Catalyst with Shayle Kann - Drew Baglino on Tesla’s Master Plan

Episode Date: December 26, 2024

Editor’s note: For the holiday break, we’re bringing you one of our most popular episodes of the year — a conversation about Tesla’s Master Plan 3 with Drew Baglino, who stepped down as the company’s senior vice president for powertrain and energy in April. Tesla’s Master Plan Part 3 lays out the company’s model for a decarbonized economy — and makes the case for why it's economically viable. It outlines a vision for extensive electrification and a reliance on wind and solar power.  In this episode, Shayle talks to one of the executives behind the plan, Drew Baglino, who was senior vice president for powertrain and energy at Tesla until April when he resigned. In his 18 years at Tesla he worked on batteries, cars, and even Tesla’s lithium refinery. Shayle and Drew cover topics like: Why Drew isn't sure that AI-driven load growth “is going to be as dramatic as people think” Drew’s optimism about the U.S.’ ability to build out enough transmission for decarbonization How to deal with the high rates of curtailment and what to do with that excess power Meeting the material requirements of decarbonization and Drew’s experience with permitting Tesla facilities  Recommended Resources: Tesla: Master Plan Part 3 CNBC: Tesla execs Drew Baglino and Rohan Patel depart as company announces steep layoffs The Carbon Copy: AI's main constraint: Energy, not chips Catalyst: Understanding the transmission bottleneck Catalyst is brought to you by EnergyHub. EnergyHub is working with more than 70 utilities across North America to help scale VPP programs to manage load growth, maximize the value of renewables, and deliver flexibility at every level of the grid. To learn more about their Edge DERMS platform and services, go to energyhub.com.

Latitude Media, podcast at the frontier of climate technology. I'm Shail Khan, and this is Catalyst. We built a model that is trying to find the lowest cost total investment to solve the balance of demand and supply. And it turns out that with the current known technology costs, what the model does is it really largely overbuilds renewables to solve the winter scenario. Well, talking about Tesla's master plan three is actually the key to Shale's master plan part one. The rest, you'll just have to buy me a drink to find out.

When utilities need flexible capacity they can count on, they turn to Energy Hub. Energy Hub works with more than 170 utilities, coordinating over 2.5 million devices to manage 3.4 gigawatts of flexibility built for the moments when utilities can't afford uncertainty. Energy Hub builds and operates virtual power plants that utilities actually stake their grid planning on, coordinating EVs, batteries, thermostats, and more through a single platform built for utility scale. Predictive, verifiable, and designed to perform when it counts. Learn more at energy hub.com. Trillions of dollars are flowing into clean and critical infrastructure, but those investments aren't driven by technology alone.

They're shaped by markets, by policy, by capital, and by the institutions that connect them. I'm Alfred Johnson, CEO of Crux, and host of a brand new podcast, Critical Capital. Each episode, I talk with people deploying capital, shaping policy and building the clean economy. Tune in as we unpack how progress is actually made. Listen to critical capital on Spotify, Apple, or wherever you get your podcasts. Catalyst is supported by Fish Tank PR, an award-winning PR firm focused on climate and energy tech, renewables, and sustainability. Fish Tank is known for generating prominent and effective media coverage for the brands they work with. If you want a PR partner that's thoughtful, shoots straight, and gets results, you'll like

Fish Tank PR. To learn more about Fish Tank's approach, visit fish tankpr.com. That's F-I-S-C-H-Fish-Tankpr.com. Hey, it's Shale. Happy holidays. For your gift from all of us at Catalyst this year, you get a rerun. But it's actually quite a good rerun. One of our more popular episodes of this year, we're replaying an episode.

of a conversation that I had with my friend Drew Baglino. We had this conversation not too long after Drew left Tesla, where he had been for 17 years, and by the end was the SVP of PowerTrain and energy and all sorts of things that he led there. Anyway, we didn't talk so much about Tesla, except to the extent that we were talking about Tesla's Master Plan 3, which I think is a document that did not receive as much attention as it should have in my mind, because it was a really good, high-level, very ambitious vision for decarbonizing the global economy that Drew and a bunch of other folks at Tesla put a lot of thought into, and I thought reflected some interesting visions of the future and technology pathways. So Drew and I had a really good conversation about that, and I'm happy to replay it for all of you this holiday season. See you in the new year.

Drew, welcome. Thanks, Shale. Happy to be here. All right. So let's talk about Tesla Master Plan 3. I went back in reread it again recently, and I actually read the first two also, and it's a pretty notably different, right? The first two are about Tesla. It's like, here's Tesla's plan to take over the world. And the third one is very much not specific to Tesla. It's like, here is the plan for the world. So I'm interested in the background of like, what was the thinking behind master plan three being what it, what it was and the departure from previous master plans? Absolutely. The thinking was there's a lot of noise out there about whether a sustainable energy economy is actually feasible, not only technically feasible, but commercially feasible.

Is it going to bankrupt the globe or something like this? And do the resources exist? And so, you know, for a company like Tesla where the mission is to accelerate the transition to sustainable energy, the broader feasibility needs to be, you know, settled. It shouldn't be considered a question. And so, you know, myself and a few people, others were tasked with kind of putting together why it is feasible, not just technically but also commercially.

And in some ways, is more feasible than the alternative when you think about not only the fact that the typical like hydrocarbon-based economy is finite and its resources and not renewable, but also because when you stack it all up and look at the investments and the materials, it's actually quite feasible. And one of the most interesting things about it is when you electrify everything, which is what the master plan part three talks about, you actually get a primary energy efficiency boost, a pretty stark one. Let's talk about that for a second. Yeah, I mean, because that's actually how the master plan three starts is talking about wasted energy, which is something I think folks appreciate in general, but the numbers are fairly stark. So like describe primary versus final energy in this context.

