a16z Podcast - How Radiant and Heron Are Rethinking Power Generation and Delivery

Episode Date: March 31, 2026

a16z general partners Erin Price-Wright and Erik Torenberg speak with Doug Bernauer, founder and CEO of Radiant, and Drew Baglino, founder and CEO of Heron, about rebuilding American energy infrastructure. They discuss portable micro nuclear reactors, solid state power electronics, why delivery rather than generation is the real bottleneck, the case for modular manufacturing, and whether data centers are actually good for the grid.   Resources: Follow Doug Bernauer on X: https://twitter.com/DougBernauer Follow Drew Baglino on X: https://twitter.com/baglino Follow Erin Price-Wright on X: https://twitter.com/espricewright Follow Erik Torenberg on X: https://twitter.com/eriktorenberg Stay Updated:Find a16z on YouTube: YouTubeFind a16z on XFind a16z on LinkedInListen to the a16z Show on SpotifyListen to the a16z Show on Apple PodcastsFollow our host: https://twitter.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

The grid is breaking. We're so bottlenecked today on the lines that run crisscross across the country. I mean, it's this very complicated, giant organic machine. New power is not the problem. Delivery is the problem. The energy services were growing over time in the United States. The net electricity delivered to accomplish those energy services stayed basically flat. We can take that momentum and bring it into a new problem statement, which is power for data centers, power for industrialization, power for economic growth and prosperity, and for sustainable energy. The idea that the grid can grow and move from the edge is just not something that we've really been able to process for the last 50 years in the U.S.

The grid itself is civilization, right? Electric power is civilization. You can metamorphosize the entire grip. Civilization can regrow off of a new architecture of moving power, right, and use all of the free energy that's out there. The sunlight is free. You put up a panel, you're getting it. It's very cool, but also your aim is free. It's in the grounds.

It's there. If you take it and we use it before it just go. goes away. But it makes this like completely new way, I think, of thinking about nuclear power. It's just, it's in the options list and it wasn't even before. Electric power is civilization. Every socket, every server assumes a grid that works. When Edison wired Lower Manhattan in 1882, he connected 85 customers across one square mile.

The model that followed centralized generation, one-way transmission, held for more than a century. Now U.S. electricity demand is rising for the first time in decades. Data centers, electrified transport, and reshoring are outpacing the efficiency games that mask years of grid underinvestment. New generation is not the bottleneck. Delivery is. This episode examines two responses, portable nuclear reactors built in a factory and solid state power electronics designed to rebuild the grid from the edge. I speak with Doug Burnauer, founder and CEO of Radiant, and Drew Baglino, founder and CEO of Heron,

alongside A16Z general partner, Aaron Price Wright. Hey, everybody, we're here to talk about energy and how your companies are playing a role in the sector. But first, let's start with, how do you guys know each other? Yeah, so I'm Doug Burnauer, Drew Begglino. And so we know each other. We were both working for Elon about 10 years ago, but at two different companies.

So I was at SpaceX, and Elon was really excited. He wanted to build Hyperloop. He wanted to put little cars in a vacuum tube and go super fast. And Drew was a VP of R&D at Tesla at the time. And I called him up and was like, Elon says we need battery packs. We need Model S motor. We need to operate these things in a never before operated condition.

And Drew was like, do we really? And I was like, yeah, we kind of do. But it was kind of like that. And then from there we did Boring Company also. There was an idea to do like 30 tons of batteries or something on a trailer to power an entire tunnel boring machine. like one and a half megawatts. So it was a story of power, actually.

Yeah, there was a lot of back and forth on what was possible and could we reuse this or that. And also a big part of the Hyperloop story because we published a big white paper was actually assessing all the technology is required to deliver the concept, right? Including for my team, it meant looking at like slingshot linear motors that would accelerate the capsule in the vacuum.

It was all kinds of fun, never explored physics for our team before. I don't know if you were had someone like. Yeah, I wasn't roped in that really early stuff. It was more like when it was for real, we're going to do it, and you had to build it really fast. That's when I got put into it. And just take the voters from... Just eight months from now,

have people from 23 different countries or whatever it was come and deal the event. It was pretty wild. And as college students, they really ran with it. I mean, it still happens, right, every year? I think they did it three or maybe it was four times, and then they stopped it, which is good. It's for the best. It was exciting.

It led to some companies. There are some startups in, I think it was one really good one based on the Netherlands and a couple of others also. It's also fun to see which universities did the best. It totally. Mostly the Europeans, to be honest. Yeah. They crushed it.

We'll cut that out there. No, but I mean, you should know your competition. And the good thing is, like, MIT was doing amazing. Delft did it really awesome as well, though. And I mean, we should learn as much as we can from that. Delft isn't a, yeah. Incredible engineers going to Delft. Speeding you have things that haven't been achieved before,

let's use that to segue into your respective companies. Talk about the moment, the insight, the why now that led you to start your respective companies. Doug, let's start with you for every year. Oh, man. Yeah, it's a fun story. So I was at SpaceX for 12 years. I joined in 2007. So I joined when they had two failed rockets, no successful rockets.

