Gradient Dissent: Conversations on AI - Elon_s Former Battery Chief_ AI Data Centers Will Make Electricity Cheaper

0:00Drew BaglinoThat's actually what a data center is. You take a gigawatt of power and you convert it into 700 megawatts of heat. You spend 300 megawatts to get that heat into the atmosphere. I guess presumably your transformer is going to be more efficient than the 10% loss. We can reduce the loss from grid to chip by about a factor of two. Our core first product, the Heron Link, the transformer part is 100 times smaller. You still need to go through an isolation stage when you're going from like a high voltage grid to like a computer. So you need to galvanically isolate.

0:29Drew BaglinoBut rather than doing it at 60 hertz, which is how we do it today, we can do it at hundreds of kilohertz because these devices I'm talking about can switch so fast. So it's a material science thing. You're trying to find materials that can block more voltage, are more thermally conductive, because a semiconductor is either conducting or not. And there's a bajillion transistors in a GPU, and a power device has one. Right, right. But it's one highly engineered device. And what silicon carbide is doing is allowing you not to make a one volt transistor like you have in a GPU, but you can make a 10,000 volt transistor.

0:59Drew BaglinoIt is actually smaller than the GPU. So we have the loss. So for somebody building like a gigawatt data center, it means they can get 35 megawatts more useful computer. OK, so why did California power rates go up? There's like a lot of weird reasons. One is you're listening to Gradient Descent, a show about making machine learning work in the real world. And I'm your host, Lucas B. Wald.

1:22Lukas BiewaldToday I'm talking with Drew Baglino, an old friend of mine, who started as a junior engineer at Tesla and ended up running all of powertrain and energy, which is most of the company and reporting directly to Elon. He then went and started Herron Power, which has raised hundreds of millions of dollars to revolutionize the energy grid. This is a really interesting conversation on topics I know a lot less about than AI. I hope you enjoy it. I'm super excited to talk to you about your company.

1:50Drew BaglinoBut I haven't known you for a long time. And I kind of wanted to go back and start with like your career at Tesla because that was a really interesting ride. I think you were there for 18 years. Can you talk about that experience? Sure. Well, maybe I'll say first, yeah, we knew each other in college, went to Stanford together, maybe lived in some funny off-campus housing together, more than one of those off-campus houses, I think.

2:20Drew BaglinoAnd I remember even back then, you and I would talk about different ways in which software would change the world. And you were always better at the software classes than I was. But I took them, too. So I have a little bit of a software background. But yeah, that's actually kind of what a little bit of what brought me to Tesla was some connections back at Stanford. So I ended up in this.

2:50Drew Baglinosigning up for this energy field trips class that Gill Masters was running where it sounded like the coolest class ever like you were gonna go to you know zero emissions buildings and you were gonna go visit like wind turbine manufacturing facilities and solar manufacturing facilities and large installations like we went to the Hoover Dam and we went to some large wind farms and solar installations including like the the SEGs like parabolic trough mirror salt solar installations and The Mojave which are actually pretty cool to check out. They're still running today. I think I see him from the airplane sometimes on Yeah, not the power tower one that one's nuts, but now you do also see these they're they're they're on that like the 15 I think but I signed up for that class sounded really fun and that's where I met JB Strable who you know the one of the early Tesla founders And he and I just kept in touch and and even though after

3:49Drew BaglinoWe lived together in Dire Wolf. Was that what it was called? I don't remember what it was called, to be honest, but that house. Some grateful dead song. Yeah, yeah, that house on in Bargadero. I left that house and went to DC to work at this think tank called Resources for the Future. And I was basically doing programming there. I was like coding up this crazy model in some bizarre like database language that I can't remember.

4:14Drew Baglinomaybe it's called Stata or R, or maybe it was both of those languages, but... There's other bizarre languages, by the way. Yeah. Classic stats languages. Yeah, totally. I can't remember which one it was, but I was working on this really interesting economic model of the Washington DC metro area and bored kind of out of my mind, mostly about the impact of it. I was like, it was interesting work, but it was translating it to nothing that I could see.

4:43Drew Baglinoin the physical world, at least not on a time horizon that was interesting to me. And so I kept in touch with JB. And at some point it was like, you know, I think we might have a role for you. I had this background in embedded DSP stuff, because I took some graduate courses at Stanford in robotics and in communications that were in embedded DSPs. And he's like, we need somebody who knows embedded DSPs. I was like, oh, I can do that. And that's how I ended up there.

5:10Drew BaglinoBecause in the first project was, hey, we licensed all this analog control design for the motor control and for the charge control and for the battery management. We licensed it all from AC propulsion. That's how Tesla got a lot of its early technology was through licenses from AC propulsion, including the motor design. And we want to replace all that analog stuff that's controlling the car. It doesn't work well.

5:40Drew Baglinowe want to do it in a DSP. And I was like, sweet, that sounds like an amazing project. So that was really the first project that I was responsible for from beginning to end at Tesla. And it meant I spent a lot of time in electric cars driving around these streets, roadsters parking in the Mission District. And also there was a really fun project involving converting a An internal combustion engine smart car that we bought in Mexico to an electric smart car and I remember parking it on like 20 second and Bryant like at the Bryant's you remember that house I lived on on Brian Street totally So I would just park it on the sidewalk it like fit on the sidewalk Right next to the staircase You know didn't even like block the sidewalk. That's how small the smart car a smart car was and it was

6:38Drew BaglinoIt was great, like learning how to, you know, going from working on code and data to model like the economy and the transportation system of Washington DC to embedded software that immediately you could see the result of it in the vehicle that you were driving yourself. Like, and I'd be driving it and I'd have them, the laptop out like showing me the signals of what was going on. It couldn't be more real time feedback.

7:06Lukas BiewaldThat's sort of what got me hooked on on what we were doing at Tesla and what sort of why I stayed there so long was just seeing The work that you do translate into the real world so quickly Well, I think you also had this awesome experience of working really closely with Elon before it was clear What a successful like entrepreneur who's gonna become do you have any fun stories of of Elon back then? Yeah, I Can I can talk through a really early my first exposure to Elon. Yeah

7:35Drew BaglinoSo I work for JB the CTO directly throughout my early years and One of the things that I was responsible beyond like all this embedded software that made the cargo and things related to that Was modeling The like physics of our products so like how far could they go how much energy would they consume per mile? You know with the car overheat when it's towing you know, about, you know, all of these sort of like physical aspects of the electric powertrain and the systems attached to the electric powertrain, how quickly could it charge and things like this? And so in the period of time when we were, you know, conceptualizing what Model S could be our first, you know, let's say mass market EV, I mean, the Model 3 was truly the mass market EV, but Model S was way more mass market than the Roadster was. I was, I was sort of in that architectural role defining

8:35Drew BaglinoYou know if we want to make the car do this that means the battery needs to look like this That means the motor needs to look like this. This is what an optimization of that system would be and You know, we got some guidance from JB and others about like what kind of car we were trying to make and So we brought our presentation To you on of like well, this is what we think the battery needs to be like how big it needs to be how much energy needs to be in the battery And you know, we went through the assumptions like we think the car will weigh this, we think it'll be this slippery, we think we can get tires that are this efficient. You know, we know we want to hit this EPA range and you know, there were a couple other like dimensioning load cases and we kind of went through most of that with him and he was following most of it. And then we get to the conclusion.

9:29Drew BaglinoWe think we need to have somewhere between 8,000 and 8,800 cells. Ultimately, it's complicated math, but it isn't just math. You have some sensitivities on how good all of these numbers could be, and this is the range, 8,000 to 8,800 cells. This is the first time I've ever interacted with Elon, and he just says, it can't be more than 7,200. It's 7,200, and we're like, okay.

9:59Drew BaglinoUm, do you want to like go through any of the assumptions? Like, where do you think we're wrong? He's like, it can't be more than 7,200 and like meetings over, you know, it's like everybody is like, what just happened? You know, like, um, and, uh, and it was such for me, it was also like, well, did we do something wrong? Like, you know, we didn't really understand how to take it. And, um, looking backwards at it, you know, he had information that I didn't have access to, like, you know, how much.

