Collectively, the electric stack (batteries, motors, chips, power electronics) has gotten about 99% cheaper since 1990. It is quietly remaking every machine that turns electricity into motion, heat, or work. Problem is, China makes most of it. Sam D’Amico, co-author of “The Electric Slide” and founder of Impulse Labs, is betting the US can still claim a piece of that stack. He’s starting with a battery-powered induction cooktop. We talk about why domestic manufacturing matters, what America would actually have to build to compete, and how Silicon Valley’s founder world really thinks about climate change, geoengineering included.
Energy analysis and journalism have traditionally focused on the supply side, but in recent years there’s been a welcome turn in the other direction, toward the ongoing revolution in the devices that consume energy. What’s happening, around the world, is an explosion of electrotech. The devices that translate electricity into motion, heat, and work are getting exponentially cheaper, smarter, and more sophisticated.
Electrotech basically comes down to four technologies: batteries, electric motors, chips, and power electronics. Together, that “electric stack” has gotten about 99 percent cheaper since 1990, which is why cars, drones, boats, and stoves keep going electric — and why, eventually, everything is going to go electric.
Currently, China manufactures most of that stack, including roughly three-quarters of the world’s lithium-ion batteries and 90 percent of the magnets that spin the motors. That leaves the US in a weak and perilous position.
That, at least, is the argument of “The Electric Slide,” a sprawling, 40,000-word essay published in the newsletter Not Boring last year. It says while America is betting the future on AI, China is betting on the hardware, and China has made the smarter bet.
One of the co-authors of that essay, Sam D’Amico, is trying to put the theory into practice. His company, Impulse Labs, makes a drool-worthy high-end induction cooktop with a battery built in. That lets it push ten kilowatts per burner while plugging into an ordinary 120V outlet. Impulse is now licensing that core platform to bigger appliance brands, and Sam has his eye on “every mid-sized machine that runs on electricity.”
I thought it would be fun to talk to Sam about why the electric stack matters, why he started with a cooktop, where electrotech might go next, and how much of it America can realistically manufacture at home. Let’s do this.
Chapters
00:00 – Introduction
02:52 – From VR headsets to kitchen appliances
07:07 – How a pizza oven inspired a battery stove
10:07 – Peak demand, at home and on the grid
13:25 – Which appliances could use a built-in battery
16:46 – Electric stack 1: the four technologies
20:28 – Electric stack 2: power electronics and motors
25:21 – China bets on hardware, the US bets on AI
30:08 – Why not just buy it all from China?
33:51 – Why engineers need to be near the factory
38:39 – How much does the US need to make at home?
45:19 – US EV policy and letting Chinese cars in
47:49 – Batteries in appliances vs. one home battery
52:04 – Impulse’s plan: its tech in other brands’ stoves
55:17 – What’s next after the stove
57:14 – Has Silicon Valley stopped caring about climate?
Resources
People & Organizations
Company & Industry News
Impulse Launches “Powered by Impulse” Platform with Thor Kitchen as First Major Appliance Partner
Impulse and Zephyr Partner to Launch First Battery-Integrated Cooktop in Showrooms Nationwide
Trump Admin Bans “New” Foreign-Made Inverters. What Does That Mean?
Books & Articles Discussed
Related Volts Episodes
Transcript
David Roberts: With no further ado, Sam D’Amico, welcome to Volts. Thank you so much for coming.
Sam D’Amico: Thanks so much for hosting.
David Roberts: All right. So Sam, you were in consumer electronics in your previous life, and I’ve been sort of curious whether it was this kind of big picture thinking about electricity and the imperative of electricity and the electric stack and the evolution of global markets, et cetera, that impelled you to get into electric devices, or was it just something about cooktops specifically that lured you into this? So what prompted you to switch over?
Sam D’Amico: That’s a great question. So basically the storyline, it’s like in the mid-2010s, I definitely wanted to kind of start my own company, definitely wanted to work in hardware. I was at the company formerly known as Facebook at the time, and I was working on VR headsets and game controllers and stuff like that.
And I’d realized that, you know, if I wanted to do something like this, like something of this level of ambition, make a device of that level of awesomeness, well, you needed to do it inside a big company. And the reason why you had to do it inside a big company was... This has now changed as of AI and all this other stuff, but if you wanted to make a VR headset, you needed to have an app store.
David Roberts: Right.
Sam D’Amico: And in order to have an app store that was any good, you needed to have, you know, developers on there. And for the developers to want to develop for your product, you know, they would need to have millions of users on the other side. And so I kind of realized that this, like, two-sided marketplace problem was, in some sense, intractable for huge swaths of consumer electronics and hardware, and kind of the reason why you’ve seen consolidation at the top, where it’s like, okay, only if you’re as big as OpenAI can you decide, oh yeah, now we’re gonna do a big consumer hardware swing, which is what they’re rumored to be doing.
This is something where if you’re a, you know, individual user, or you’re a smaller company, or an individual founder, you’re like, “Okay, well, I need to get my millions of users before I can even get to the next thing.” And so that was a key thing that I was like, “Okay, that’s the first thing.” The second thing was, with stuff like smartphones and VR headsets and all those other things, they had already kind of been touched by these electric stack technologies, you could argue.
You’re kinda pushing the limits of, like, what the off-the-shelf chips are doing. You’re gonna... Like, Apple obviously has a huge fundamental lead in their smartphone chips, for instance. And, you know, to make a better product, you need to actually have, you know, a billion users, enough demand to do that, and being willing to customize things from the display, the camera, the chips, et cetera. And so I kind of looked at this in a way where I was like, okay, so there’s like three reasons why you really shouldn’t go do the smaller devices now. Because, like, you’re gonna get the floor mopped by like these, you know, trillion-dollar companies.
And then the other option is like, okay, well, you can do the larger machines. You can go do cars and rockets and all this other stuff. And I’m like, maybe it doesn’t make sense to compete with Elon on the first go-around, was kinda the thinking. This, like, medium-sized machine market is kind of everything left, and these are devices that have, you know, basically been built on kind of like the previous generation of technologies when there was, you know, a brief wave in the mid-2010s of kind of smart appliances and putting, you know, smartphone guts into stuff. But, like, there hadn’t really been a ton of core innovation in how the systems worked and how they were, like, architected.
And so I’ve zeroed in on that as like, there’s an opportunity here. And then the second thing is like, well, you don’t need multiplayer for a stove. Your friend doesn’t need to have the same stove as you to make the stove good, right?
David Roberts: No network effects.
Sam D’Amico: Fortunately or unfortunately, there are less network effects in this space. And so that means that you could start with, you know, selling one fantastic product to one person, and you’d still be able to, you know, be successful there. And so that was just an attractive thing. In that kind of early formation process of the company, I was like, I kind of realized that these were all key pieces that made this workable, whereas like me being like, “Hey, I’m gonna go raise for like some sort of newfangled AI smartphone thing,” or whatever, like that would’ve been a much more difficult ask, and arguably like more difficult to actually be successful at as well.
And then the second thing was like, okay, well why a stove, period? And it was the realization on my own was, this was like 2019. I’m at this like IEEE haptics conference in Tokyo because I was visiting, because I was working on the VR game controllers at the time. And my pizza comes out of this pizza oven in like 45 seconds and I’m like, “Okay, this is fantastic. How many watts did that really take to cook that pizza?”
David Roberts: Mm. An electric pizza oven.