And so I got to work on the first ones that worked. And you're like, this is the company to be at. Yeah. I just wanted to work on an important mission. And I didn't, I really just cared. Can I like polish one stone of this, like, great big pyramid that is like some, lifetime achievement for someone else even, right? That's what I wanted.

So, yeah, joined. I did that. I did the first two Falcon 9s, did the ground system for it entirely, which involved all the permitting also. So this is like launching a rocket from a military base. There's a lot of like regulatory stuff there. Your first four A into the permitting wilderness.

Totally. And not to eat up all the time. I worked on like the first rocket with legs called Grasshopper back in 2011. It was a four person team really designing, building the whole team. the whole thing. And we were reporting directly to Elon, like just Elon to us four and then building the whole thing. And it was awesome because we did really well. We got lucky a lot of times, but we may rocket that flew and landed on legs. And then I did all the weird Elon side projects

and ideas. So Hyperloop, when he got really serious, I got tapped into that and into the boring company and then Mars colony design. And in doing the Mars colony design, I was looking at how do you take Starship there, make fuel from what's on Mars, make fuel from ice that's there. And if you do that, you need megawatts of power. And I was trying to do that. I was trying to do that, I was trying to do it with solar and getting totally stuck and showing Elon these plans that were like four miracles we need on a single mission. And it was just ridiculous. And so Elon was like, you probably should look at nuclear. And that's really the jumping off point. Right. I started to learn. And then three years later, I left, I founded Radiant and left to go run it. And really

trying to make mass-producible, portable microreactors, not for space, but, you know, currently we're focused on a trailer-sized thing. But also needed in space. Right. Yeah. So, I do eventually want to do products for space, but we got to have customers. We got to have funds that are actually there. Yeah. So a similar story, I guess, and that for me it goes back to similar time.

So I joined Tesla in 2006. And the reason why I even went there in the first place was because I had done this undergraduate thesis in how to enable New Zealand to meet their Kyoto commitments, if you remember the Kyoto Protocol. And they're an island nation. So it's almost like a micro-good study in disguise. So what are all the resources there?

How can you reduce the carbon emissions from all these different sources? I was focused on transportation in particular, and I became convinced that electrification of transport was not just a way to solve a carbon problem, but actually just the best thing to do from an economics perspective. And that motivated me to come to Tesla. And then over the almost two decades that I was there, all the technologies progressed over those two decades, right? Like the power electronics became cheaper, the batteries became cheaper, what you can apply these problem statements to just grew up. in scope, right? We proved that electric vehicles could not just be the best vehicles, but also affordable vehicles that renewables can come with really affordable storage to help decarbonize the electricity sector. And towards the end of my time at Tesla, I had the opportunity to

work on this project called the Master Plan Part 3 project where we were effectively studying, okay, now let's do it for the whole globe. Can we have a sustainable all-electric or largely all-electric sustainable energy future? Is it feasible? Are the resources there? Because the investment, reasonable. And the answer was like yes, resounding yes in many ways. And so I got further commitment that not only was electrification coming to transport, but to like everything. And so the electricity grid needs to grow immensely, you know, three to five X, depending on how you calculate and how much intelligence we need to electricity. We need a lot. We need a lot, a lot, for sure. Everything can't just be like little intelligent generators and microgrids that like mesh together

whatever and however they're needed. All of the above. We don't have that already? Yeah, right. Yeah. Which planet is this? Yeah, like maybe humans or that, I'm not sure. Yeah. Very energy efficient. So, yeah, I got convinced that we needed a massive scale up of electricity generation, you know, production, distribution, transmission and stuff. And I had seen at the same time that while there's so much innovation happening on one side of the wire, there's really been almost no innovation on the grid side of the wire. And that's where I got the conviction to go after power electronics for accelerating electrification, really increasing the scalability and affordability of doing everything that's required.

to enable electric generation in use. And that's how I found it her in, yeah. What feels structurally different about the increase or the rise of energy demand across AI, defense manufacturing, whereas previous areas had more flat growth in demand. Feels structurally different. Yeah, high level, we benefited from energy efficiency,

from the 70s, really, were really the 80s until like the 2010s. And you saw efficiency everywhere, like an industry with variable speed drives, with lighting, with heating and cooling, progressively more efficient refrigerators and air conditioners. And all of that energy efficiency more or less

negated the growth in energy services. Because it's absolutely like the energy services were growing over time in the United States. But like the net electricity delivered to accomplish those energy services, stayed basically flat. And the same thing even happened in data centers. What's actually really interesting.

If you look at the efficiency of compute, even with all the AI things that are going on in terms of like flops per watt, like it's really gone up an immense amount from like the first data centers to today. And that is allowed, you know, more and more compute capability for the same percentage of GDP,

electricity production. Now, it's changing, it's changing now that we can find more applications, I think, for AI than ever before. And so we're gonna see, we are seeing, you know, electricity growth outpacing, well, higher than it's ever been before. And that's not just because, the benefits of energy efficiency are tailing off, but it's because we're seeing like broad-based

demand for more electricity, compute, transportation, industrial electrification. It's an exciting time, I think, because electricity as a carrier is like incredibly flexible and as a pathway to a sustainable energy future, whether it's for intelligence or for heating, I see it as a great thing and want to accelerate it. Yeah, I think though, you know, on the flip side of this, the last 40-ish years of efficiency gains we've gotten, which have been super important. It's allowed us to kind of look the other way about the real bottleneck in energy generation on the grid, which is just, which is delivery. And we're so bottlenecked today on the lines that run crisscross across the country.