10:30Drew BaglinoEnergy how much would the cells cost he was in the background pushing Panasonic our cell partner way out of their comfort zone on how much energy density they could get per cell I didn't know any of that But and I don't actually think he did either He just was like doing that math in his head of like oh my god If it is 8,000 cells like how do I fit that many cells like how much is it is the car still gonna be affordable accessible. He didn't communicate any of that, but actually that's part of, I would say, his genius is like, by putting out what seemed like impossible targets, like, yeah, you said it has to be 10 to 20% more than this number, but I can't be, I'm telling you, it can't be more than 7,200. You know, what it forces, and we did it was, okay, we got to make the motor more efficient. We got to figure out how to

11:28Drew BaglinoYou know make the car more slippery use less energy on the road cooling itself and and conditioning the cabin We do have to push Panasonic really hard to make the sell more energy dense and and sort of You're breaking everybody's back at the same time. I mean you're trying not to let's say you're loading everybody's back equally to sort of achieve this goal and and doing it in a way that that forces kind of like the A game out of everyone. So that's sort of like a classic example of I think how Elon operates. At the time I was like, I never want to be in a meeting with him again because it was such an intense interaction with not a lot of context. But when you sort of step back and appreciate sort of like why that's the way it's communicated and how that drives results, like it certainly drove the result we needed, which is like...

12:25Drew BaglinoEverything that I just said happened like we did get a more efficient power train and the battery energy density was Was much higher than Panasonic was initially proposing And it's almost good too that we did that because other things changed along the way like the way range was measured by the You know the EPA changed in a not so favorable way. So like you kind of needed everything to

12:53Lukas BiewaldTo go this way just to have a complying product anyways Like it's one thing that I've observed from the outside watching you and a lot of my Other friends or mutual friends kind of work for Elon as people seem to burn out pretty quickly for this reason I mean everyone's kind of got a story like that and most people seem to walk away from that story being like this is like ridiculous and I kind of told him the truth and he just didn't listen and then he fired me when when I said turn out to be the truth But I hear that all the time, but you

13:19Drew Baglinoseemed to kind of enjoy the experience and you managed to work from quite a long time and really rise up inside Tesla. Do you think there's something about you that's different than most people? You need to have a thick skin. You need to decouple your work product from your ego. Not everybody can do that. I think that's actually probably one of the hardest things for engineers, especially in high-paced companies to do, is to accept that What they just spent the last three months on might get torched and for really good reason You know may not as it may not initially be obvious to them. Why? But in retrospect as if they think back on it Whether they're at the company or not, they'll be like, oh, yeah, that was the right decision And I think that is just hard for engineers to do But you kind of need to be able to do that if you're gonna be nimble if you're gonna respond to the newest information if you're gonna

14:19Drew Baglinofind the failures in what you develop, whether in software or in hardware, you can't attach your ego to your work product. And I think that's something that, I mean, with JB's support, I mean, JB like helped me through that, right? And having a great manager early on, I had two amazing managers early on. One was JB Schraubel, the other was Craig Carlson, who was like, I would say, highly contributing.

14:48Drew BaglinoEnergy sorry engineering executive At Tesla from like 2008 or seven or something until 2013 and he ran software At a Tesla because I was responsible for a lot of this powertrain software like I worked for Craig for a while Both of those leaders really taught me how to How to how to like check your ego at the door basically if you're gonna if we're gonna do a good job in it in a like hardcore engineering organization I think that is hard for any young Engineer to do But if you can do it you can you can go far and I think that's true in business generally not just like in an Elon company Well, is it like when JB left so you kind of stepped into his role, right? I mean what what was that like?

15:41Drew BaglinoIt was fulfilling to be exposed to so many new areas of technology. I think that was really what like retained me at the company was all of these new problems. So yeah, when when JB left, I was responsible at the time for kind of this like powertrain architecture. So that's the physical modeling of all of the systems in the vehicle and how do you optimize them to deliver the best results? And so that was a team of like uh, 20 or so people that, uh, did what we would call a model versus actual. So you are both like responsible for predicting what the range of the vehicle will be or the water per mile will be or, or whether it'll overheat on a grade going up to Tahoe, but not just predicting it, but also then going out and running the tests. And if it doesn't match figuring out why and, and, and like using the gap between model and actual to make everything better, not the actual better and the model better. Um,

16:40Drew BaglinoSo that was that team. And then I was also leading this energy engineering organization. So the team that was building and designing and building and kind of installing the power wall and the power pack. And we were already starting to work on the next versions of that stuff at that point. And the superchargers also, we were responsible for engineering the superchargers. And so when JB left, I took on the battery pack like how our batteries made and the cell roadmap of that's inside those batteries so you know how do we improve the cell energy density the cell affordability and also took on the propulsion system so like the how do you actually like physically design and validate the motors and the gearboxes and and then also all the power electronics

17:34Drew Baglinothat are in all of these systems. So I took on all of these like new areas of technology that were really fun. And then I also started to play a more significant role in the business aspect of energy, like how do you make money with our energy products? Cause that was something that JB had also sort of been the executive steward of. And then the last thing I became was one of the officers of the company and joined all the board meetings and the earnings calls, which

18:02Lukas BiewaldI mean, it's just fascinating. It's fascinating. It was a story. It was a story if people might not have heard of that experience. I feel like you got this inside look at a really iconic company and a really interesting time. What the... I mean, yeah. It was always... Let me get some clicks here, Drew. Come on. Yeah, okay. Look, I don't think it will be a surprise to you at all that Elon leads a very busy life.

18:33Drew BaglinoWhat was always impressive to me was, you know, we are ahead of an earnings call. Like, you know, I work closely with Zach Kirkhorne. I work closely with the other leadership team, like Jerome and others. And we're like, we have a lot of anxiety about the, these calls and like prepare a lot, you know, and, and frequently like Elon would be coming from probably neural link or, you know, maybe, maybe even, uh, uh, open eye at the time, to be honest.

19:04Drew Baglinobecause he was heavily involved still with opening eye at the time. And then like, just show up with and maybe have like five minutes in the room to like chat with us. And then the recording button starts for the earnings call. And like, you know, it's soup. That's I don't know how he did it. To be honest, I don't know how he did it. I mean, that's just part of his genius is that he can do that. But it was certainly anxiety is high anxiety times for us.

19:33Drew BaglinoUm, and, and, you know, like, are we aligned going into this? Are we going to say the same things? You know, um, those were, those were difficult times. Um, but, but we, of course, maintained good alignment with, with Elon through other avenues, not just, you know, the, what comes up ahead of one of those earnings calls, trying to think if there's anything out like any specific event. Um, but I mean, you can imagine it's, it's not quite like a live recorded podcast in a studio like this. Hello.

20:04Drew BaglinoBut you know, you have the people that are going to move the stock listening on every word you say. And if you're not prepared and if you put your foot in your mouth, yeah, you're probably not going to be at the company the next day. So probably some of the more stressful moments in my life. Anything else in your experience with Sharon? Yeah, I've got some other good stories. I think there's been a lot made of production hell.

20:35Drew Baglinokind of ramping model three was very intense. Um, and there were so many reasons why, like the company was in a major growth period in so many facets of like how to operate itself, not just how to produce those that many vehicles, but like get the parts in, get them inequality. Once they've been produced, deliver them to customers. Once they're delivered to customers, keep them serviced and operating well, every aspect of the company to grow.

21:02Drew Baglinomassively in Model 3 or like orders of magnitude. But there's been a lot made of like the production hell part of it and I think this is another area where like a lot of people, as you would say, were like, oh, I said the truth and then I was fired. And there's always two sides to the story and I'm not here to defend Elon, I'm just here to provide like another layer of context to all of that.

21:34Drew BaglinoYou you have to step back and think about how are you how to run this company that is good trying to do like probably one of the hardest things any company will ever do you know go from delivering 10,000 to 20,000 cars a year maybe 50,000 that we were doing at that point I don't know maybe it was 100,000 model X's and S's that we did in that time frame per year I think it was to you know doing millions of cars a year And just when you think about the sheer tonnage of material, the millions of dollars and billions of dollars of work in progress, it's time. And any given interaction that any executive would have with anybody at that time, you as the executive, you're trying to drive the end outcome. And while maybe in an interaction on a production line,

22:32Drew Baglinoyou know An executive might say like just delete that thing or like we don't need that, you know Is that actually what they really mean or what they try to say is like we need to produce more units and so any interaction with any individual bottleneck like You don't want to over index on like oh well He said we shouldn't do that here. So we shouldn't do it anywhere. Well, actually what really the point was was how do we make more product like holistically And I don't want to, it's not like anybody wanted to make bad product. It's like you just wanted to make more product. And sometimes you need to make a point of like one thing that, and actually one thing that we, I'll actually get into details on this one. So we had this robot that was trying to install this like,

23:33Drew BaglinoI think it was like a moisture, it's kind of like a, it's like this flimsy bag filled with a desiccant or material or something. Try to avoid condensation building up between the battery pack top and the, and the, and like the underfloor of where the seats, you know, where your feet foot foot is. And it was like partially to, to maybe act as a desk and partially to be like a noise barrier.