Sam D’Amico: I was like, okay, so electric pizza oven. I like hosting barbecues at my house, so I like to cook for people. And so the realization was like, oh, and then I’m noticing that the guy’s not continuously putting pizza into this pizza oven. It’s like he puts the pizza in, comes out 45 seconds later. There’s a period where like the oven’s empty and I’m like, okay, so if you had a battery, you could just buffer the spikes when you actually have the pizza inside, and suddenly, you know, you could trickle charge that pizza oven, you know, over the rest of the day.
David Roberts: Right. But whereas you can’t get the wattage you need straight out of a 120-volt standard outlet.
Sam D’Amico: Yeah. I mean, if you look at like toaster oven pizza, you don’t really get good results out of those. That’s effectively what you’d otherwise get, you know, from a 120-volt outlet. And so this was kind of like one of the inception points where I was like, oh, there’s now a set of technologies, back to the electric stack framing where I didn’t have as cogent of a description at the time, but it’s more like if I could take basically the guts of an electric vehicle and I can apply it to spaces where it hadn’t been applied to before, well, fundamentally I can make a better device.
And I can make a better device without having to spend a billion dollars on R&D for new displays, cameras, all this other stuff, because the automotive industry already did this for me. And in some sense, like the stack was ready effectively.
And then, you know, with a bit more kind of like startup journey details, it was kind of like, okay, well, what’s the device that probably has the most touch time in the home that where also the flame is actually the UX for, in the case of a gas stove. And that, that’s what led to doing a cooktop first, was kind of like, okay, most people don’t know how their dryer’s powered, even if it’s gas. Like that was something I ended up learning over the course of... I think I smelled a gas leak in my house, and like the guy checked my dryer and I was like, “Oh, I didn’t realize my dryer was gas.” And so meanwhile, like you definitely know if your stove is gas.
So that was kind of the rationalization for like why pick on the stove as the first thing. But then it’s also just like a great thing to pick on because it’s got the most disparity in peak power versus average power. I think the Energy Information Administration says that an induction cooktop is how many watt hours per day? It’s like 200 to 300 watt hours per day.
But the peak power of a cooktop with all the elements turned on could be like, well, except for us, but like maybe like 7 to 10 kilowatts or something like that. And so you end up wiring like half of your electrical panel if you’re in an urban area to something that’s using like potentially less power than your computer.
David Roberts: I mean, this is just basically at the household level, the same problem we talk about at the grid level over and over again, which we’re being, up till now, we have been forced to build infrastructure for peaks, right? Which means a lot of the time, a lot of the infrastructure is not being used, and batteries solve that problem at the macro level, and now they’re solving it at the micro level too.
Sam D’Amico: Yeah, and I think there’s a metaphor here where if you think of the grid, the joke I would say is like we’re evolving from like the Tesla, the Thomas Edison electric stack to the Tesla, the car electric stack, you know, with a lot of these transformations. But the metaphor that I think has been very useful for me at least in terms of thinking about the grid is like the grid before the advent of storage looks kind of like a rotating shaft that’s spinning 60 times a second, or 50 if you’re in Europe or whatever.
David Roberts: Mm-hmm.
Sam D’Amico: And you kinda have to clutch into that shaft. Like to participate in the grid, you gotta basically make sure that like, if you’re trying to add energy to the grid or whatever, you gotta like turn that on and like, you know, slowly dial up the friction until you’re spinning at 60 times a second too. And if you do that poorly, like obviously this is why the interconnection queue is so long.
This exists at the micro level with like, “Oh, I wanna get permits for home storage for my house,” to the macro level of like, “Can I plug in a data center?” And so that was kind of like where electricity 1.0 gets you to. It’s like, okay, you’ve got a spinning shaft. You could even think of it even simpler. It’s like, where did the grid come from originally? It’s like, well, you were running factories, and the factories needed to use each other for backup. And what were those factories running on before electricity? They were running on water in many cases.
David Roberts: Mm-hmm.
Sam D’Amico: Like the Lowell Mills in Massachusetts and stuff like that. So it’s like, okay, you’ve got a water wheel. The water wheel is spinning, let’s say 60 times a second. That’s going into a shaft. That shaft is then powering all of these machines that are all running in synchronized, you know, off of the power being pulled off the water wheel. Or like, okay, then we evolve to generate electricity and then power all the machines with electricity, but again, they were still spinning at that same frequency, basically. And so what storage lets you do is it changes it from this sort of like, you could almost think of it as like a single shaft rotating to a network that looks like the internet.
David Roberts: Mm-hmm.
Sam D’Amico: Where you can basically be like, “Okay, well, there’s now packet buffers everywhere on the internet.”
David Roberts: Right.
Sam D’Amico: Battery storage, could be pumped hydro, et cetera. And now that makes the grid a lot more resilient. It makes it not like one single spinning shaft. It makes it very much a, like, self-healing network. And I would argue we’re, like, only partway through that transition. And this transition happens at the device level, obviously. It started with, like, your smartphone.
David Roberts: Mm-hmm.
Sam D’Amico: Actually, before the smartphone, it was like, what were the expensive things that needed lithium batteries initially? It was, like, camcorders or something like that. But basically it was, like, expensive consumer electronics now to all consumer electronics now have the lithium batteries in them. Okay, and then automotive-derived technologies on the other end, and then we’re kind of in this pincer movement where every device in between is getting filled in now at this point with having some level of battery storage or at least, you know, kind of smart computer-defined power electronics, basically.
David Roberts: I’m curious, so the cooktop is perfect for your purposes in that the difference between average draw and peak draw is huge, and that’s where batteries are handy. They can store up, and then they can be the ones to supply the peak. I guess I’m trying to figure out how big that space is. Like how many appliances fit that description?
Sam D’Amico: I can give you a good example of this is like, I think it’s pretty much everything.
David Roberts: Really?
Sam D’Amico: Let’s use your computer as an example. And I think this is kind of funny because it’s like your laptop already kind of does all this stuff. You know, it’s trickle charging off of its charger. If the GPU’s drawing a lot because you’re playing a video game or whatever, it could actually locally discharge during that period or maybe not charge as fast in some cases. And so like we’ve already seen this sort of stuff. Like so on computer side it’s like it just turns out that like you’re not drawing all the power all the time.
Actually, I can even give a better example is like the VR headsets where you spend all this compute to render the frame. You try to do it as fast as possible, then you show it, and then you basically wait and you often are like waiting and it’s, the chip is asleep after the frame has been rendered basically. And so that’s how you can get the battery life to not be like 45 minutes and to maybe get closer to like two hours or something like that.
And so it turns out that there’s kind of like lots of things like this. Data centers are like this. If you look at what’s happening with like training where like they have to like basically you do a ton of compute across all the racks, and then they have to share all that data. And then in the sharing of that data, that takes less power because it’s bandwidth constrained, not compute constrained. So that’s why these facilities all have giant batteries outside of them right now.
David Roberts: Yeah, but I’m talking about the medium-sized electric machines that you are honing in on.
Sam D’Amico: Yeah. So like oven preheat’s another example. It’s like your oven doesn’t use a lot of power except when it’s preheating, is a good example. And then most of these devices are just off all the time. So like another example is like your laundry machine might be relatively consistent in terms of how much power it’s drawing.
But like I live at home by myself, so maybe it’s like, maybe a little bit of an exception, but I use it once a week. And so as long as you can capture kind of like one session for use of all these devices, like you’re pretty much good in a bunch of the cases. And then even within that session it’s like, well, is your dryer running full bore all the time or only for part of the cycle, basically?