I mean, it's this very complicated giant organic machine. We probably don't have to talk long about that. Everyone's talked about that plenty. But the grid is breaking. You know, so when we think about just, you know, from an investment perspective, the types of things that we're interested in, it's like how do we actually go about kind of solving that really complicated transmission and delivery challenge with technology?

Yeah, decades of not a lot of change means like the brightest and most talented people are like, I got to go do something else more impactful with my life. And a lot of those folks actually, like, for example, a lot of the talent in GE went to Korea, right? Or Japan, because those were the really. growth markets. And so you got this almost like hollowing out, or China, right, hollowing out of the knowledge base. And so yeah, we need to reinvest because we're growing again. It's exciting. I mean, it's opportunities for both of us and others to come to make it happen. Yeah. When you flesh out our energy investing thesis a little bit more in terms of what are the types of opportunities we've been involved with or interested in the why now, you know, is it technical breakthrough? Is it regulatory? Is it some combination? When you flesh out a little bit more? Yeah, I mean, you know, this sort of AI moment in time is, is, it's a good, it's a really good forcing function. I'm actually, I'm very, I'm very grateful for AI for lots of reasons. But it's just, it's bringing to a head a lot of things that have sort of been brewing under the surface for a long time.

And it's kind of forcing us to remember how it builds large scale infrastructure in the U.S. again and reminding us that actually energy, energy is really important. Electricity is really important, getting it to the place. where it needs to be is really important. So whether it's data center compute or the broad-based reindustrialization of sort of the industrial base of the U.S., whether it's defense use cases on the edge or more centrally,

all of these technologies that we're investing in at A16.Z are kind of demanding access to power. And it's not strictly access to power on the grid. I think what we're seeing is the importance of having lots and lots and lots of different types of power generation and a lot more resiliency in the overall system and network. So I think, you know, we all take for granted that when we turn our lights on in the office, you know, when you boot up your computer, there's sort of like a unidirectional line of

electrons that go from some sort of massive central power generation station to, you know, your power outlet. And what we're seeing and what seems to be necessary with much more spiky, much more complicated loads on the grid and off the grid is that resiliency, decentralization, and software-driven workflows really matter. And it's going to be impossible to build out the grid for sort of the max power capacity that we could need, even at the most remote edge use case, like on an Army base or something like that. So when we think about our energy thesis, it's how do we turn this network into a much more software-defined, much more resilient, much more decentralized, you know, much more decentralized things?

And so, you know, I think both of your, both of what your building sort of feeds into that sort of micrograde behind the meter, on grid, off grid, hybrid. We just need to be a lot more flexible and a lot more adaptable to a variety of changing conditions. Doug, it seems like the tide turned on nuclear a few years ago in that, you know, more and more people start to realize the, you know, the borneous criticality of it. What is sort of the progress that we made as an industry? What would have we achieved and, you know, what are the biggest bottlenecks remaining in terms of, you know, really make progress as a country? Yeah. It's a good question. I think there's a bunch of fun ways to answer it.

I mean, the one thing I like to say, it sounds a little sensationalist is that there is no nuclear industry. But it's really true. You know, we're kind of, it's almost like we're getting excited about flight before Kitty Hawk, right? To a certain degree. there's a really coming very soon deadline. A lot of companies, a lot of little nuclear startups have actually been given access to fuel and facilities and just expedited support from the subject matter experts

required to regulate to make sure that these are going to be safe tests. And so by July 4th, several companies will have reactors built that go critical that are fundamentally new designs, completely new and from scratch, but it hasn't happened yet. So it just feels like a little bit of cart before the horse. Are you, does that worry you at all? Not too much. You know, I've been doing nuclear, well, thinking about it since 2016, but I founded Radiant in 2019, and then for a year just learned how to do reactor design and then raised money in 2020.

And just, I never founded a company before, never intended to really do that. And I kind of slow rolled into it. I could have tried to go much faster. But I've stayed totally committed to just building. And actually, the funny thing is like in 2020, I said in 2026, I will put a full. scale reactor and get it critical and get it up to full power. And we're on schedule to do that, which is kind of wild. And it's not like that was really the actual plan, but it was just, I was resilient to

all the challenges that were put in the way. We are now the only reactor permitted of these new reactors to go to full power. So a lot of others are getting to critical, which doesn't mean you get to high temperatures or high power. And those things are very challenging on all the parts in the system. And they require careful consideration of the thermal gradients and the alloys, You need high strength materials to do that. So that's really exciting, but we're like not quite there yet.