24:01Drew Baglinobut it was this super flimsy object. And there's this robot trying to get it onto the top of the ferry pack. And I think Elon may have coined it this, or somebody else did, but we call it the fluffer bot or something, because it just looked like this ridiculous thing and was always erroring out. And when you step back and looked at this and you said, why two things? Why do we have this part?

24:28Drew BaglinoSecond why do we have like a robot installing it? It was like it had to be like taped down. It was really complicated and In like classic outcome like this is true of so many things in life Like first we got rid of this robot the line was stop because like the thing wouldn't get installed correctly right like That's like you don't need to stop the line like a person can fix it like There's space for a person to go and just like correct it, like don't stop the line. Then it's like, why are we even using robot at all? Like if you're going to have a person fixing the robot, just get rid of the robot. And then it's like, wait a minute, this part should just be deleted. Just get rid of this part. You know, and like you go through that process. I think in this case, we went through that process in a matter of like two or three days, you know, and just by taking that like one flufferbot or I can't remember the name of it. It was something like that. Somebody is going to respond.

25:22Drew Baglinoyou know comment on this and say this is what it was actually called just by like you know going through that example with the team like they find like seven other versions like that you know seven other other similar opportunities where like the park could be deleted the process or like the way we're automating it um and sometimes those interactions are really painful especially if like the person that is in that interaction with Elon is the person that spent the last like two weeks coding up this complicated robot and like defending their work. And then, you know, maybe not there the next day, but like then the broader team is like, Oh yeah, I get it. I get it. It's not really about the fluffer bot or the part. It's about this overall system. Like there's other examples where we're making decisions that if you viewed them from another lens, you could delete, delete, delete or simplify, automate, whatever.

26:14Drew BaglinoSo what about now that you're running a company yourself and you've seen this one example in CEO of this really iconic, inimitable CEO? What do you take from it and are there things that you leave behind? How do you find your own style? Yeah. Well, even working within Tesla, it was like, yeah, I learned a lot from Elon. I learned a lot from Zach. I learned a lot from Jerome. I learned a lot from Doug, Craig.

26:44Drew Baglinoyou know, I think I was developing my own like merger of all of these individuals that were my role models or influencers, you know, peers. Um, uh, what, what I value more than anything else in, in my organizations that I build is, is like humility and, and transparent communication. Um, and, uh, and sort of this like commitment to trying to fail fast. Um, This is another reason why ego attached to your work product is a bad thing. So if you attach a lot ego to your work product, you're more likely to take like the positive side of any finding. Oh, um, this test result, it's a little funny. I, it, you know, it's not like it passed or failed. It's funny. You know, it may, maybe we need to retest it. That's like, okay, I got to stop you here then. But yeah, but totally, but like the stories you've told Elon's been right.

27:45Drew BaglinoBut like, I guess my experience of Elon is mostly from Twitter, where this doesn't seem like a guy with zero ego, who's sort of dispassionately talking about Elon. I'm not talking about Elon. I see. Yeah, I'm not talking about Elon here. I'm talking about how I try to run my companies and why I reinforce the divorcement of ego from your work product. Yeah, Elon has, in his companies, Elon is the alpha, for sure. He's the, yeah, there's no question about That like he brings the so that's a different style I'm talking about what's important in for the people in my team that I asked from them and it and that's not that's not a question of like my style I'm not talking about my style necessarily I say I'm talking about what I Want to see in my you know team members And the reason the end I'm just providing another I guess analogy like

28:43Drew BaglinoIf you put a lot of ego into your work, you're not going to find the problems in your work. And if you want your product to get to market and be the best in the market, and you want to get into the market quickly, and you want to iterate quickly, you need to find the problems quickly. And that's where trying to detach yourself from your work product and being able to attack it as openly as anybody else could, as openly as Elon undoubtedly would.

29:12Drew Baglinoto bring Elon back to the conversation. I think that is key. And so that aspect of what I learned at Tesla, I try to bring to Herron Power, where I'm the founder CEO now. And it's, anyway, so that's key. The other things that I value are like a commitment to hiring people who are better than you, which I think is not...

29:42Drew BaglinoAlso doesn't always come natural to people like people have a habit of wanting to throw their weight over their You know reports and and and one of the ways in which they do that is they feel better than them and And one of the ways they feel better with them is they think they're smarter than them or whatever But actually you want to do the exact opposite like you want to bring on people that are you know smarter than you Have done have more experience than you do you know, fill in the blind spots that you have and really elevate them because that's how you grow too, you know? I have like a whole management philosophy, I'll send it to you. That I've kind of developed over time. I'm forgetting, there's a bunch more, but those are the humility, the higher better than you are two key ones. I'll think of some more as we keep chatting.

30:41Lukas BiewaldWhat about like really aggressively pushing? Is that, I mean, that's something I've actually noticed that people that come out of working from Elon often have is a style where they push really hard. Do you also do that? I do. I try to use, but I try to do it pragmatically, but I do. I think the, it's something I learned from my father. My father was an electric engineer. I met him, yeah. Yeah, yeah, yeah. But awesome guy. What was your father? My father actually, it's funny. My father does consulting on

31:12Drew Baglinoelectricity grid. My father was an engineer and he was also an engineering manager and led divisions within Pterodyn as like the delivering the products. And he was always telling me, even before I was at Tesla, that basically when you develop a timeline, you have to do the bottoms up thing where the team It says this how long I think it's going to take. And then you have to kind of bring your top down view of like, well, what is the business need or what do I think the team can do? And you sort of like push as hard as you can for something in between because you give people 12 months, they'll take 15 months. If you give people nine months, they'll take 10 or whatever. And that's just human nature. So yeah, I try to bring pragmatism to it. I don't.

32:10Drew BaglinoI don't think it's worth denying some reality. I think that is where you run into problems if you're like, okay, well, we've got this large stamping tool and it's going to take whatever 40 weeks to get it cut and delivered and ramped up as a tool. I've been through enough programs to know that that's like...

32:39Drew BaglinoThat's probably that's not that compressible. Like maybe you can make it 38 or something. But then you have to, when the team brings that as the reason why the project is two years long, then you have to say, well, we could soft tool it. We can work around that part for a lot of the early prototyping. We can find ways to compress the program that maybe add a little bit of cost, but don't.

33:08Drew BaglinoDon't require the project to be this long. I think people have this habit of long pole in the tent Whack-a-mole where they're like trying to put a pole up to make the schedule be what they want the schedule to be And it's sort of your job as a engineering executive or an executive of company to Just keep like cutting the poles down with the long pole though, right? They want to cut down the long pole. Yeah. Yeah keep but the people just keep putting up new examples It's like, oh, it's okay. Okay. So we figured out how to have the stamping up your problem. Now it's, now it's something else. Like, and you just have to keep kind of going through that a couple of times and, uh, and sort of like train the team on, on what, how you're willing to let the schedule be developed. Cause I think when people, when people land on a long pole, then anybody that isn't the long pole just sort of like rests and is like, well, we gotta wait for that stamping to show up. So like,

34:07Drew BaglinoI don't need my part to be done until, you know, 36 weeks from now, because my lead time is only four. And it's like, uh-uh. No. If your lead time is only four, you know, do your work now, get in here in four weeks, rev it till it's perfect, and go help the guy with the 36-week lead time get it done sooner. You know, it's funny. My experience with software is kind of different, where I feel like what usually happens with software projects is if you ask people how long each component's going to take,

34:34Lukas Biewaldthey weigh underestimated, right? They're talking about the first 90%, the second 90%, the third 90%. And so if you really do a bottom-up analysis and you let a team do it, especially a junior team, you'll end up with this ludicrously short timeline. And you end up really needing to pad it and watch it. And then you kind of watch it expand over time. It's not usually taking a longer timeline and pushing them to compress it. It's more like trying to get people to be realistic. I mean, yeah. So with software, I totally agree. Software is so different than hardware. I think with software,

35:03Drew BaglinoIt's all about what's the list of like integration milestones or like demonstrate functionality demonstration milestones that you want to rally around and like architecting that list and like sequence in a way where you're kind of getting the nights and weekends work, but you're also like focusing folks because I think software is there's a lot of hazards and software. You can basically fill time.