And then when you kind of look into that, you’ll notice that the same thing I mentioned where it’s like, okay, well, you’re using your stove an hour a day, but then within that one hour only two minutes are at max power. And it kind of like fractally is true through that whole thing. It seems like this pretty much happens on everything. Your water heater, it’s only when you’re taking a shower or filling up a bathtub.
David Roberts: Right.
Sam D’Amico: Otherwise, it’s like, okay, yeah, you put a little bit of hot water into your sink in your kitchen to wash some dishes. That’s not material. And also like, are you running water all the time in your house? No. And so HVAC is probably the one where it’s like it has more continuous load profile versus this, but even then it’s like, okay, well, if you’re looking at AC, there is a peak period, right? The grid even like prices it in a lot of cases with time-of-use rates. And so it may push you into more of a like four hours of storage type domain, but it’s still something that you can largely address with energy storage versus not.
David Roberts: Okay. So I wanna pull back the lens a little bit. We kind of jumped into the guts of it real quick. I wanna pull it back the lens a bit and then get back into the guts later. But let’s just talk about, for people who are not familiar, maybe for people who have not read “The Electric Slide,” or one of my ongoing agonies is that, you know, you and I are aware that electricity specifically is the thing. Electricity is the thing, storage use, et cetera.
But I think a lot of people just have kind of electricity and clean energy, capital C, capital E, kinda mushed up in their head. They don’t have electrification as a distinct category with distinct technologies and distinct dynamics in their head yet. And my mission in life is to try to do that for people, try to help people see that clean electrification is something distinct from the larger category of clean energy writ large. So maybe just talk a little bit about the electric stack. What are the four technologies? How do they fit together? Why those four? And, like, what is the sort of revolution that you are trying to capture in “The Electric Slide,” that big essay?
Sam D’Amico: Yeah, that sounds great. So basically, the four technologies are batteries, power electronics, electromagnetics, which include like electric motors and other systems, embedded compute. And the reason kind of all these things are together is like this is the first time that there’s a fully coherent set of technologies that are like addressable by software, is the other thing to really kind of underscore here.
David Roberts: Mm.
Sam D’Amico: And so you could argue that Tesla was revolutionary in making the first practical electric vehicle, but in some sense, the Model S being kind of the first car that was produced in some sense like a consumer electronics device, was equally revolutionary, in some sense depended on the fact that they could use the electric stack technologies basically. Like these were codependent, but in some sense, that was as big a revolution as being an EV, period. And so go back to history and kind of talk about why we’re here now versus, you know, earlier, it’s like basically we needed to ride down a bunch of cost curves to make this stuff work.
David Roberts: Mm.
Sam D’Amico: So lithium-ion batteries, I think early developments were like with Sony, and like these things, it really worked for consumer electronics if they were above a certain price, and they needed like exquisite battery life, basically. So this landed in portable camcorders first and then went to laptops in the 1990s. By the time that Tesla was able to start looking into this stuff, like the key revolution that enabled Tesla was like the founding team was like, “Oh, these laptop batteries are good enough, and you could build a car that would have like non-trivial 200-plus mile range just with these cylindrical 18650 battery cells that are already in scaled production for the consumer electronics industry, particularly in laptops.”
David Roberts: Yeah, the beauty of the cost curve, which I don’t know that everybody fully grasps, is that every little increment they get cheaper, it broadens the use space, the application space, the things you can do with them. And then if you broaden the things you can do with them, you create new industries, you create new scale, you end up manufacturing a bunch more of them, they get cheaper. The use case then expands again, et cetera, et cetera. And so Tesla was sort of right at that moment where kind of realization like, oh, they’ve gotten cheap enough now that EV has entered the viable use case space.
Sam D’Amico: Yeah. Laptops effectively enabled an electric sports car. Now, what enabled the subsequent development is actually wasn’t laptops, but it was actually just the EV industry itself in some sense. Once there was a certain class of high-end vehicles available, everyone realized that they could then invest in this and make it happen. Then the second core technology is power electronics.
So basically it’s like, if you go back to how the grid, you know, historically works, like we kind of avoided having switching or storage in a lot of the grid up until like the ’70s and ’80s. It’s kind of crazy. And it’s like if you look at what was done, it was like, okay, well yeah, you have circuit breakers. You’ve got the ability to turn things on and off, like kind of a macro level, but not at the like microsecond level, if that makes any sense.
David Roberts: Right.
Sam D’Amico: And so the invention of the transistor was obviously really huge, but the big story is like, I think in like the ’70s and ’80s, we started getting kind of these like higher power insulated-gate bipolar transistors and higher power MOSFETs. Those have just been improving ever since.
And so again, when Tesla was, you know, putting together their initial like motor controller and all this other stuff, it’s like, okay, yeah, it was possible then because the semiconductor parts existed to actually drive a three-phase motor off of the battery pack basically, or brushless DC motor or AC induction motor off the battery pack. That was also just an enabling technology that had to take time to develop and for the cost curves to go down. Because obviously like when you’re trying to do something with relatively high power, it ends up being a decent amount of semiconductor die area.
And then what’s happened since then has been just like a revolution with gallium nitride and silicon carbide, and now we’re starting to see like data center power where you’re able to do, you know, close to 20 kilowatts in like a single unit of rack space. You’re able to do like all the solid-state transformer work that folks like Heron Power, Enphase have been doing.
Like all this stuff started… And then even more mundane, it’s like your power brick for your laptop can be like half the size that it used to be. That revolution has been crazy and enabling. That means like, okay, you can throw a lot more power around with, in some sense, downstream of software control as well, so you can do a lot more, you know, crazy experiences as well. And then the next piece is electromagnetics, and this goes into like the motors and the supply chain there.
So like basically, I’m trying to remember. It’s funny, is like I wonder if the biggest use case for neodymium magnets a while back was like those toys that you could like construct things out of. That like, you know, kids ended up like accidentally eating and like having to, you know, get surgery for. I forget the whole story of that. But basically the neodymium magnet situation has been nuts because like in some sense, if you actually look at what’s been happening over the past like 20, 30 years is the magnets themselves have gotten better.
David Roberts: Yeah.
Sam D’Amico: And this effectively means like you could have more magnetic field density and all this other stuff, so like the motors have gotten better, and this is like-
David Roberts: Yeah.
Sam D’Amico: And what’s been very interesting with this is only now is like the US government realizing that, okay, this thing was kind of like, in some sense, the rare earth conversation was just ignored because it was like, okay, it was the electric vehicle thing, which was, like, not big, and now it’s big. And then it was, like, toys. And meanwhile, like, investing in mining and materials and all this other stuff is like, you know, historically was kind of, like, underappreciated in the US.
Now everyone’s waking up, but it’s kind of like, I think there’s definite arguments of, like, are we too late and all this other stuff. The other thing that’s really interesting about this is like, well, this is how you turn electricity to motion. And now that AI has just shown that, okay, robotics is probably coming pretty soon, there will be a, quote-unquote, “ChatGPT moment for robotics.” You know, if you’re on tech Twitter, you’ll just see that someone’s claiming that about every other day now.
