Jacob Effron?

  Yeah. Do those AI friends also intersect with like, you know, I feel like people have tried this, you know, like a group chat with some AI people and some real people or, you know, how separate do those worlds end up being?

• 42:07Speaker 01

  I think it's probably like, maybe you'll call that friend in a group, but it's like, if you've ever had the experience, you have two different friend groups who have two different hobbies, you can bring them together and it's sort of awkward, you know? So I think that like, I think the actual value is in a non-judgmental on your team, like, what do you want in a friend? It's like, I want you to be on my side, like, I want you, you know, and have my interest at heart. And I think more people will find that with AI chatbots than we realize.

• 42:31Jacob Effron?

  I guess reflecting, you know, back on your time, you know, standing up there and obviously ran it super successfully at Meta, you know, hindsight to 2020, we've learned a lot about AI, but anything you look back on and kind of regret or you're like, oh, I wish we'd done X or Y thing?

• 42:45Speaker 01

  I mean, there's a lot to be excited and proud. I mean, sort of, you know, PyTorch, you know, Llama, you know, I think even in the last couple of years, people are coming around to, you know, when Meta really went all in on open weights for Llama, that was not a particularly common decision. And I think people are coming around to sort of the power of that. So I think that that's one of those things that we've had a long-held belief that I think, I think was right, we're on the right side of history there. You know, I do think that, you know, this debate of sort of scale up versus algorithms, I think there was a period of time where we were having that same debate internally, and I think we probably could have jumped on the scale-up side a little faster.

• 43:21Jacob Effron?

  Well, I always like to leave last word to you, so, you know, where can folks go to learn more about you and the work you're doing today, or really, anywhere you want to point, folks, the mic is yours.

• 43:29Speaker 01

  Okay. I didn't provide my list of sponsors. I like to thank Powerade. Oh, no. So in all seriousness, I mean, look, you can find me at gigascalecapital.com. You can find me on LinkedIn. I'm on X occasionally and threads often. So you'll find me in any of those places. Awesome. Well, thanks so much. This has been an awesome conversation. Awesome to see you.

• 43:51Jacob Effron?

  Hey, guys. This is Jacob. Just one more thing before you take off. If you enjoyed that conversation, please consider leaving a five star rating on the show. Doing so helps the podcast reach more listeners and helps us bring on the best guests. This has been an episode of Unsupervised Learning, an AI podcast by Redpoint Ventures, where we probe the sharpest minds in AI about what's real today, what's going to be real in the future, and what it means for businesses in the world. With the fast moving pace of AI, we aim to help you deconstruct and understand the most important breakthroughs and see a clearer picture of reality. Thank you for listening and see you next episode.