35:34Drew Baglinoendlessly with more code or rearchitecting over and over again. And that's why I think short deadlines are really important in software, right? Because you get people to cut scope and do the thing that actually matters. Yeah. And so one of the things that because I started writing software and like physical systems and the physical systems had, you know, real deadlines, like, well, the car needs to be driving for this event. Like we're going to have, you know, we're going to do this with road service crazy. We did like.

36:02Drew Baglinowe had like a 400 person event at a airport in LA. I remember Arnold went and lots of the celebs went and like every single one of them was going to do a zero to 60 drive. Like they were all going to experience the four second zero to 60 in the roadster and you know the date was fixed and it's like the software everything needs to work like there's no it's incompressible you know and so it's super focusing and and And and the only thing we couldn't figure out how to deal with was keeping the whole system cool and all the back-to-back Zero to 60s. So actually the the like vehicle technicians came up with this like liquid No, it wasn't like a nitrogen. It was like dry ice Like rapid chiller setup that like the car would go behind the curtain and they were like doing this thing with dry ice to like cool everything down and then like send it back out

37:01Drew BaglinoAnd we did all those, they were in the 60s and everything was fine except for the magnesium cast motor mount cracked at like 300 or something. And so the, luckily the motor was pinned between the back of the battery pack and the, there was this like rear trunk thing. And so when you floor it, the motor would like slam into the battery pack. So you heard like a thump, but it still worked.

37:29Drew Baglinonice and then and then when you slowed down and went on regen it would like thunk into the trunk but everything was fine um but uh anyway that was a fun example but yeah no you need you need integration milestones uh that if you're gonna if that date was like 10 weeks from now or you know and this is something that Craig Carlson taught me and he was amazing at it's like okay what are we doing this week what are you doing next week what are you doing next week and like

37:57Lukas Biewaldjust laser focusing the team on that stuff. This is total aside and I do want to get to Heron Power but I had this interaction with you I think like 10 years ago where you're talking about like luxury cars and what makes a luxury car and it was things that were so stupid like having like a bigger piece of like metal or something that it's completely maybe not ever want to buy a luxury car and always drive cheap cars.

38:24Drew BaglinoDid I take the right takeaway from that conversation? I'm trying to remember the conversation. Well, what makes a car like fancy? Like are they actually sort of faster? Well, okay. What makes a luxury car? I think some of it is the choice of metal. I think, yeah, so there's some interesting...

38:49Drew Baglinoexamples like the NVH package, the noise vibration and harshness package, which is basically the content that you add to absorb the sound that the wheels make on the road, or the tires make on the road, that they're the primary sound generator. And then in an engine car, an internal engine, combustion engine car, the engine makes noise. So you basically spend money to absorb that sound. And in a budget car, you spend not a lot of money. And in a luxury car, you'll spend, I don't know, four or five hundred dollars. And so the difference between the regular sedan where you spend fifty bucks on NVH material and the luxury sedan where you spend five hundred dollars on NVH material is the difference between a quiet cabin and a acceptably noisy one.

39:47Drew BaglinoYou know, that's that's an example. That's pretty good. Yeah, but that's the kind of thing you're doing with luxury car another example would be like Closing the door, okay? So a luxury car it sounds nice when you close the door. It's got this like satisfying thunk You know, it's like as a well-built car like I will I'm happy I bought this car right and I'm not luxury car. It's like the car door closed and stayed closed when it was supposed to stay closed in this accident and that accident and like that's what it did you know and and the difference between the satisfying sound and the not so satisfying sound is like the stiffness of the structure of the door and the the like uh how many lips on the seals and all of these things contribute to like that thunk and the solidity feeling of the door again it's it's weight

40:41Drew Baglinobasically like materials that you've added, you can always optimize it. But the thing is if you take the optimized luxury car, it just means the not luxury car will be lighter. Like that is kind of the thing you're trading or trading some mass. I think it is mass and some cost for these sort of creature comforts or these perception aspects of the vehicle. And I think like a car guy after all these years, like we're gonna cars do like Look at cars and think about, no. I never was much of a car guy. I mean, I love, I will never drive a non-electric car or at least not happily. Yeah. I mean, I love how much quieter they are, the responsiveness. I can understand like the linearity of steering and what good damping feels like on a suspension system. Like I can appreciate all of these things, but.

41:35Drew BaglinoI was never, I'm not like on the weekends going and like rally driving on the racetrack or really following, I mean of course when I was at Tesla I was following all the competitors but like as soon as I left Tesla I really wasn't. Although you do kind of drive like a maniac. Is that because you are confident in the specs of the car?

41:55Drew BaglinoWas that before Tesla? No, that Tesla definitely. Yeah, I spent a lot of time on racetracks at Tesla Developing the trash control the stability control, you know, I was responsible for that aspect like sort of how the car stays on the road from their perspective of braking and steering and and the and the You know the propulsion system So yeah, I've been spent times on you know dry racetracks like slick like where you put water on to understand what rain driving the rain is I've been on I've been in Alaska Minnesota Sweden on snow and ice driving and you learn how to interpret Where you are at relative to like the edge of stability on a vehicle just from the feedback you get to the steering steering wheel the sound of the tires like you learn these things And since I did it for so much time

42:53Drew BaglinoAnd I'm still driving those same cars today. Like I still, I drove a Model 3 here today. You know the car really well. You can understand its limits and you can have fun. Nice. And I'm always chronically late. So I think that's another reason I drive like a little bit faster than most people find comfortable. What about autonomy? I mean, you were in the first people I knew talking about autonomous vehicles and then it got really hot and you were right in the middle of it. Yeah. How is that?

43:20Drew BaglinoTell me about what you're thinking when it started and then how is the trajectory surprised you? You know it was yeah So I JB one of the things that JB asked me to do in 2012 2013 was start working on this problem statement and like You know, we needed to be competitive the model us need to be competitive with all these German cars and You know have adaptive cruise control and a lot of these like active safety systems So I went out and you know met with Bosch to like see their radars and I talked to Mobileye and like you know understood what was out there and I hired some people I hired You know some some controls folks to start developing the autonomy stack At the time we didn't really call it the autonomy stack. We called it adaptive cruise control, but to develop the

44:18Drew Baglinothe car driving itself functionality um and was responsible for autopilot like in the first year or so as we were putting the team together um and you know i didn't have your background like i didn't have any exposure to machine learning at that point i didn't have any exposure to what computer vision could be you know i was and still am like very much like a physics first control systems i took i've I was an electrical engineer under God and I had a concentration in, you know, signals and systems, which spaces like control theory and in electrical circuits and in and in software. And so I came at it from like a physics based perspective, like, okay, if I'm going to have a car drive itself, you know, I'm going to create a physical model of the world around me. I already can fully model the car and can plan optimal control for like the car from here to there.

45:17Drew Baglinoif I understand where the car needs to go. And that worked really well for adaptive cruise control. But it pretty quickly became clear that that wasn't going to work well all the way to the end, even back then. And I think what was interesting to me in that whole period of time, this was 2014, I had 2015 I think, or I think it was 2014.

45:47Drew BaglinoSo Mobileye had a demonstrator car to sell their cameras. And their demonstrator car was an Audi paid or something like that. And they would, you know, have the car drive itself off of their single monocular, like single camera image sensing stuff. Wow. Just one camera. Just one camera. Wow. They have the car drive itself. And I think they also...

46:11Drew Baglinodid some fusion with the radar or something. So they had a camera and a radar, and they'd have the car drive itself. It was a really impressive demo. And Elon got a demo from the mobile ICO. And we were doing adaptive cruise control. We weren't talking about the car driving itself, nothing like that. We were developing the platform of the vehicle to meet the Euro NCAP crash safety standards, or active safety standards, like detect pedestrians, cyclists.

46:41Drew Baglinoyou know, automatic emergency braking, do all this stuff with like some radar, some camera. So we had been developing, this was me and this guy Satish, we're developing this like platform that could hit all the European requirements and also have the car do adaptive cruise control. This is, yeah, early 2014 and Elon comes out of this mobile iDrive and it's like, we're gonna have the car drive itself. And so then it's like, you know.