So it’s probably gonna happen. I’m not saying I’m skeptical here, I’m just saying people are gonna- it’s gonna happen. And so now everyone cares and wants to have motors and actuators built in the United States, much to my benefit as an investor in one of those companies. But basi-
David Roberts: The way I would wrap all this up is these four key components of the electro stack, batteries, motors, power electronics, embedded compute, all have been riding down this cost curve, their own cost curves together, getting exponentially cheaper-
Sam D’Amico: And riding up a capability curve.
David Roberts: Right. So every bit they get cheaper, the capability curve goes up, the space of things they can do goes up, and the combination of the things they can do together now that they’re all cheap is sort of like, has grown exponentially, I think, faster than our brains and institutions can keep up right now.
So there’s, like, an explosion of things that electricity can do now that we are, like, groping around, trying to even figure out how large is that space. And we’re just, like, groping our way through that space now because it’s happened so quickly, like comparatively so quickly for all these things to get cheap together. There’s just been this combinatorial explosion of things they can do, and we’re just now groping, kind of checking the things off the list one at a time. But we keep discovering, like, oh, it could do this too, and this too, and this too. And one of the things it can do is AI.
So here’s one of the aspects of the essay I thought was really interesting. So one of the things you argue in this essay is that China is taking a different path than we are, has made different choices than we are in regard to the electric stack. You argue that it is trying to grab hold of the manufacturing and the expertise in making these substrates, these pieces of the electric stack, while it’s trying to commoditize AI, whereas we’re just going after AI and trying to make that our domain, and you think China’s got the smarter strategy there. I’m just curious what you mean by commoditizing AI and what are the implications of that?
Sam D’Amico: Yeah, so that’s a great question. So basically, what China’s been doing is basically, I think you can almost imagine it as like you ever played the video game Civilization or something like that where you, like, imagine, like, the technology tree. I think this is also how Elon thinks of it as well, to be very clear.
So I would lump, like, the China and Elon thinking in, like, one zone. I would lump, like, let’s call it, like, American finance minds in the other one. My stereotype of the US view is basically that, like, okay, figure out how to make the money, and then you can kind of demand pull everything through, which is generally a good strategy from a, you know, “Hey, you wanna make your business work,” right?
But if you think about it from the building blocks sorta layer, it’s like, okay, well, how do I do that? I need to have Lego blocks A, B, C, and D. I need to have my batteries, power electronics, embedded compute, electromagnetics, whatever. I need to have those building blocks. And I would argue that, like, China and Elon have specifically identified that they need… like, the bottlenecks will be their ability to have those Legos. And so making sure that you’ve got a complete Lego set is extremely important.
And then the other piece of this is like, well, AI is going to do this thing called, like, FOOM, AI doomers, or like, whatever it’s called, like FOOM, where it’s effectively AI will get to the point of self-improvement. But also, because of, like, Jevons paradox and all this other stuff, like, we’re gonna be using a lot of AI in the future. Like, it’s gonna be kind of unreal and a little bit, you know, beyond, I think, what people are expecting. And so demand for all the technologies that AI runs on is also going to explode.
Then if you wanna actually build things in the real world, like, you’re gonna be starved by the fact you’re competing with all of these AI demand, basically. And so I think the opportunity that America’s in danger of missing is that we’ve created this demand engine with frontier AI, and we’re not necessarily steering all of it onshore. If we had decided, like, hey, if we’re looking at, like, tens to hundred-plus gigawatts of new data center development over the next couple years, which is what I’ve been hearing, well, that’s a lot of hardware to be produced.
It’s a lot of hardware that is kind of like electric stack shaped, actually. And it would be pretty compelling if we use that to kind of reindustrialize America, build a lot of factories, do all this other stuff. That hasn’t exactly been happening yet. And, you know, there’s just tons of domestic companies making money on the AI boom, but you’ll notice that the tariff treatment of AI hardware is, you know, not less than, you know, what a lot of other things are. We’re intentionally not tariffing it to try to not slow that out, slow that down.
But what that also means is that we’re not necessarily onshoring production of all these things. That’s something that, like, I’ve definitely noticed and realized that there’s a gap here. Now, going back to Elon, because I think this is actually really interesting, is like, if you look, like, SpaceX is not considered the top frontier lab in AI.
David Roberts: Mm.
Sam D’Amico: People are preferring Claude and Codex to Grok in a bunch of cases, for now. But he is in a very enviable position in the space because he’s probably one of the few people in America who knows how to still build stuff.
And so they’ve been very successful at getting data centers plugged into power and being able to build these, like, large-scale coherent clusters, and that puts him in a position that he actually ends up having leverage over Anthropic, over, you know, some of these other folks, despite the fact that his models aren’t necessarily as good yet. And you could argue that this looks awfully similar to what’s happening in the robotic space with folks overseas in China, where it’s like, okay, well, you could build all the robots, you could be training the frontier models for robot control, like Pi or Skild or these other companies, but who’s building your motors and actuators?
And so I think there’s this really interesting thing of like maybe there’s gonna be a bimodal split in the winners of the AI race, which is, like, the folks that kind of control the electric stack technologies, and then, like, maybe the AI labs.
David Roberts: Clarify for me, though, like, I think maybe talking about this in the context of AI is gonna be driving a lot of listeners crazy, so let’s just talk about electric cars instead. What would be wrong if we want to get the benefits of electric vehicles, which relate to pollution, less maintenance, et cetera, et cetera, and we just bought them from China, which is very good at making them cheaply? Or, you know, same thing, like heat pumps.
We’re electrifying industry, electric furnaces, arc furnaces, or whatever, like all these... If China can manufacture enough of them for cheap, what is wrong with us simply buying the pieces from China and rearranging them in clever new ways? Our innovation is, you know, assembling new ideas out of these Lego blocks, and we just leave the manufacturing of the Lego blocks to them. What is wrong with that? What is the danger there?
Sam D’Amico: Yeah, so this is something where you can see recent FCC and I think commerce actions, the reception on this has been kind of bipartisan in some sense, which is, has been very interesting given, you know, kind of our current political climate.
But basically, the FCC banned new foreign-source power inverters. And you could argue that, okay, so let me define what an inverter is and all this other stuff, basically something that converts DC to AC so it can, like, you can send power into the grid. And if you go back earlier in the episode, it’s like I mentioned the grid as kind of a network. And then obviously these things run software on them, firmware. Okay, so you’re allowing foreign network devices to connect to our network, is kind of like the simple way to put this.
David Roberts: Yeah.
Sam D’Amico: And so it’s like, do you wanna do that without, you know, some sorta checks? Likewise, for any sort of large loads in the grid, it’s like, okay, well, if you give other countries software control over those devices, like, there are cybersecurity risks, right? Like, this ends up being kind of the same as the Cisco and Huawei disputes of, like, 15-plus years ago where, like, China kind of pushed Cisco out of their systems, US pushed Huawei out of their systems. And it’s like, in some sense, there is this sovereignty aspect of it. Like, do you actually control the grid or the network that you’re connected to?
David Roberts: Okay, but that’s an argument for then maybe manufacturing our own power electronics and the software controls. But, like, rare elements, you know, motors, those are not networked in any meaningful sense. Those are just kind of dumb, I guess, pieces of the puzzle. Is there a reason not to just let China do those?
Sam D’Amico: Yeah, so I mean, I think that the argument then is what happens if there is some sort of national security issue, and then you end up in this position where you don’t have the production capacity available to respond to a crisis because, like, okay, we could make a decision that, you know, uses up all of our interceptor missiles, for instance.
David Roberts: We would never do something so dumb, Sam, come on.