And I think if we're doing the same discussion next year, it's going to be dramatically different because we're going to be able to point at all these different designs, what you could do with them. And I think the products, like nuclear reactors as products, has never been seen before. There are always usually these giant mega projects where you dig a huge hole in the ground and you take five to 10 years or up to 15 for the slower, the bad projects out there. But reactors that can just come, ours, you know, we're targeting one per week coming off of a production line

from our Tennessee facility, which is an 80-acre site we just signed for in October, not even a year ago. But I want to tie back into the grid because I was just, I had some interesting thoughts. And we really, our product is for off the grid. Right? It's a megawatt reactor on a trailer. And you can, we build in our factory,

we drive it or fly it to where the customer wants it to go. and then turn it on within like 48 hours. We go, you know, wheels stop moving, and then we go to power on your site in that amount of time. And then it lasts five years, which is like a full oil tanker worth of diesel equivalent. It's two million gallon diesel equivalent.

So it's sort of an unbelievable thing where you can grow the grid or put a microgrid anywhere. But it's like a totally different problem, I think, from the grid itself is civilization, right? Electric power is civilization. If you go in those sockets and you pop something into them and you just get power, that's very well developed. That's civilization and that's using electricity to do what you could otherwise only do with human muscle or animal muscle.

And then I'm excited about Drew's product and company what it can do for the grid, but also for microgrids. And it'd be fun to hear your thoughts on how these things mesh together actually. Because we had talked about where products actually go. together. Yeah. Yeah. So maybe, and talk about your product. I know you have like a power building block of a certain size.

Yeah, for sure. We're building. Our first product is Heron Link, and it's a, it's a 5 megawatt bi-directional solid-state transformer that goes from DC anywhere from 800 to 500-500 volts DC to 34,000 volts AC, which is effectively, that AC voltages is the sub-transmission voltage. of data centers of large, you know, battery power plants of solar facilities. Really, it is the highest distribution voltage on the grid around us. So if you look at the wires on top of a pole, you know,

specifically the ones that are way up top, you know, the highest voltage those will ever be is 34,000 volts. So we're going after all of the distribution voltages in the world. Europe is the same, so as Asia. And our first product is DC because that's about a 500 gigawatt market, growing quickly of, you know, data center, solar and batteries. But future products will be AC to AC,

and that allows you to do, you know, all the utility use cases and use cases inside of, you know, commercial buildings like this building we're in right now, they can all benefit from AC to AC. And what is, what is it actually doing, right? It's power semiconductors and software, and instead of converting voltage at line frequency using large coils of wire around like magnetic steel

in a bucket of oil, which is how transformers generally are done, you're doing it at really high frequency with much, much smaller, simpler magnetic materials to produce called like ferrites using switching devices. And you've charged a smartphone before, like a little object, that little power brick that you have in your hand, you know, it's doing conversion from 120 or 240 volts AC to 5 volts DC to charge your phone.

And it's switching at a million times a second. There's like tiny little GAN, gallium nitride devices, switching a million times a second, voltage across a transformer that's smaller than a pencil eraser. And, you know, if you remember back to maybe you had a laptop in the 90s or something like that, or you did, I don't know. And the giant power brick that you were carrying around that was really hot when you stuck in your backpack after charging your computer,

like just in a couple decades, you've seen more than an order of magnitude power density improvement there and efficiency improvement. Now you can do like multiple outputs and the seven different voltages. And we're trying to do the same thing, but not for commercial electronics, but for industrial scale electronics.

And our building block, as you said, it's a modular architecture. So that 5 megawatt, Harenlink, it's got 30, 165 kilowatt, you know, modules inside. The product itself was fail operational. If one of those fails, we just keep operating. So we're really oriented and rugged. A whole building could power off of one of those blocks, right?

You can switch your distribution to some modern. Yeah, I like to. For control. Yeah. I like to think of it as though, like if the grid is civilization itself and everything is routed in AC and there's some much, much better way, you can, like, you know, metamorphosize the entire grid like civilization can regrow

off of a new architecture of moving power and I of course like to think about the first power that you put on another world like if you put a megawatt on the moon or on Mars distributing that will like set the precedent for how you do like what the technology type will be will you plug into a DC thing

or will you plug in the two prongs that we're used to in the US at least well either way because I'm not going to take a big stance one way or the other It's like you should be using software and semiconductors, you know, which are higher efficiency, you know, much less mass and size, especially that matters when you're going to space. Like, you don't want to scale. I think there's less oil. I think the big can of oil, the big bucket of oil thing for the transformer, it's hard on Mars.

Yeah, exactly. It's limiting. So it's really said. Several times of steel and oil. It's just the leapfrog thing, right? Like let's use software and electronics rather than like mechanical systems to accomplish our power distribution. We went like super nerdy on this, which is perfect. But our reactor, like Flanameli makes DC power because we actually run a really compact power generator.

Perfect, merit. And it runs. Yeah, exactly. So the greatest civilization, you are pushing civilization that inch forward. Yeah. It's civilization anew, right? And it can grow from the edges where our system makes sense for people with a critical need for power, for resilient energy at a military base or hospital or for disaster relief. Yeah.

Yeah. And I think that's... The reason you have to use AC actually is the way to move that power. And the idea that the grid can grow and move from the edge is just not something that we've really been able to process for the last 50 years in the U.S. Like the grid has been a very top-down project. And if you want to attach back into the grid as all these data center people are realizing now, it's a huge nightmare. It's a nightmare. And so how do you like make it easier to do that kind of organic...