47:08Drew Baglinoforget about meeting the European requirements. We're just going to have the car drive itself, uh, fully. And we're like, so this is the hardware we got. We're going to use this hardware. And he's like, yeah, that hardware can do it. And we're, and it was like, okay, you know, and that's when the like people have two, two, two eyes, like, you know, they drive. Why, why can't we drive the car with, uh, with just vision? That's when that sort of started. It really came out of that mobile eye drive. Um, and, you know, I was in the position to meet with Elon every week to be like, so when's the car going to drive itself from California to Boston? You know, that was the every meeting. That was his question, um, to me and the early autopilot team. And he's like, can it be next week? Like that was the question. This is the, this is the, uh, the amazing value of like the distortion reality distortion field that like folks like Elon and others bring to a conversation. It's not like, so how many years is it going to take? You know,

48:05Drew BaglinoIt's like, can we be tomorrow? You know, like, why can't it be tomorrow? It's just like, that's the question. And you know, the team is ramping up, we're getting like the computer vision people in, we're getting all the people in that can sort of start thinking about how the car could drive itself. And every week, that's the question. And at some point, I'm like, you know, I'm an energy nerd. I'm like a physical systems person. I think I'm going to go do something different.

48:35Drew BaglinoAnd that's when I started working on Tesla Energy that, you know, when the first like Super PAC Powerwall 1, you know, the team was like three engineers, I started doing that. And Autopilot moved on to, you know, a series of leaders that tried to work with Elon through the Why Can't It Be Tomorrow reality. And we're still in the Why Can't It Be Tomorrow a little bit.

48:59Drew BaglinoAlthough it's really impressive really awesome. Yeah, this is incredible totally and I mean the team a show I actually interviewed a show like back when he joined, you know Because I was part of like building that early team I mean a show because amazing the team is amazing the Tesla team is incredible like what they've done to go from heuristics and physics-based like optimal control planning to nothing but nets like pixels to actuators is incredible. Could I have predicted that? No, I just didn't have the context at all. I knew how to do estimation and Coleman filters and all of these things that are in control systems where you are working with unclear, where you don't have all the information and you do have to do effectively a recursive on the fly state estimation. I was familiar with all of that type of work.

49:58Drew BaglinoBut I just had and I had done I took like an AI class in college But that I knew I knew about like Bayesian trees and like all of that. I think we took the same classes Yeah, I knew about it, but like I had no idea where it would go. I could never have predicted it and And kudos to the team for like sticking with it and getting to where they've gotten it's incredible totally all right, so let's talk about electricity grid so so What what made you start here in power? Yeah, it was this conviction that electrification is accelerating, that there's gonna be more electric cars, gonna be more electric heating, we're gonna electrify industry as well. Just in general, we're going to see a growth in AI also. It's gonna say you're like two cents is in, you haven't said data center. I know, I know, I know. Sometimes I need to stress the other things, because AI is said so often. But yeah, we have like low growth.

51:02Lukas Biewaldin AI. Which, by the way, when I started, Heron wasn't as obvious as it is now. Totally. Okay, but maybe backing up, because I'm actually really not an expert on this, and probably people watching or listening are even less experts. What's the problem here? I turned the lights on, the lights consistently go on. The electricity here seems pretty stable. We complain about power.

51:28Drew Baglinoprices in California for some reason but you know of all the things in my life like the electricity grid seems pretty stable and okay like what's the issue yeah just an interesting fact about California so the retail rates are really high in California but the average household electricity bill in California is like middle of the pack in terms of average electricity bills in the country and it's because there's just people don't actually use a lot of electricity in California very few people have like air conditioning or everybody's on like, you know, fancy LED lighting. And it's a moderate climate, right? So like it's not like Texas. Right. And so while the retail rates are high, the average cost of electricity to a homeowner is middle of the pack, which is just a fascinating thing. So it actually means that, you know, if you go back 10 years before the retail rates were high, electricity was super affordable in

52:24Drew Baglinoin California, which is kind of like contrary to the thesis. But I'm not here to defend the retail rates in California. I think they need to go down and they are going down as we bring more load on. And maybe that's kind of the point. So from the 70s, maybe 80s till just a couple of years ago, there really wasn't electricity load growth in the United States and in the Western world. And it was because of energy efficiency in all of these technologies. Lighting.

52:54Drew Baglinoair conditioning Computing in as well, you know data center load growth was not very much and actually as a percent of total end-use electricity did not go up very much From like the 90s all the way till like the 2010s because the efficiency of compute kept improving dramatically and not just compute but like storage like everything that networking how to cool them Like everything just kept getting more efficient. So you kept adding more compute, but like you kept making it more efficient at the same time. So the overall load of the United States hasn't grown since the 80s? Yeah. From like 80s, the 80s and the 90s and the 2000s, 2010s, like through that whole period of time, like average load growth was 1% or less. So very little growth in the system. And what that did was.

53:50Drew BaglinoI would say almost like calcified the industry that supports the utility sector, like the supply chain to the utility sector, because if you're not in a growth industry, what are you doing? And so a lot of the talent left the states and went to areas that were growing like Japan and Korea, China, India, because that was where, you know, the demand was going gangbusters where new things needed to be deployed.

54:20Drew BaglinoAnd so like the new energy technologies were being developed and improved other places. And the other thing that added to this calcification is that the utility incentive model was sort of built around the fact that like electricity use isn't always going to go up. And so the way utilities make money is not by increasing the sales of electricity.

54:49Drew Baglinobut by getting a guaranteed rate of return on the CAPEX they deploy. So now you imagine this period of time when there's not load growth, really. So the utilities don't have a lot of reason to deploy a lot of CAPEX. What's going to happen? The suppliers are going to charge more money for what they are deploying, because that's how the utilities get a return on investment. And this is all natural incentives. It's all just incentives. The utilities incentivize to spend more money.

55:21Drew BaglinoThen yeah, yeah the utility the utilities have an incentive to deploy more CapEx here every year so I don't think we'd maybe make some awesome projects that get all kinds of wild New technology, I mean they're they're regulated so like they have to justify their rate case to the public utilities commissions and the public utilities commissions are gonna say like Hey, like why are you why are your expenditures outpacing inflation like they have some pushback? Okay, so that's sitting here in California. Yeah Why are the rates going up then? It sounds like nothing's changing. What happened? Yeah. Yeah. Yeah. Okay. Let me, let me just finish this thought. Okay. So, um, so you're in this environment where it's like stagnant, like the, the amount of hardware being deployed is almost met, like maintenance capex. The.

56:10Drew Baglinothe utilities are incentivized to spend kind of like more money on that maintenance capex because that's how they make the return. So then the suppliers that are like shipping a flat number of units are just kind of getting like fat and happy like charging more for that same unit over time because that's actually what the utility wants to do is spend more money on it. So that whole supply base became siloed and a little bit uncompetitive.

56:40Drew BaglinoAnd if we're going to have electrification growing at the pace that it is now growing, 3%, 4%, 5% per year, I think we need to triple the total amount of electricity generated and consumed if we want to have an all-electric sustainable energy economy, which I'm really passionate about. We need new suppliers because the suppliers we have are like, we used to call it at Tesla, it's sort of like towing a dinosaur is kind of what's happening or a boat anchor.

57:09Lukas BiewaldLike the supply base is almost like a boat anchor on the path towards electrification. Wait, sorry. I keep interrupting, but why would electrification need triple the amount of electricity? Like isn't the main way we consume power through electricity? Like is there so much like gas cars that is that going to be the main thing that? Yeah. Well, a third electricity represents like a third of end-use energy consumption right now is like the simplest way to think about it.

57:40Drew BaglinoReally? So the gas is more than? Yeah. Well, you have fossil fuels in industry and buildings. You have fossil fuels and transportation. You have fossil fuels and like the whole chemical sector of like fertilizers and, um, and, uh, building material, you know, plastics composites. So, uh, yeah, there's if you, I worked on this, um, a paper called the master plan part three, which was effectively, can you build a sustainable energy economy? Like what is involved? Like do the resources exist? Do the technologies exist? Is it an economical thing to do? Short answer, yes, you can. But yeah, it does ultimately mean that we will produce three times as much electricity as we do today. But in so doing, we will have like a effectively a fully sustainable.