Sam D’Amico: We would never do that. And now there’s, like, rush to, you know, fix these gaping holes in our defense supply chains to produce, you know, be able to produce, like, that, basically restore our ability to produce things at volume. And so I think that us having a baseline level of capability is important, but there’s another reason why I think this is important, and it actually comes down to the innovation cycle.
So if you go to southern China, like Shenzhen, stuff like that, you go visit places like BYD or companies like that, one thing you’ll notice is, like, the engineers are pretty close to the factory. So, like, the ability for them to iterate is downstream of the fact that, well, the decision chain and the ability to, like, speak to the folks doing the production, that loop is really close.
David Roberts: Yeah. Let me insert here ’cause you’re teeing up my next question, which is one of the other arguments you make in the paper is that there’s enormous benefits to having design and manufacturing close together, like you’re saying, in a closed loop together. But, like, no American company has that. Like, Apple computers are famously designed in California and manufactured in China. Everything’s manufactured, your own company, Impulse, designed the stove and it’s manufactured in China. So, like, if that’s an error, then that’s a really fundamental error that the US is very, very far down the road of.
Sam D’Amico: I would argue that… So Apple actually is interesting here. So Apple addressed this sort of concern of co-location with a clever solution that a large megacorp could solve, which is they pay for business-class flights for their, you know, early-career engineers to just basically post up at the factory until stuff is done. And-
David Roberts: I guess you can close the loop by throwing a bunch of money at international flights.
Sam D’Amico: Yeah. They are, like, the number one United customer, if I remember correctly. I think there was some signage at some conference that was like, it was, like, $100 million a year in business-class flights on just United. They obviously spend on the other airlines too. But basically, closing that loop is super important. I would push back on there being no American companies that have done this. There’s a couple, and they all seem to be run by one dude.
David Roberts: This is the vertical integration that everybody talks about, that Elon’s so big on, is basically that, is basically bringing manufacturing and design under a roof so that they can iterate with one another, they can communicate back and forth faster.
Sam D’Amico: Yeah. And I would also argue that, like, there’s a lot of people who, like, speak the cult of, like, SpaceX vertical integration, but the real story is that the supply base just isn’t really there for the volumes and demands that they need. So it’s like there’s no satellite supply base that could make stuff for 10,000 satellites a year. Like, that was never a thing that anyone in the defense industrial base was doing.
David Roberts: Mm.
Sam D’Amico: Because of that, it’s like, okay, if you’re doing things that have never been done before, and your problem is actually like, they could have probably gone to Asian contract manufacturers and been like, “Hey, Foxconn, help us build some satellites.” The issue is they couldn’t because of ITAR, so they had to do it in the United States. International Traffic in Arms Regulations. So basically it’s like, you know, satellites are, you know, ballistic missile-adjacent, et cetera, et cetera.
But basically it’s like they had to do it in the US because of those regulatory barriers. What did Tesla do for a lot of their power systems and computer systems like that? Well, they work with Asian contract manufacturers for a good chunk of their system. They’re not fully vertical on every printed circuit board in the Model 3, for instance. Like, they’re working with folks that know what they’re doing and stuff like that.
And so that said, they now are at a volume that they can kinda pick and choose who they work with and all this other stuff. But the supply base, like Tesla’s not as vertical as you might think because they’re working within that ecosystem. And this kinda goes into, like, the big problem with the US on the innovation side. It’s like your ability to iterate in Shenzhen is just so much faster because it’s like three phone calls gets you, you know, 20 suppliers.
David Roberts: Yeah.
Sam D’Amico: And like things can be driven between... And like these are all the driving distance, and like prototyping goes way faster. I’ve been talking to founders who are like, “I’m just gonna spend two months in Shenzhen, just get the first prototype out.”
David Roberts: Did you go through this with Impulse?
Sam D’Amico: Yeah, yeah, yeah. Well, so we had a fun part where we started the company in 2021. You couldn’t really go to China until 2023, but I basically spent a huge amount of time getting stuff up and running, and now we’ve got, you know, building this kinda scale to arbitrary volumes. You know, mostly downstream, we’re just capital in terms of like pre-ordering parts and stuff like that. But we were a little bit early to kinda the second wave of this, but like if you’re some founder trying to build some, like, AI doodad or whatever, or a robot or whatever, it’s like this is clearly the path to getting you a prototype faster.
The difference too is because you’re prototyping in the production ecosystem, you’re kind of like what I would call like playing for keeps, where effectively if the decisions you make, if they’re, you know, made and applied to like a Shenzhen production line, it’s like, well, those will scale to your final product, versus if you go and prototype in the US with like McMaster-Carr, online hardware store, or whatever, it’s like, well, now you’re gonna have to remap that to a production process later. You’re gonna go through, you know, what certain people would describe as like manufacturing hell or whatever.
So this is the big gap is like the US lacks kind of this mature ecosystem that lets you kinda prototype anything. Meanwhile, Shenzhen is turning into this like what I would call like a matter compiler. Like it’s like, okay, cool, you can like almost like vibe code your hardware just by asking for it. Like we’re getting the Star Trek replicator, just happens to be a city.
David Roberts: Yeah.
Sam D’Amico: But then the frustrating bit, and this goes back to the electric stack, it’s like there’s only four things you need to be good at, so why don’t we just get good at those four things?
David Roberts: This gets to my question. Let’s cut to the chase then, because like- standing up a manufacturing ecosystem of the scale and skill that China has built honestly strikes me, and I think a lot of people, as basically fantastical. We’re pretty far down the road of being a service economy. A lot of things would have to change for us to get that, for us to truly manufacture our own stuff on the scale that China does.
So like, clearly the people who are saying this kind of thing, who are advocating what you’re advocating, are advocating for something short of that. So like what is the right percentage, right? Like, how much of our own stuff do we need to be producing, or what kinds of our own stuff? You can’t just say we need to become like China, because that’s the work of, like, a half-century. You know what I mean? And we’re in a hurry on a lot of fronts here. So like, what do we mean concretely when we’re saying this?
Sam D’Amico: I think the reality is it’s like, it’s not that many discrete things we have to get good at, and we’re honestly still pretty good at most of these things. Like, the automotive industry provides us enough backstop scale. Like, our existing automotive industry provides enough existing backstop scale that, like, we have some of these capabilities in-house, like still.
But a lot of it’s just like, can you assemble printed circuit boards with automated equipment, and then can you put them in boxes? If you wanna stick, if you wanna be really simple, it’s like that’s not rocket science. Like, that’s relatively straight. The big issue with all this is that the customer profile of someone who goes through domestic manufacturer, until very recently, was, “I am a fancy medical device company-
David Roberts: Right.
Sam D’Amico: and I’m building, like, an artificial heart, and it can cost a gazillion dollars,” or, “I am Raytheon’s sub-sub-subcontractor and I don’t care if this bolt costs a billion dollars.”
David Roberts: Yeah.
Sam D’Amico: Or it’s an automotive player that is like, you know, building something from the last century. And so I think that we ended up kind of pigeonholing ourselves, where it’s like no one was really doing the things that were going to inflect. And if they were, they were building exquisite systems for the military or for medical or whatever. It’s like they weren’t sold at enough scale to ensure that the pricing was competitive and all this other stuff. And so I think you’re asking me, like, what is the 80/20 of, like, US manufacturing?