Part of that is because the underpinning of the grid is these mechanical systems that are not fast responding that don't have a lot of telemetry. No software. If there is software, it's very slow to respond. And so you don't, it is harder in a world like that to imagine a bid directional grid, right? Yeah.

It's the central planning from inside out, you know, when you're thinking about protection and like, you know, can I stay within the load ratings of these lines? is when you don't have the infrastructure to dynamically control it the way you want, like you're stuck. So I think it's an enabling alternative. Yeah.

I think one of the things that it could do for people is you have a DC battery. We go DC to AC, right, on these, all these, to attach some like large battery system. Yeah. And then maybe you have a solar grid, but that's on some different DC voltage. And that also needs to get converted AC to them,

put on a grid to use it. But you could have little cells, like these things that we grow at edge. It could be like a megawatt hour battery packs, a few megawatts of solar for during the day, and a reactor for at night. And all those things actually merge and work perfectly on a DC grid. Yeah. Like a DC microgrid. That's, that's sort of what it, I think, what it could do, right? It's interesting for people to think about this. I think it could be demonstrated at some military installation or some other place. And so it's like a fun thing that Drew and I have

chatted about a little bit. But there's no, we don't know when we'll be able to do it. Exactly. Soon. Soon. Let's get critical. That's right. Full power. Get to power. The only thing you consider is compute also is natively DC. So, you know, you're in this interesting world. Yeah, exactly.

Compute batteries, solar. All the name technologies. Micronuclear. Yeah. Actually, all DC. Micro nuclear. Yeah. I like it.

I think it's, I think I've talked to other folks about like what is, what is a modular reactor? Right. Like you hear of SMRs. And it's like this is, this in my mind is like the definition of SMR one megawatt versus like you've heard about this 100 megawatt. SMR. Yeah. Micronuclear is a term no one is using, I think. Well, I think it's new right now.

Like, we just said it. Oh, good. And then someone else said it. And we'll say it. That's it. It's creation. It's done. It's Andle and hammer.

It's in stone now. It's like mainframe versus PC, but flood nuclear. Yeah. And the PC one, like the data centers look like PCs that don't look like mainframes. And there's a reason why. Yeah. Well, SMR can represent like a hundred megawatt thing. Building it, digging a big hole in the ground.

It's not necessarily going to run. It has to run faster. than 60 Hertz for it to be DC source, right? Actually, I make AC because every heat engine, you spin something. But we do it at such a high speed that we then use active rectifiers to convert to DC as the first step. So I think one thing that's interesting about both of your approaches, and I'm curious how much of this comes back to your time at Tesla's basics, respectively. But you're both very focused on manufacturability and modularity and modularity of design.

Yeah, maybe you should talk about it. But be curious here, both of your thoughts about that. I find that there's this sweet spot between like capital investment and like manufacturing cost of goods sold. And you're always trying to find that with any product, right? And if you just compare stick building a power plant in the field versus, you know, fully integrating like on a highly automated line, that same function, the total cost will always win. If you can do it on a highly automated line. And so, you know, we're building for our first factor, we're building a 40 gigawatt factory. And people are like, wow, that's a lot.

Well, contextualize what is 40 gigawatts? Yeah, 40 gigawatts a year. So it's about 10 to 15 percent of the X-China market for our product category. And it's equivalent to half the state of Texas in peak power, if you want to think about it in that term. That's kind of useful. I'd say 4% of the whole country is electric power, isn't it? Yeah, something like that. Between 500 gigawatts and the Turawad, depending on how you think about it.

So, yeah, 40 gigawatts a year is, I mean, it's significant in the overall U.S. And why 40 gigawatts? One, you know, looking at it from market sizing perspective. But the other is, hey, like right around 60 second tack time is where you maximize that, like, capital efficiency of building a factory. And if you look at our module size and you look at 60 second tack time, it sort of works out to 40-degree. gigawatt spot. And, you know, I, we're going to ramp that factory. I hope, I expect us to exceed the demand for that factory. But you get so much quality and cost benefits going for full on as much automation as you can, not too much, but as much automation as you can. Yeah. It's more than just that. But that's the first two things that come to my mind, quality and cost. Yeah. I'm using it as such a huge scale. I definitely, I definitely 100% agree with less, like, I think you said, stick building. And it's like the opposite of that is mass production.

Right. Think about doing it all in a factory. And that's our entire approach is like a nuclear reactor. It was pretty radical for a nuclear reactor, the idea that you would build it in a factory. Absolutely. Yeah. So it's really, we're doing nuclear reactors as products for the first time ever. And it's so that you can just, you can say, yeah, I want nuclear power.

And we can deliver it. It operates. And then when it's done operating, we take it away. And there's no waste or other tricky. consideration you have to make. It's totally safe on the customer site. We actually use a meltdown-proof fuel. And it makes this like completely new way,

I think of thinking about nuclear power. It's just, it's in the options list. And I think it's, it wasn't even before. Not only is it there, but it can look better than almost every other form of power. And I don't like to, you know, imagine that it's the only thing you want to do.