58:35Drew BaglinoAnd more affordable energy economy in the future when we do that. Okay, so why did California power rates go up? Yeah, there's like a lot of weird reasons One is the cost of wildfire abatement and the liability of fire prior wildfires. It's probably like the signal single individual Like adder in California relative to other places the other is little bit what I'm talking about of like the incentive structure and you know the cost of the components that go into maintaining and building out the grid. Inflating over time. The other is like California is actually kind of like a big state and it's in its it's a pretty like on average sort of rural state so there's actually a lot of distribution lines relative to the number of people.

59:31Drew BaglinoAnd we do have, you know, and like the utilities have a mandate to like serve the rural folks in Northern Mendocino County, you know, as much as they do in downtown San Francisco. You know, you compare that to like Delaware or like Maryland, it's just like the ratio is not in favor of the California utilities. And California kind of got built out long enough ago that a lot of stuff needs to be like repaired and replaced. So that's like a unique thing in California. And then Another one, and this is probably one of the ones that is the most controversial, is for a long time California had net energy metering for solar, which meant that you as an end customer get compensated at the retail price of electricity for putting your solar onto the grid. You don't get compensated at the same price that the utility pays to send you electricity from the grid.

1:00:28Drew BaglinoSo if you think about how that works conceptually, like when the sun is not out, the utility pays the spot market price of electricity, which is whatever it is. Plus it has to pay for transmitting the power from where it's generated to you and then distributing the power to you. So the utility kind of like pays for all those things. And then you pay a retail price of electricity for the power that you consume. But when you go the other way, the utility has to buy it back.

1:00:59Drew BaglinoAt the retail price not at the spot price whereas they could have bought power from the market at the spot price So effectively they're paying you way more than they would pay a generator anywhere else where they're not making money They would have otherwise made well the simple way to think about it is If you have all the kilowatt hours if the way that that you amortize the cost of your infrastructure is all of your costs divided by all the kilowatt hours that you serve. Well, the kilowatt hours that are served by rooftop solar, you don't get to put in the denominator. So those fixed costs get divided over fewer kilowatt hours served. And so the fixed costs go up. And rooftop solar is so popular in California that it has a significant impact on the denominator. The denominator got a lot smaller.

1:01:56Drew BaglinoOr didn't grow So the numerator even if it stayed fixed it means the like per kilowatt hour cost has to go up for the same infrastructure and the thing that is changing that now is You know, there's a high penetration of rooftop solar. So like fewer people want to have it NEM has been changed so that you don't get compensated in the same way and We have low growth again. We have electric cars in homes. We have heat pumps in homes. We have data centers, maybe even in homes like extra is this like idea of data centers in home. So all of those things have meant that for the first time in like many decades, PG&E actually had the numerator go up. The numerator of that equation went up. And so that's why the rates didn't go up in California. And they think the numerator is going to go up next year. And they think the rates will even go down. If you listen to Patty Poppy, CEO of PG&E, who I know, Patty is great. Hi, Patty, if you're listening.

1:02:53Drew BaglinoThat's what she's saying. She's saying if we keep having this load growth and the regulatory structure doesn't change, we're actually going to reduce rates in California, which would be great. So what are you going to make? I would have thought maybe you look at this and think, okay, I should make solar panels or some kind of new energy generation, but that's not what you're doing. Yeah. So what we're trying to do is provide alternative, let's say, software and silicon-based solutions to the largely mechanical passive solutions that utilities and people building power plants and people building factories have to choose from today. And again, going back to why this opportunity exists, it's because there wasn't a lot of load growth. There wasn't a lot of need for innovation in that supply base. And I saw it directly, like I was responsible for deploying mega packs. And we did over 10 years, we did

1:03:51Drew Baglinofour different versions of utility scale storage. We did like PowerPack 1, PowerPack 2, MegaPack 1, MegaPack 2. In the course of like eight to 10 years, we iterated and dramatically improved the cost, made it easier to install, did all of these things. But over that same period, we were still interconnecting to the same annoying stuff that was hard to get from the same suppliers that were largely made abroad that weren't improving. So what are these things?

1:04:21Drew BaglinoWhat are we talking about? It's like switchgear that controls the flow of electricity or interrupts the flow of electricity, think like large mechanical switches that take like hundreds of milliseconds or maybe even seconds to open. So switchgear. This is like a physical thing. Physical object. Is it like those like blade switches from Frankenstein? Total blade switches from Frankenstein. But imagine they're motorized.

1:04:48Lukas BiewaldNice. And sometimes they're inside of a vacuum, or maybe they're in some other inert gas that helps suppress the arc. But I guess I better look at that and say, how could you make a motorized blade switch better? What more can it do? Like, what's the power through or not? Yeah, I'll get there. Or pick any ways. No, no, I'll get there. Yeah, yeah, yeah, yeah. So switch gear is one. And if you go to somebody building.

1:05:16Drew BaglinoA data center or a factory like the one of the longest lead time things they have to buy is the switch gear So why is maybe the question because it is just simple mechanical stuff And then transformers And all the things that around the switch gear and the transformers which are like meters and sensors for voltage and current and all of that gear is kind of like this cottage industry that partially because The utility industry is really balkanized. There's, you know, 3,000 utilities in the United States, and they all have their own rules and requirements. And, you know, when they all were formed, you know, 100 years ago, there wasn't a federal highway system, there wasn't the internet, you know, so they all have like their down-the-road supplier, so they all develop their own requirements and their own custom stuff. So...

1:06:13Drew BaglinoThat is the ecosystem that we're looking at is this world of passive steel and oil and copper wound transformers, mechanical switches, ancillary components that have a little bit of software wouldn't around them to control them. That is what we're trying to provide alternative solutions for. And how we're doing that is by using wide band gap power semiconductor material.

1:06:43Drew Baglinoin the form of like really robust, capable proven devices that have been deployed in electric vehicles and grid scale batteries to replace these mechanical objects. Sorry, what is a wide band gap? Is that a wire? A wide band gap semiconductor. So, you know, you've seen, you know, an integrated circuit, you've seen a GPU, right? Now, imagine that that highly engineered, you know, multi-layered device isn't engineered to store ones and zeros and move ones and zeros really quickly and do math and all that kind of stuff but instead is designed to optimally conduct and interrupt the flow of electricity and block voltage. So that is what power semiconductor materials do. Wait let's talk about that because it makes sense to me that like a compute would need a really complicated system but I feel like interrupting power

1:07:44Drew BaglinoAnd letting power through that feels like very very simple. Like how do you actually? What are you actually doing? Did you did you take semiconductor physics at Stanford? I didn't and I'm sure people watching this have it. So yeah It's pretty awesome, but the way you know the way all semiconductor devices work, you know is like you're selectively choosing or doping different regions of the base material which is a the semiconductor of choice and in, you know, CPUs and CPUs of silicon. In some power devices, it's not. It's silicon carbide or, uh, GAN or other materials. Um, so you, you start with some substrate and then you kind of like dope different regions, etch different regions to selectively channel, um, or, or, or be able to control the flow of, of electrons or even like enhance the

1:08:42Drew Baglinothe number of electrons in different regions of the material. So it's it's a very like mechanical process ultimately of how you make these semiconductor devices. With where you have like masks for photolithography where you have you know you're doing like vapor deposition or atomic layer deposition of like atoms like doping atom by atom into the matrix of the

1:09:10Lukas BiewaldThe the the substrate like I'm like on GPUs like the flow of actual electrons is really light right these are really small Current at the level of the chip. I mean, oh you add them all up currents are nuts Well at the GPU because one volt and it's like a kilowatt and so it's a thousand amps, but like I think on each individual Part of the chip. It's a pretty small flow, right? No, it's I mean that's actually how the power density and the compute performance

1:09:41Drew BaglinoOver time as improved is you're actually driving higher and higher current density in the chips and figuring out More effectively how to like pull the heat out like the heat flux out Because because uh, you know effectively that you know all of the operations that are happening in the chip are Especially at high frequency, you know your you're just every clock cycle, you're charging capacitor, discharging your capacitor. And it's all these like parasitic capacitances in the device. And then you're actually like moving electrons and like electrons are moving through a resistive medium, the resistor that they're generating heat. So you have like capacitive discharge and charge and then you have the actual current flow, all of those things have

1:10:31Drew Baglinoare generating massive amounts of heat. I mean, that's what data centers basically are. You take a gigawatt of power and you convert it into something like 700 megawatts of heat and you spend 300 megawatts to get that heat into the atmosphere. That's actually what a data center is. And the 700 megawatts of heat is almost all inside the chips. There's maybe 8% lost in power conversion on the way to the chip, maybe 10%, depending on the architecture.