David Roberts: And not just that, but like do we need also to be digging the minerals out of the ground? Do we need to be doing initial processing of the minerals into materials that go into these pieces of the electric stack? Like, how much of the... China has, like, so comprehensively grabbed ahold of all this. What bits of it do we need, or how much of it do we need? Like, what is the-
Sam D’Amico: The reverse engineering I’ve been doing is like, how is Elon the only guy who seems to be able to do this? And if you kinda look at, like, what’s crazy about this is like there’s all these technologies that were invented in California, for instance. And if I was like, “Hey, I just wanna make a factory to make this,” it wouldn’t be more expensive. It would actually just be illegal for me to do it.
And so the examples are like if you wanted to make printed circuit boards in California, I believe you have to deal with California Air Resource Board and your local water agencies and stuff like that because, like, you know, you’re plating, like, gold on- You’re doing electroplating and doing all this other stuff. And like, basically doing chemistry with metals has become largely illegal.
And so kind of the trade we made in the ’90s was like, as we were kind of deindustrializing from the fall of the Soviet Union, and then in the post-2008 recession when, you know, we didn’t really resurrect a lot of our US manufacturing base except for the auto industry, was like, “Okay, yeah, you can just do this overseas.” And like, we didn’t realize that we had made it impossible to reactivate these facilities because of, you know, environmental law and all this other stuff, because we kind of went into a like, it didn’t matter because you do it overseas.
Now, I think this is kind of ironic because it’s like, okay, we all like solar panels, we like batteries, we like all this other stuff, but like it’s kind of hypocritical if we can’t, you know, manufacture these things in places like California that ostensibly are like the biggest fans of these technologies.
And so I do think there’s gonna need to be a little bit of a reckoning on the environmental side for like just allowing, basically legalizing green jobs is probably the way that I would phrase it. It’s like there’s an opportunity for tons of blue-collar, quote-unquote, “green jobs” that frankly are not currently allowed in places like California, close to where the engineers that develop these technologies are. And so that’s a big piece of it as well.
But like going back to what would I say to prioritize is like, I think if you look at what’s missing in our industrial stack, it’s like… Frankly, it’s like we just need capability to build the replacement-level stuff, which is kind of ironic because the US likes to do the thing of like, this is the doomer version of it. It’s like we’re kind of like Germany entering World War II, where we had all the wunderwaffe, but we didn’t have the volume production of all the stuff that actually matter. Versus America entering World War II, where we could like turn washing machine factories into munitions factories.
Yeah, so not to kind of bemoan our like late-empire situation, but basically the point though is like being able to just have baseline production of the basic stuff, like, okay, can you go get drone motors in the US? Now, is the volume of drone motors from the entire Ukraine conflict enough to stand up a notable industrial base there? Honestly, maybe not, which is actually kinda crazy.
So then it’s like, okay, well, you probably need to have a domestic robot actuator business with, you know, drone motors for defense purposes being some sort of side effect, basically. And so you kind of end up navigating to like, I think ultimately you need to anchor with some sort of consumer demand, not just like the national security demand. The consumer demand is the industrial policy we need to be, you know, architecting.
But then to make that work and make it cost-effective, we need to do all the things like reform, you know, environmental law, permitting, make it offer cheap financing for factories, get states to compete off each other. It’d be like, “Hey, I’ll give you, you know, your tax abatements, incentives for workforce development,” et cetera. Like, there’s a lot of states that are actually starting to do the, do really clever things with this, but we need to start getting into that game too because that’s ultimately how China works.
Like, China’s ultimate thing is that they have like competition between all the provinces. The provinces, you know, step up and give people extremely cheap financing, like really good access to land, power, you know, facilities, et cetera. And they understand kind of this industrial policy finance ecosystem that like I think America’s starting to get, but it’s like that’s a key piece of it. We used to have this thing. They used to be called local banks.
David Roberts: Well, the, China also has central planning that decides on the overall industrial direction of the country. Then they have furious competition at the local level or within sectors, you know, and between provinces. But they also get the benefit of just deciding we’re moving in this direction. Let’s all row, you know, we’re all rowing in this direction, and then everybody’s competing. You get the benefits of competition down there at the sector level, but you don’t have this sort of either total lack of planning or flailing back and forth just about the macro direction of the whole country.
Sam D’Amico: Yeah, and I would argue that there is, the most interesting thing about the current moment we’re in is, I was joking with this on a panel at a conference like a year or two ago, but basically it’s like… You know, the tanks are gonna go woke even because basically, you know, the electric stack is gonna absorb even the most, like, masculine of combustion technologies at a certain point. That stuff, sort of stuff’s gonna happen. Very funny about this is like, I don’t think anyone in current administration, you know, in, disagrees with that statement actually. I just think that they might message it differently.
So there might be more alignment here than you would expect, you know, at the street level, if that makes any sense. I think that there’s definitely a kind of like, okay, what are we gonna do about the US auto industry and the EV transition and all this other stuff? And I think there’s gonna be big macro strategic disagreements. Are we gonna have a domestic battery industry that is, you know, underwritten by some sort of consumer demand? Yes.
David Roberts: I mean, what they’re doing now on EVs is the opposite of what you would do if you took the imperative to develop your own electric stack, et cetera, seriously. Do you not agree? Like, it is the opposite of what you would want to do. They’re protecting combustion long after its shelf life.
Sam D’Amico: Yeah, yeah, yeah. And then I would argue that, and this goes back to what you were saying, is, like, should we allow BYD and, like, Geely and Xiaomi to sell their cars in the United States? I’m like, I think the answer is yes, but I think if the answer is yes, and it’s kind of in this, like... It would just show how behind we are. You know, it’s kind of like, it would be like a Sp-
David Roberts: It would be embarrassing for many years.
Sam D’Amico: It would be a Sputnik moment, and so I do think that, like, we need to step it up and be competitive here, and that might mean that, okay, the Big Three need to, you know, need to reboot. There’s a lot of interlocking issues for why we’re in this spot. I would argue that, like, Jim Farley at Ford, like, they understand it deeply. It’s just that there’s a lot of structural issues that make this tricky. I think he does drive a Xiaomi SU7, by the way. Just to sh-
David Roberts: He was very impressed, very verbally, vocally impressed when he went and visited. We’re almost out of time. I want to get back to the stove for a minute. A little bit, a couple of stove questions. So this is your little piece of the electrotech revolution is cooktops. Couple of questions about cooktops, which is why, you know, the battery to increase performance, that’s a straightforward argument. Like, you’re gonna get really high levels of power at your burners, and so you can do certain cooking things you can’t do with a normal cooktop.
But that’s limited and a high-end, definitionally limited and high-end thing. Part of what you’re trying to do is get batteries in and then control the batteries. So a couple of questions about this. One is, why should you control the batteries rather than the utility controlling the batteries? Two, why a bunch of little batteries in appliances rather than one large home battery and a very good control software? Like, why the benefits of the batteries beyond performance? Talk a little bit about those.
Sam D’Amico: Yeah. So I think the performance angle was basically like, wanna convince people like, “Hey, you should buy this new product from a startup.” And like, the general attitude was like, “Okay, we have to just make sure this is the singular best stove on market.” Like, it was kind of, that kind of the argument. Because also, like, you’re not gonna win on price.
You don’t have pricing power and volume and all this other stuff when you’re starting, so you’re like, “Okay, you gotta do it that way.” The second thing was like, okay, so we put a inverter in the product. We haven’t turned it on yet, but we’ve got the capability to push power out of the battery into the house, and obviously there’s a bunch of changes with balcony solar laws and all this other stuff to make permitting that easier.