I really like the idea of like solar and battery and nuclear and like put that, that whole block somewhere. Yeah. Right? And use all of the free energy that's out there because the sunlight is free. You put up a panel, you're getting it.

It's very cool. But also the uranium is free. It's in the ground. It's not flying through the air. It's there. If we take it and we use it before it spontaneously undergoes fission, it just goes away,

which has been doing. So like when the Earth formed a few billion years ago, we had like 128 times as much uranium 235. So we better get it before it's gone. Yeah, the same way as like you'd put up a panel and catch some sun. Like, take it up out of the ground to use it.

Don't let it turn into radon and other stuff that is actually harmful to someone's health. It's like it's more dangerous. from health perspective, leave it in the ground, and it's a total waste of the power. And there's a bunch of cool ideas like this. And a new way of really seeing nuclear that I've written a thing called Adams for Prosperity.

That's on our website. I've released it about a year ago, but it has this and like, how should you think about radiation and waste and reactors and what's possible? Yeah, I think there's another thing that you're not stressing enough about that infield work content being really low.

So, you know, when I was a Tesla, I was responsible for the megapack product development and then mass manufacturing and the business of selling them. And the less involved, the onsite project is, just the faster everything about it will go. Yeah, the per room. Racks of batteries. Yeah, just like land the cabinet.

You know, in fact, we even work past the pad, like we got rid of the concrete, wherever the seismic would allow. It's awesome. It just did soil nails. Because, again, you're disturbing less. dirt, you know, there's less concern about, like, you know, how much you need to grade or what are you going to do with, you know, civil and architectural type scope. The local community, especially in your case, if it's like a temporary installation, they're

going to feel less concerned about it. And the fact that you can just pick it up and remove it, if for whatever reason it needs to be, which you can also do with megapax. It eases this, transmit this transition into these new technologies. And consider that, compare that to like a giant, nuclear cooling tower, visible for miles around. Yeah. It's such a different approach.

And in today's world where like not everybody is a Ymbi, A Yes in my backyard, having a, you know, rapidly deployable, let's say, I thought NIMBY meant nuclear in my backyard. NIMBY. So we should re-ran NIMBY. Let's bring it back. YMD is a new. Dang it.

I got a problem here. I love it. I love it. Sorry. Go ahead. No, I mean, I think that's what you get with that modular approach. You get logistic simplification. You get quick install and you get simpler permit and access to no skyline.

It's really infrastructure free. Just clean power wherever you want it over a weekend. Well, but so, I mean, you built, you helped build factories that made rocket chips. How it seems like building a factory that builds a nuclear reactor that's pretty hard. What's that, you know, what's the experience then like? So what you're just getting. Everything is a factory. We're building a nuclear reactor in our first building that we had,

and it's like a 70,000 square foot that we have right now. We have two buildings. We had to get a second one in November because we filled up the first building because we start doing a little bit more vertical integration, a little more of the machining in-house and all that's happened to us already. But we'll be able to build up to 10 reactors in those facilities that we have. We both know in-law operate out of a tent. So this is like quite a bit fancier to have.

real walls. But in Tennessee, we have an 80-acre site. We have the first building going up, which is made for just fuel handling because that's the tricky bit. And so we're working the regulatory permit path right now on that. We should be able to put fuel reactors there. And so we'll initially be building everywhere. We can possibly find a building with enough power, right? And then moving all the parts to Tennessee to then get the fuel loaded there, to then take it up to the customer site. It's nice that it's mobile. So we can actually do all of that. But the factory itself is a bunch of other buildings on that site and most of it is like normal assembly work like you've got big structures that you're welding bolting things together right putting wire harnesses and things on

on the unit but the factory will have to evolve multiple times also right a factory before gigafactory right is a different thing you know the process right automate is last yeah yeah yeah yeah if you're trying to delete the steps and done all the other yeah all the other smart stuff exactly of course yeah So yeah, we're just building and learning what is actually the factory look like. But parts of the production line that can be automated and should be will then go in those newer buildings. Yeah, we're doing the same thing, right? Like we're building our first prototypes largely by hands, you know, in our engineering facilities. We're doing about 10 of those this year.

We'll do another like 30 to 50 prototype systems this time in our factory location, which we hope to announce next quarter. And then only from the learnings of those two builds, we go and automate from there. And that's the, you just got to get the reps. How should we think about how micro reactors fit within the broader energy landscape? Do they compete with large centralized plants? Do they complement then? Are they serving in different categories in demand?

How should we think about it? Yeah. So they're definitely an off-grid product. So they don't at all compete with larger reactors. Really, if you can build, if you have time to dig a big hole in the ground and put a reactor in that way, then you can do a larger reactor, maybe five or ten times as big is the one megawatt size that we're looking at, and it's going to win on economics.

It definitely should. We're already using one of the fanciest forms of fuel and that it's so that we can set it up anywhere and have it not be a risk to people or facilities nearby. And so we don't compete at all with those things. One of the ways I like to talk about this is you can run a diesel generator or you can run a nuclear reactor, and you're really deciding between those two things.