1:11:00Drew BaglinoSo you get a gigawatt at the grid, you get 250 megawatts to 300 megawatts of cooling power, and maybe 650 megawatts of, like, useful compute. I guess presumably your transformer is going to be more efficient than the 10% loss. Yeah, yeah. We can reduce the loss from grid to chip by about a factor of two. So we have the loss. So for somebody building, like, Gigawatt data center it means they can get 35 megawatts more useful compute out so I say cool and the switches presumably can switch faster or what's the yeah, so so back to the like what's happening at the semiconductor device so so wide band gap means that you can the Effectively you can intrinsically the way the material works you can block more voltage per unit like micron or nanometer of material and

1:11:56Drew Baglinoyou You can be more efficient like there's less loss like the highway that the electrons travel through when the semiconductor is acting as a conductor is more efficient and because you Can block more voltage per unit of the material you have like less parasitic capacitance between like where the electrons are flowing and the and however your cooling device or the bottom of the device and So you can actually switch the device faster Because that parasitic capacitance that like charge has to charge every time you like imagine lifting the output of the semi of the power semiconductor like what you're doing is you're if you if all you know This is like the top voltage that is connected to the device. This is the bottom of the of the of the voltage that's connected device the like Output of the device needs to swing from like the lowest voltage potential to the highest voltage potential

1:12:52Drew BaglinoAnd there's all these like parasitic capacitances in the material, which are all like physical. And when you go to these wide band gap materials, you can make them physically smaller. And that means they can switch faster. Or for the same loss, they can switch way faster. And that's ultimately what silicon carbide and these other materials are, GAN, is even better at that. And so it's a material science thing. You're trying to find materials that can block more voltage, are more thermally conductive, and are low loss when they're in the on state. They can conduct electronically well when they are trying to conduct. Because a semiconductor is either conducting or not. That's what makes them a semiconductor. You can control that with the gate voltage on a transistor. You can control whether the channel in the device is conducting electrons or not. You turn that on and off with the gate voltage. And there's a bajillion.

1:13:51Drew Baglinotransistors in a GPU, like I don't even know how many trillions or whatever. And a power device has one. But it's one like highly engineered device. And what silicon carbide is doing is allowing you not to make like a one volt transistor, like you have in a GPU, but like you can make a 10,000 volt transistor. And in actually a almost smaller, it is actually smaller than the GPU, and it's able to block 10,000 volts.

1:14:21Drew BaglinoAnd that's that's what material science has enabled over the past like three or four decades. And so now when you're thinking about this like mechanical switch that we were using to control the flow of electricity that literally physically needs to move the distance of my arm because otherwise we're going to have an arc and free air. Now you can do that inside of a material like a solid state material. You can start and stop the flow of electricity across 10,000 volts.

1:14:49Drew BaglinoAll within a solid-state material know nothing moving That is what power electronics is. That's the power of power So you're imagining that these big like transformers that like freak people out that like we see around, you know San Francisco in the world. Do you imagine those become tiny? Yeah, so our core our core first product that her and link, you know the transformer part is a hundred times smaller volumetrically per unit power And the way we're doing that is because we're switching the, you still need to go through an isolation stage when you're going from like a high power grid or high voltage grid to like a computer. Cause otherwise, um, if you had like lightning or something like that, you know, you might see the voltage from the grid side come over to your computer. That would be very bad. So you need to isolate galvanically isolate, um, uh, the

1:15:46Drew Baglinohigh side from the low side. But rather than doing it at 60 hertz, which is how we do it today in these giant oil filled gray boxes, we can do it at hundreds of kilohertz because these devices I'm talking about can switch so fast. So you take that 60 hertz power and you use these power devices to like chop it up to super high frequency hundreds of kilohertz. You run it through a high frequency transformer that because of the way the physics works can be much, much, much smaller. It's actually like inversely proportional. So as this frequency goes up, you can make the transformer that much smaller because it's really an energy product thing. And the way you store the energy in a transformer is a simple way to think about it is that at 60 times a second, you can move a packet of energy through versus 100,000 times a second, you can move a packet of energy through. It was the simple way to think about it because if the voltage is the same and the energy is basically current at that voltage moving up and down, if it moves up and down 200,000 times a second instead of 60, well.

1:16:45Drew BaglinoIt could be 200,000 times physically smaller. And that's the principle at work. And that's the objective is, yeah, we're making the transformers way smaller. And then these giant switch gear objects that are interrupting the flow of electricity. Well, now we have power electronics on either side of this transformer. They can natively interrupt the flow of electricity. You don't need another mechanical switch.

1:17:14Drew BaglinoBecause the passive transformer is that. It's passive. It can't change voltage on one side to the other other than in a static ratio. It can't stop the flow of electricity. You can't do anything. So you need to have these like large mechanical arms, levers do that work for you. So do you imagine replacing all the infrastructure with these new devices? Yeah. Over time, I think as, you know, We've got 50-year-old stuff out there and needs to be replaced. Yeah, you're going to replace it with active solid-state transformers instead of the passive ones, because you can get all of these interesting functions integrated into one object, and they're smaller. They're easier to service, easier to maintain. They're not flammable. The oil in those transformers lights on fire. I don't know if you have seen it, but when I lived in Bernal, more than one transformer lit on fire down the road.

1:18:08Drew Baglinoand you have a power outage, and they're just, it's a flammable oil, and you've got this thing that's getting hot inside, and if you don't maintain it, and even if you do maintain it, sometimes these things happen. So there's lots of reasons why you'll use it to replace the existing, but the primary focus we have right now is the grid is growing really quickly. There's people building solar, people building battery facilities, people building factories, people building data centers, and they need, they want a more scalable, better solution.

1:18:36Drew BaglinoSo that's our first focus is those folks building the new stuff. Do you have a point of view on data centers and their impact on the grid? I feel like that's become a pretty controversial topic lately. I think properly designed data centers can be like a grid positive because data centers are like the best utility customer. They are baseload. They consume a ton of kilowatt hours per kilowatt.

1:19:07Drew BaglinoYou know if you look at your house where you have You know a 200 amp service You maybe Which is like 40 kilowatts like I mean you're probably never 40 to 50 kilowatts you're probably never using more than 10 kilowatts peak and on average you're probably using one or two kilowatts maybe even less So the utility is invested in the infrastructure to serve you 50 kilowatts and you use one kilowatt on average Data center, you invest in all the capacity to deliver a gigawatt, and the data center uses 800 megawatts on average. They're a way better customer. If you think about the simple equation I described before, numerator, total cost to keep the grid healthy, denominator, total kilowatt hour served, the data centers are the best customers. They're basically like aluminum smelters or steel mills. They are the customers that make the grid affordable for everybody else.

1:20:06Drew BaglinoThe way that could not end up being true is if you know they're The utilities because of their incentive structure are building more than they need to to serve that data center or they're amortizing the costs of like connecting that data center more on the residential rate payers than they are in the data center. I mean, I actually think the data center should just pay for the infrastructure that comes to them because the return on investment of these data centers is so high and that infrastructure actually isn't that expensive compared to the GPUs. But if you think of data centers as a utility customer, they're the best ones. And if you actually look at the numbers, the states with the highest penetrations of the data centers overwhelmingly have had

1:20:54Drew Baglinothe lowest electricity rates and actually have had rates reduced. So you actually think data centers will cause electricity rates to go down. I do. Interesting. And the other thing is because of the uniqueness of like training, data centers need to have some energy storage, participating in absorbing all of that training ripple. And once you put energy storage at a data center, now the data center can support the grid through like peaks of uh, like peak periods of high load, um, they can stabilize the grid frequency and voltage rather than disconnecting whenever there's a problem. So they can do all these things to be like grid support networks, um, in addition to being an awesome customer that should also drive down the cost of, of the utility. Whereas historically they've actually done some things on the opposite side. The rectifiers they used have had like flicker problems. Utilities have had to kind of clean that up with extra hardware.

1:21:53Drew BaglinoOr like just happened this week, three gigawatts of data centers all turned off at the same time because they didn't have storage set up to do that. You know, and there was a little bit of a grid wobble and they all disconnected and then that can cause a liability concern. So historically, data centers were maybe a grid liability from a stability perspective. In the future, I think they can be an asset and they can drive rates down. What do you think about nuclear power? Yeah, I mean, I'm a.