David Roberts: But to be clear, you can’t plug into anything into the stove, and the stove is not involved in any kind of demand response or anything like that yet.
Sam D’Amico: Yeah, yeah. We haven’t turned these features on. We’re piloting stuff like this, but basically the, there’s a bidirectional inverter in the product. It’s code compliant to like UL 1741, so it’s just like a solar inverter or any of this other stuff. One caveat here is like with a central home battery, you have this problem where it’s like you gotta either back up the whole house or you gotta back up part of the house, and then you have to like... At least that’s the killer consumer value prop. We don’t necessarily have that. The value prop for us is the stove’s really good, and it also backs itself up.
So like there’s no need to do the like partial home backup thing that like if you install a single Powerwall, that’s what you would probably do, versus like installing two of them and you can back your whole house up. That’s actually what like Base Power, for instance, has been solved, where it’s like-
David Roberts: Yeah.
Sam D’Amico: just make the battery so big that-
David Roberts: Yeah, yeah...
Sam D’Amico: you can back the entire house up, and then you don’t have to deal with this like weird half-assed installation situation that is like, adds like 50% more complexity on the electrician side, or if not more, 50%, way more. And so we were like, “Oh, well, the battery’s gonna be smaller than that by nature. It’s gotta fit inside a device.” And then it’s like, okay, so you just need to do grid-tie. You don’t need to do grid forming.
And then you can still do all the virtual power plant stuff, but you take backing up the house off the table, which makes the installation way, way more straightforward. So then going to like the second piece of it, which is why many small batteries versus one big battery. Well, it’s like I think there’s a role for batteries at kind of every layer of the network.
If you think of the grid as a network, going back to what I was saying earlier, there’s packet buffers at every layer, like down to the router in your house. And so you end up doing this in a couple different layers, and the reason this would make sense is like, okay, let’s imagine I have a central home battery, like got a Powerwall. It’s only got so much peak power out of the inverter. I don’t know, it’s like somewhere in the 7 to 10 kilowatts. I don’t know, I forget the exact number.
But like your house could draw more than that. And now if you could basically say, “Okay, cool, my major appliances all draw way less,” well, now that central home inverter doesn’t have to handle induction cooking, doesn’t have to handle all these things. And so again, makes the installation and distribution problem significantly less if you kind of scatter batteries everywhere.
Then the next piece is you gotta then kind of reaggregate them into something that is like legible to the utility. Ultimately, the utility is the one who actually gets the value out of the batteries that are distributed on the grid, because it’s like they’re the one who’s saving money by not having to upgrade transmission, because they can then harvest the battery storage. And so ultimately, that’s the person who gets the most value here. And then it’s all these other things are just structures to kind of share that value with the people who deployed the batteries, basically.
David Roberts: What’s the roadmap? Right now it’s a $7,000 countertop cooker, which is obviously going to be of, you know, limited sale. That’s a limited market. Is the idea to bring down the cost of the Impulse cooktop until it becomes a mass-market product? Or is the idea to take the basic core technology and power electronics that are making it work and embed those in other countertops that are cheaper? Like, what is the product roadmap for the Impulse cooktop specifically?
Sam D’Amico: Yeah, so we’ve got several partners in the appliance industry that are coming out with products at various price points, and these are both cooktops and ranges, so, like, basically with the oven integrated-
David Roberts: Mm-hmm.
Sam D’Amico: and other products like that. But basically we take the underlying electronics, largely the same stack. We can tune the performance in software depending on what the customer’s price point and all this other stuff is. There’s other things we can do in hardware to, like, defeature, you know, upfeature, et cetera. But this allows us to go from, like, hey, we’re have, like, a 30- and 36-inch cooktop available to a bunch of different SKUs in a bunch of different sizes and shapes. We’ve announced some of them already. We’re gonna announce more in the coming-
David Roberts: These are through other bigger manufacturers, other bigger appliance-
Sam D’Amico: Other bigger manufacturers. Now, we’re still making all the electronics ourselves, so it’s like that’s coming from our supply chain. Then it’s kind of like the goal is to set a new standard where it’s like, okay, well, one of the key features with our product is, like, you can set and hold an exact temperature, and you can set and hold an exact temperature inside the pan through some technology we’ve developed.
David Roberts: Yes. I think you tried to explain to me exactly how that works once at a demo, and I’m still, I’m st- I still confess to be totally baffled by how that works.
Sam D’Amico: Yeah. Effectively, if you control the temperature sensor and you control the induction, you can kind of measure the temperature through the pan versus just kind of, like, seeing some sort of second-order, like, the bottom of the pan is hot or whatever signal.
David Roberts: Wild.
Sam D’Amico: But that sort of stuff, super exciting for people. The fact that we’ve got, like, a screen, can put a bunch of content and stuff on there has been super exciting for some of these partners as well. And so, we’re kind of being like, “Hey, this is the new standard for where induction and kind of power systems for appliances are going, period.”
David Roberts: So you wanna be like the Intel Inside of the induction stove world?
Sam D’Amico: Yeah, but it’s also like Intel plus Microsoft is probably the way to think about it, like in the ’90s, because like you’re doing the software and cloud platform and all this other stuff as well. That’s how we’re taking this like more mass market than just like, hey, one brand, one SKU, or two SKUs, one industrial design, et cetera. Because people have opinions on all this stuff too. Like, people want some like La Cornue thing or, you know, whatever.
It’s like people have all sorts of preferences what they want for their kitchen, and so you end up in this pickle where it’s like the appliance industry is actually closer to like the car industry in a sense, where there’s like dealerships and like it is kind of the dealer model and like multi-step distribution and all this other stuff. So this allows us to kind of do the compelling thing right at the node where they’re, the most interested people are to buy it, which is like the appliance base or the appliance manufacturer specifically. That’s been a lot of my focus on that end, and then it’s also just like what other categories of devices to kind of go into and innovate there next.
David Roberts: That’s my next question is, you know, we’re talking about these medium-sized electricity, electric devices where, you know, like a lot of the smaller ones have been perfected, a lot of the bigger ones require a lot more scale than you have to get into, but like an individual company can get a toehold in this medium-sized electric appliance world, what else is in that world and what’s next on your list?
Sam D’Amico: I’m holding a little, a bit of this a little bit close, but basically it’s like there’s a lot of stuff in the home that’s really interesting. There’s stuff in cooking that’s interesting that, like, I would love to do a microwave, for instance, but that’s probably a little bit later.
David Roberts: Ha ha ha.
Sam D’Amico: The other stuff is, like, there’s industrial equipment is interesting because it’s like, okay, if you can buffer power to, let’s say, like a CNC machine or something like that, like now you can install more CNC machines in a building.
David Roberts: What are CNC machines?
Sam D’Amico: Like computer numeric control, like you have cutting metal. The machines that cut metal, like it’s the thing that machines like your MacBook’s body or something like that. But basically it’s like, you know, people are trying to reindustrialize America. There’s actually a pretty interesting opportunity in terms of just like power buffering industrial equipment, so that way, you know, again, it seems like everything, if you kind of look closely at power traces from pretty much any machine, they’re like not on all the time, and then when they’re on, they’re not at full power all the time. And you can just look at that across huge numbers of sectors, and there’s really interesting stuff to do there.