And we don't beat like super cheap diesel. Like we beat diesel at like $6.50 a gallon, that kind of a number. So that's where our initial customers need to be. But if you go start looking at what people pay for diesel and what they pay on the edges, not on like the center of the bell curve, the average for like a country or an area, like you look at the tough regions. They're paying a lot. And so there's plenty of customers out there.

And one's like some examples, I think. Oh, like $10 gallon is the average in Hong Kong. I think like Iceland and Scandinavia, Northern Europe. those regions are like $7, $8, $9 per gallon for a whole country, actually. So like it's very easy to see the market is massive. And islands. Yeah. Yeah, islands.

Absolutely. I mean, Hawaii is pretty high electricity costs and it's, I think, 80% diesel powered, actually. It's got wind and solar that make up the remainder. But yeah, you could have a cleaner form of power, right? No emissions. The nuclear reactor operates and then rain, it takes it, and we handle all the complexity. But the amount of power people need, right, they need in the gigawatts for the grid. And so we don't really do that.

We have the niche customers on the edge and we don't want to make, right, thousands and thousands of reactors. At 50 a year, we'll have something in the range of like a thousand or two at the most. But we don't consider, we don't look at it and go, hey, could we make it work for 10,000? There's different products and we can do it at better economies. And there's a couple of ways to do it. But we're, Radin doesn't want to dig a hole in the ground. and solve that other miracle. It's too many miracles.

I think it's important to be able to do it again, but it's not on us to fix it right away. Yeah, series miracles. You don't want to have too many in a startup. Yeah. You need some. Yeah. Yeah.

One miracle that leads to then a product and revenue and right, that's the way. And then you have time to think about another miracle. Totally. So you mentioned that we're very early in the nuclear industry, you know, we're in pre the nuclear industry in some sense. So what is the mind? milestone or the KPI or what would need to be true for us to say we as a country,

the nuclear industry is here and flourishing? I think a couple of things. So we could have access to nuclear fuel and enrichment that are in completely competitive free markets where there are innovative startups fixing and solving those challenges. We should have waste storage facility that's some centralized repository, which is way safer. for the existing nuclear fleet that's operated since the 60s. That's an unsolved problem.

And those things alone would cause everything else to flourish because we already have this middle layer of me and a bunch of other startups, trying to get fuel and operate reactors and then if we're able to, as a country, really have a better system to deal with nuclear waste, which actually radium doesn't need. Like our uniquely at this really small size,

we can just put it into dry cask on about 10 acres of our 80 acres site. and that works for like 60 years worth of reactors and we can always expand and do more. The nuclear waste has got high reactivity elements that the last like 100 years. And for that, it's pretty benign. But we already have a waste isolation pilot plan

in New Mexico, which is like this deep borehole down inside of a salt structure. So it's like a salt dome. This is where defense waste already goes. And they just said they were going to build it and they built it. And meanwhile, we structure, struggle still on the DOE side to build a repository for big nuclear plants. And because of that,

these gigawatt scale plants are operating, generating nuclear waste and they have to store it at the same site where they're making power. And in California, this is like coastal regions that are risky, where like you can have a tsunami or something, instead of taking it and putting it in a saltland structure in the high desert where there's no water, no risk of like certain natural disasters. So it's just a smarter, safer, better idea. And we don't. don't do it. Yeah. And actually it was a huge cost.

It's a commitment. That's right. Like, you've got to demonstrate that commitment. It's the, and it's also the NIMBY transition, the not in my backyard to. Nuclear my backyard. Yeah. We need that. Well, you, I mean, I think that you mentioned also the sort of nuclear fuel supply chain.

It's something Drew, you and I've talked about on the, on the power electronics side, how important the supply chain is, how focused you are on it as well. I think you're lucky that your biggest silicon carbide supplier is a U.S. company. It's a technology developed in the U.S. But what other parts of your supply chain, and are you worried about? Yeah. Or thinking about. Yeah, for sure. So I mentioned there's ferrite is a pretty important

ingredient in these high-frequency transformers. It's basically just iron oxide, so it doesn't have maybe a little sprinkling of manganese. So it's not, there's no rare earth in the ferrite, but the world's largest ferrite companies are in Asia, right? It's like every other, you know, complex supplied good. Now, there have been ferrite manufacturing facilities in the U.S. before. I'm working to bring them back.

In fact, one of them is nearby in Georgia, and it could be brought back with some coaxing, and I'm working on it with the parent company. Another is thin film caps capacitors. There's a decent amount of, they call power electronic. but they should really call it like power capacitors because the things you mostly see are the capacitors. And the same sort of thing, you know,

supply based largely in Asia, working with those same vendors to bring it closer. No rare earth materials there, you know, mostly polymers and, you know, thin copper or thin aluminum conductors. We already have a pretty well-established, copper and aluminum supply chain. You're just saying like a small wire gauge?

I mean, it's like micron thin. Sheets. Sheets, yeah. Okay, yeah. Copper aluminum and foil. There are really foils. And then that's what's in a thin film. Okay, sorry.