1:22:24Drew BaglinoI'm down for nukes nukes. I don't really have like any aversion to nuclear power. I'm happy that Diablo Canyon was recently extended here in California I think that with proper engineering controls like nuclear power plants can be safe And of course you need to maintain them and do all those things I would love to see the US like focus on its like waste plan because there are plans in place in other countries And we don't really have one but we should have one I think The history of nuclear in Asia has been a history of nuclear power plants continuing to be more affordable to build. Like as the countries get reps, the thousands of dollars per kilowatt have progressively come down. And I think it is... The US hasn't seen that trend because we kind of stop building nuclear, but I'm confident if we start building nuclear again, we can find ways to do it more affordably.

1:23:26Drew BaglinoLike has happened in Asia and they did that without a safety any safety problems or anything like that So it's not like you're trading safety for cost So yeah, I think nuclear is needed As part of the mix like you need some base load renewable. I'm actually a big geothermal plant fan I think geothermal is gonna give nuclear Iran for its money in terms of affordable base load 24 hour By seven renewable power. What about just solar plus batteries? Like when I look at the simple math it seems like that's actually the cheapest way to get a Base load like why isn't there more focus on that solar plus batteries can absolutely provide base load power one of the things we did was we did it on islands like a lot we we took Some islands in American Samoa like fully renewable with solar plus storage like you can definitely do it and But like when you think about

1:24:20Drew Baglinothat the like lcoe the total throughput of a cost of electricity with solar plus storage when you consider Kind of that worst-case period the winter without bad storm or the number of cloudy days like if you really want to have 100% solar plus storage the lifestyle like the lcoe the lifecycle cost of energy is it's still like basically 10 cents Maybe a little bit more depending on how good the solar asset is. Maybe you can get it to eight if it's like an incredible solar asset, like a desert and you don't have to worry about the winter. So like the Middle East or something like that, you get to like seven or eight or something like that. But geothermal, it can beat that, I think. Like that's definitely what Ferro is trying to do. They're trying to get down to six or five. And nuclear, you know, when you look at the nuclear plants we have today that are like fully depreciated, like they're like two or three cents.

1:25:19Drew BaglinoAnd the question is, can you build new nuclear and beat that? Like, that's definitely not what the Georgia facility that we just turned on. Georgia, the Georgia facility we just turned on was, I think, more than 10. But like, could we find affordable nuclear? Could we develop more affordable nuclear that gets down into that below that? And is that happening in China or is the government subsidizing it? Because China is actually ramping up nuclear quickly, right? Yeah. Korea has built a lot of nukes, and they've been able to get to this.

1:25:48Drew BaglinoLCOE China to Whether they subsidize it or it's like such a tricky question, right? Like there's so much government support in the like raw material industries in China So you go and you build a nuclear facility like you're benefiting from the nuclear from the government support in the concrete in the steel and the in the like fancy metals, you know like

1:26:13Lukas BiewaldWhether they directly subsidize the nuke or not like they've subsidized everything that went into the nuke in some way or the other Okay, your your new transformer is like order of magnitude maybe or you're saying double the the efficiency Do you feel like that enables new uses of transformers that where people aren't using today? Or if that's not the intention, that's fine. I just I like the question What we're really doing is is

1:26:49Drew Baglinois changing the thought process of how you go from medium voltage down to the load. So if you're building a factory right now, you might actually have like a number of transformers in series. Like you probably have some main interconnection to the transmission grid that's 300 kilovolts or something like that. That is a transformer there that goes from 300 kilovolts to something lower like 13 kilovolts or 34 kilovolts and you'll distribute that medium voltage around and then you'll have like another transformer maybe to 480 and then you'll have another transformer from 480 to 1208 or kind of with like these cascading series of of Transformers and protection switch gear and all this kind of like stuff in a row And with you know this power electronics approach you can kind of elegantly do it in like one stage

1:27:49Drew Baglinoand also have multiple different output voltages. And so you're just like simplifying the build out of factory infrastructure, electrical infrastructure, or energy facility infrastructure, or data center infrastructure. And I think that's a good thing because ultimately if we are gonna triple the electricity sector, we need to be able to sort of kill three birds with one stone, right? We're just doing more with less.

1:28:20Lukas Biewaldplaying around with building little robots in my garage. And the motors you need actually, they have a significant power draw that they can kind of melt the little wires and cause problems. So I've been using like a little transformer off the battery to get down from like the 24 volts that the lipo battery puts down out to like five volts. And that thing is actually surprisingly heavy. Are there like mini versions of what you have that could be used in sort of like consumer electronics?

1:28:49Drew BaglinoYes. Definitely. I'm looking at one right now that is a probably like a 200 watt, 120 volt, actually it's 240 volt AC to, you know, 15 volt output, you know, it's powering the laptop over there. That is an example of a solid state transformer. And, you know, that's, you know, an Apple, an Apple AC to DC for, for a MacBook, right?

1:29:19Drew BaglinoThat thing is probably switching 800,000 times a second power semiconductor devices inside of it. But that's like 60 bucks probably, even if you bought an off-brand one. Shouldn't it be cheaper? It's definitely cheaper than if you were doing it at 60 Hertz. That same thing if it were 60 Hertz would be like 10 times bigger, way, way more, and cost a lot more. Interesting.

1:29:50Drew BaglinoYou have to like there's like a grasses greener phenomenon and like that is much smaller than it would have been for the same power like five years ago also with power electronics and I like to pull out this like 70 watt Tiny Gan adapter actually. Oh, yeah, let's see it. It's like 70 watts two outputs USB-C Does like five or six different Voltages, I mean it's and power shares like from one to the other Wow, it's tiny How many amps can it do? Oh, plugs right into the wall. Cool. So basically what we're doing at Heron is like the same way that this technology has made these, you know, consumer electronics chargers tiny and super capable. We're doing that, but for industrial electronics, you know, megawatts instead of 70 watts. So 5 million watts is the rating of our product instead of 70 watts. Bottom. Same concept though.

1:30:50Drew BaglinoAnyway, what's inside this so yeah, you've got you have an isolation transformer That's probably the size of my pinky that is isolating the 240 volts AC or 120 volts AC from the 5 to 15 volts DC and on either side of that isolation transformer you have these power electronics semiconductor devices that are the die area of each is probably like 10 millimeter by 10 millimeter or maybe 20 millimeter by 20 millimeter, probably more like five millimeter by five millimeter. I want to take it apart and look at it. Yeah. So yeah, you have like five by five die area power semiconductor devices that are GAN, that are switching a million times a second the voltage waveform from the grid to make the DC across that transformer. Cool. I mean, one of the things lately, I've noticed my USB-C cables, they like power it like wildly different.

1:31:45Lukas BiewaldSo then I got like little meters to like measure like how much power I'm getting from my USB Cables I think you buy them like amazing and you realize that some of the cables are actually limiting The amount of the current that you get and then I kind of realized actually it's upstream that that converter Different ones will put out different amounts of yep amps and so then I really wanted to get like good ones where I could like you know There's a ton of charge it's my laptop and and charges super fast But actually like couldn't find ones that actually got to the top of the USB-C rating rating So are we gonna should we expect like better converters coming in the near future? Yeah, I think it's I think it's actually happening in tandem It's like the batteries also have to become more capable because they're they're also have a bottleneck like I like to think of it Have you heard the parking lot problem? No, so lithium-ion batteries

1:32:34Drew Baglinocan be thought of as like a giant parking lot at a stadium. If you imagine that the parking lot at a stadium has exactly the number of spots, as there are people going to the stadium, like it's got 60,000 spots. If the parking lot were set up as like a square, it would take forever for the 60,000th person to figure out where to park. Totally.

1:33:04Drew BaglinoBut if the parking lot were set up as a line, and the person could literally just like drive along until they see the hole, they could find their spot really quickly. Now, the line doesn't really help you get to the stadium. You need something to like get them from the line to the stadium. But that's effectively the battery problem. There are parking spots for Luthymions in the anode and the cathode.

1:33:33Drew BaglinoThey can only get to their parking space so quickly and and making a Linear parking lot in a battery is actually super expensive because it's like a ton of surface area interesting, so you're always like trading this like 2d 3d problem Is that where the charging slows down so much? Yes, exactly. It's like you can find the first spot easy in the parking lot You can find the last spots really hard. Well, that's great analogy. It's the best analogy for lithium-ion batteries. All right, maybe good spot done. Thank you so much

1:34:03Lukas BiewaldThanks so much for listening to this episode of Gradient Descent. Please stay tuned for future episodes.