And so one of the things we’ve been doing is just like figuring out kind of like what the minimum viable like core module is for that. And then, okay, that becomes the thing that we can kind of like repeatedly sell across a bunch of these spaces. And then it’s got all the same software and control systems, and if you want the temperature sensor for some other use case, you could use it. You know, it’s like it all ends up being this kind of modular platform. That’s been kind of the way we think about it.
David Roberts: And the modules basically are like battery for buffer and power electronics for fine-grain control. You’re just applying that to machine after machine.
Sam D’Amico: After machine after machine. And then the thought here is like, how do you connect to the grid? Well, we got this grid-tie inverter. You now have that. You can do that in that sort of modular way as well.
David Roberts: All right. Should probably leave it there. Actually, I have one additional question, which I may or may not cut if you have like a minute or two left, which is semi-unrelated to the other questions. You’re kind of in the Silicon Valley founder world and community, which, you know, the rest of us, whether we like it or not, have to be aware of-
Sam D’Amico: You can’t escape, yeah.
David Roberts: what they think, what they’re doing, what their sort of macro thoughts are matter to the rest of us, whether we like it or not. And I’m just sort of curious about, I’ll tell you my external impression is that these people, through some mix of like AI sci-fi scenarios and effective altruism arguments about the distant future and et cetera, et cetera, have basically talked themselves out of caring about climate change much. Is that accurate?
Sam D’Amico: No.
David Roberts: You don’t think so?
Sam D’Amico: I think there’s a couple things going on that are, like, interesting here. So, like, I think what people have realized is that there is a kind of like AI will solve all problems sort of EA-adjacent, you know, thing-
David Roberts: Yeah...
Sam D’Amico: that I think is totally happening. But I have to be a little bit quiet about this, but basically it’s like, I think that there is relative conviction that geoengineering will work among a certain sector of folks, and it’s probably some sort of like stratosphere aerosol, stratosphere particulate injection situation.
And I’m reasonably convinced that we’re gonna have mitigations that look in that shape, or we’re gonna have to do those mitigations because again, we’re gonna get so far into trouble in the 2040s, 2050s that like, it’s gonna be inevitable that we’re gonna have to do that sort of stuff. I think the bigger aspect of this is, the bigger story is that the language has changed, but I’m not sure if the mission has changed. Does it make some sense?
David Roberts: Hmm.
Sam D’Amico: And when I say that, it’s like, well, if we’re reindustrializing America, we’re building, you know, stuff with all these new technologies, what are those new technologies again? Oh, it’s the same tech stack that we anointed as the saviors for climate change, basically. It’s the same stuff. It’s like, okay, well, how are we gonna solve climate change originally? It’s like batteries, power electronics, embedded compute, electromagnetics, rebuild every machine on that stack. Now everything is suddenly electric.
Oh, and then we need to make sure that we have clean firm power. Okay, well, building lots of batteries. Every data center is getting gigawatt hours of batteries attached to it or whatever. And then what’s the cheapest source of electricity? Well, it’s probably solar out in the desert in South Texas. And so it seems like we’re doing all the things that we were supposed to be doing. We’re just talking about it differently. I’m not sure if that’s reassuring.
David Roberts: It looks to me like a lot of them are perfectly happy to power AI on gas, and in fact are embracing that affirmatively and less and less talking about that as a bridge, as far as I can tell. Like, so that would, it seemed to indicate to me that they care a lot more about AI than they do about climate change.
Sam D’Amico: I do think that the thing is real is that, like, I think people are still underestimating how transformative AI is gonna be. Someone said, like, 2% of Americans have a paid ChatGPT subscription or something like that. Like, this is maybe six months ago. But the fact that that subscription rate is so low, which means that everyone’s just talking to some, like, low-end model and not talking to, like, the stuff that’s actually at the frontier, means that I think most people’s perception of this stuff is fundamentally inaccurate in terms of how transformative that’s made.
And so one thing to really put a point, to put a fine point on it is, like, I’m definitely in, like, this, like, you know, low tens of thousands of people, and I’m not even in one of these frontier labs. I’m not seeing what’s quietly happening at OpenAI or Anthropic or whatever. And it’s like, what’s actually happening there, to my understanding, is, like, people are, like, throwing thousands of agents at complex problems and walking away with solutions to, like, math proofs and stuff like that. It’s starting to get pretty spooky and, like, outside of the domain of, like, what humans can comprehend of the capabilities.
So we’re at the point then you have that group, then you have people like me who are, like, deep in the sauce with this stuff, and then you’ve got, like, typical software engineers that are getting dramatically more productive with this technology, and then you’ve got kind of people on the outside of this being like, “What is going on? This is wacky.” I think we’re in the biggest inflection point in the technology industry ever, basically.
Maybe since, like, the dawn of the invention of the transistor and the integrated circuit might have been, like, about as big. This could be the big one, basically. And I strongly suspect that I even think that I’m not taking this as seriously as I should be based on what I’m seeing in all this other stuff.
David Roberts: But what is the implication of that for climate change? Spell it out. What are we supposed to take from that about this gathering problem of climate change?
Sam D’Amico: So complex engineering work is going to be increasingly more tractable by machine intelligence over the next two to three years. The transformation of the economy with AI is gonna basically buy, you know, more batteries and solar than I think ever were bought before. Now, yes, they’re gonna buy gas turbines, but gas turbines are fundamentally bottlenecked by hard materials problems, whereas solar is not bottlenecked, right? And we’re not gonna go give China gas turbine technology, you know, to scale it. Like, that’s not gonna happen. That’s not gonna happen.
And so basically, I think what’s gonna effectively happen is, like, the economy will increasingly just electrify. Cars will go electric because of autonomy, because that’s, again, like, you’ll want the more controllable electric-native vehicle. Like, if you look at Tesla Cybercab, the brakes are electric. It’s like it has no hydraulic system anymore. I mean, you could argue that Starship is electric, an electric vehicle, just happens to have, you know, primary propulsion is methane. But, like, other than that, it’s, like, all of the actuators and all this other stuff, it’s an electric vehicle.
And so I think what’s gonna happen is we’re gonna see a slow or a very rapid transformation of, like, everything becoming kind of like a robot, and thus to be a robot, it’s gonna have to be electric. Then the last question’s gonna basically be like, “Okay, well, what do we do about the fact that, you know, Earth is several degrees Celsius warmer than it should be?” And I think the answer there is probably geoengineering. This is my guess. And are there people working on this right now? The answer is yes, but I also suspect that that’s gonna be the spicy topic in the next, you know, 10 years or so.
David Roberts: Well, everything you said could just spin off into a pod of its own. But I probably shouldn’t have cracked that can open at the very end of our pod.
Sam D’Amico: The point is, we’re gonna have to deal with climate change, but the tools we’re gonna have to deal with it are gonna basically be the entire economy is going to be electric within the next 10 years, period, driven entirely by forces, entirely by demand pull, basically, on robotics and data centers and all this other stuff. And then it’s like, okay, how do we clean up the mess we made of the past century? And it’s like, it’s genuinely a good question. Do I think people are not caring? It’s like, no, I just think that what’s happening, it’s almost like climate change was the concern and now there is a bigger concern, if that makes any sense.
David Roberts: All right. Well, I’m gonna resist the temptation to pull on all those strings and end it here.
Sam D’Amico: That’s another pod for another day.
David Roberts: Thank you for talking this through with me. It’s always, as you know, I love talking about electricity, so thanks for indulging me. All right.
Sam D’Amico: Cool.
David Roberts: All right, see ya.