Got it. So, yeah, those are, I think, the critical aspects of power electronics. Everything else is already, like, very abundant and easily this for us in the United States. So, you know, your plastics, your sheet metals, your aluminum castings, boss bars and things. So, you know, because of that, we can really focus on those three key commodities. And we do have plans for. each to both near shore and onshore if they're not already. And like, I'm excited about that because I see power electronics really moving from the like

device skill I was describing of, of, you know, charging your laptop or where it's largely been stuck more recently, like in EVs and solar and storage to the grid itself. And when you do that, you know, you're like, oh, 40 gigawatts. So it's the state of Texas, but it's actually not like that because you have power conversion all the way along the way, right? You have some power conversion going on at the generating facility to go from some lower voltage to a media voltage, like 34,000 volts. And then you have another power conversion system going from that intermediate voltage, 34KV, let's say, to the main transmission voltage, hundreds of kilowatts. And then you have to do it again on the other side when you get into the community. So you might have an 80 gigawatt peak grid in Texas,

but you actually have like 800 gigawatts of power electronics actually supporting that grid. So I'm excited, I guess what I'm saying is while, and this is a useful piece of context, last year about three terawatts, and these are big numbers, of power semiconductors when it's electric vehicles. And the peak grid power is less than a terawatt in the U.S. So these are really like on the same scale kind of opportunities. Like this is solvable. It's a solvable one, yeah.

And I'm like really motivated by these, how much these supply chains have scaled up to support electric vehicles. And electric vehicle growth is slowing down, which means we can take that momentum and bring it into a new problem statement, which is power for data centers, power for industrialization, power for economic growth and prosperity and for sustainable energy. So, yeah, it's an exciting time, I would say. Let's talk a bit about data centers. There's a lot of controversy around them. How should we think about what is the impact? Are they causing problems on the grid, et cetera? Yeah, it's a, there's two sides of this one.

But I think in general, like if you zoom out, data centers are overall going to be really good for the grid. And I'll kind of explain maybe way they get a bad rap and how that's going to change. So just I think yesterday I saw a headline about two gigawatts of data centers turning off instantaneously in Virginia over the weekend or last week or something like that. That is the reason I see why there's a lot of concern about data centers on the grid,

like from the grid operator perspective. And they are designed to date. They have been designed to do that, right? They want to keep their compute up. They want their six-nines. So anytime there's anything funky on the grid, they isolate and run off of their backup generation or UPSs and then ultimately backup generation. And when a data center is 10 megawatts in a grid that's hundreds of megawatts or gigawatts, that doesn't matter.

But when you're building gigawatt data centers, it starts to really matter. And the grid stability is at risk. And so that is very solvable with software, modern power electronics, you know, dynamic grid forming controls in your rectifiers and the data centers. They can stay online through those cases with a little bit of energy storage and actually stabilize the grid rather than DC. stabilize it. So that is like a solvable one. But it is, it is a problem that is real. Like it happened in Washington State, it just happened in Virginia and needs to be resolved as these data centers keep becoming bigger and larger percentages of the grid. And I think it's a, it's a function both of the data center design up until now and how they are able to connect to the grid, which today is,

you know, very dumb systems. Yeah. And it needs to be more interactive. Absolutely. And it can be so much more interactive with, with better software and with more understanding of the capability. Yeah. But Then there's this other commentary about how data centers are going to increase rates, which I think if doesn't make sense to me on a big picture, and it's really simple like physics of electricity rates, right? Electricity rates are costs to maintain the electricity grid or to deliver electricity altogether divided by total kilowatt hours delivered, right? And the data center customer is like the ideal customer.

They're consuming like near their maximum power. almost all the time compared to like your house, where you're like maybe at 10% of the maximum power of your house, like an hour a day, right? So they are the best customer to serve in terms of delivering more kilowatt hours. And then the way utilities generally do is they like take all the kilowatt hours in and they look at all of their costs and they, you know,

spread it across everybody. So the more data center load, like more loads we have like data centers, like factories that are steady constant loads, the cost of serving electricity to everybody will go down because they are the they are increasing that numerator the denominator right like utilization yeah the utilization is going up right um so the average is getting better for everybody yeah and and i think you know there's concern about the power side oh well there'll be enough power but i think what we've seen and actually last year the u.s had had one of the highest power additions to the grid ever and this year it's by far going to be the highest capacity addition to the grid ever So new power is not the problem. Delivery is the problem. And data centers, by increasing the utilization of the delivery system, make delivery more affordable.

So I think net net, they will actually drive rates down. Cool. I think that's a good place to wrap. Go to Drew, thanks so much for coming to the podcast. Thank you. Thank you. That was great. Thanks for listening to this episode of the A16Z podcast.

If you like this episode, be sure to like, comment, subscribe, leave us a rating or review, and share it with your friends and family. or more episodes go to YouTube, Apple Podcasts, and Spotify. Follow us on X, a A16Z, and subscribe to our substack at A16Z.com. Thanks again for listening, and I'll see you in the next episode. As a reminder, the content here is for informational purposes only. Should not be taken as legal business, tax, or investment advice, or be used to evaluate any investment or security,

and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see A16Z.com forward slash disclosures.