Yeah, primary energy is the resource in the ground that is being converted through mechanical or other transformation into the end use. And so when you look at that for, say, petroleum pathway where you've got to kind of pay the piper at every step along the way, usually using petroleum as the primary energy source, you know, if it's shale or tar sands, there's a lot of energy involved in first getting the resource out, then you have to use energy to refine it. You have to use energy to distribute it. And eventually goes into an end use where sort of the best efficiencies at the end use in a vehicle or something like that. that might be 30%, maybe a little bit higher. And so when you stack that all up and you replace that with a renewable electrified pathway, you can get almost a tripling of end-use efficiency, well, sort of well-to-wheel or primary energy to end-use efficiency boost, which actually is one of the first things that really makes this all look feasible. All right.

So you alluded to this, and I want to get into good detail on the actual path that the master plan basically proposes. But as you said, it's a heavy electrification path. And I'm curious before we talk about that in the first place, was the idea to start from, let's consider all the different possibilities, one of which might be electrify everything that you can, but that's not the only one. Or did you come in with a presumption? Obviously, Tesla has been like pretty bullish on electrification since inception. So was it presumed that that is the, that's going to be the primary mechanism. Well, this paper, which I do encourage everybody to download and look at is, is really just intended to show one path that could be taken, right? And, you know, we went

and described the path that we knew the most about, I would say, and could articulate with confidence. But there are certainly many paths towards a sustainable energy economy looking at the resources that exist on planet Earth. And, yeah, I would not say. that it's the only one. And so, yeah, we wanted to be able to confidently describe the path. We published the paper and ask people to sort of pick at it and, you know, suggest alternatives. And there's been some feedback, but it really is just a conversation starter about, and trying to be both a conversation starter and a conversation ender, like, is it feasible? Yes. But also, let's start the conversation about the best way to go about it.

Yeah. The reason that I have been drawn to it is that my, the heuristic when I talk to people about like, how are we going to decarbonize the world, the very simplest version of it that I've always described is decarbonize electricity, electrify everything that you possibly can and then basically basically pick up the pieces, everything that's remaining, like figure it out through clean fuels or carbon management or whatever it might be. That third one, obviously, the first two have like a nice little half sentence that I can describe it in. The third one. is more complicated, but to a first order, that's always how I've described the path to decarbonization. And that's sort of what the master plan describes. So let's get into it a little bit. So it basically is structured by talking about demand in this context, specifically electricity demand because it is a heavy electrification pathway. So how much electricity demand will there be due to the electrification of various things? Then how are we going to supply that electricity? And then it addresses a question that I think people often ask, and I suspect this was your, like, preempt the complaints people have component, which is material requirements in order to do that. So let's talk about those in order, starting with the demand side.

So just to put numbers on it, today we're at, this is focused on the U.S. because that's readed the deepest modeling. So today, I think we're at something like four, a little over four terawatt hours of power demand in the United States. and this model's basically getting up to like 11 or 12. So it's like roughly a tripling of electricity demand in the United States. Knowing what you know about electricity, the market structure, the everything to do with electricity, the prices, etc., how speaking to feasibility, how feasible do you think that is? Can we triple electricity supply and demand in the United States? Yeah, and it has to just, well, you mean, can the demand happen? The demand side, I'm less worried about than the generation side, I guess.

Yeah, I guess right, right. But even the demand side is a function of, it's a function of, obviously, of supply, but it's also a function of stuff like price. Yeah, well, I think what we've seen with EVs is that given they are flexible, but when they charge, and actually, that's one of the big things in this paper is we leverage the fact that that vehicles can charge at the best time for the grid and to minimize the total investment in storage. Because EVs are flexible when they charge,

they can charge when renewable resources would otherwise be curtailed. Now, assuming that the renewable resources can actually get to the end use through the transmission, but also the other nice thing about once you've electrified everything is you can also do that at the edge, right? So you can be charging your car at your home off of your own rooftop solar,

and especially in the world we're moving where things like straight net metering are going away and being replaced with schemes where you're not really being compensated to export your power. Well, now that power is basically free to you because you're not going to be compensated any other way, so you might as well charge your car with it.

So I think the ability to affordably, at least on the EV side, we're really the average person driving, you know, less than 40 miles a day, on average, in fact, like the 95th percentile trip is is like 40 miles. You only really need to charge today's EVs with ranges of 200 to 300 miles once every three days, four days, which means you can be very thoughtful about when you charge, assuming charging infrastructure is ubiquitous, which is another thing that I think needs to happen for EVs to be successful and is happening pretty quickly as we see, especially

in Europe and China, which are ahead of the U.S. and their EV transitions. I was just in the Netherlands and there's charging everywhere. It's really amazing. And China is similar. But yeah, back to, so that's EVs as an example. Now, that's just one form of devices that need to be electrified. I think it's going to be a little bit harder with heat pumps. And there's a couple reasons why that is.

One is the main panel can be a bottleneck. Not everybody has room for a heat pump, especially in, let's say, colder areas where the heat requirements are higher, and so the heat pump sort of rating, the kilowatt rating needs to be higher. Yeah. But the next thing is, you know, we're in a weird world with like plentiful natural gas in the United States and barring some weird regional transmission constraints like in New England where natural gas can't get to New England because people keep voting against pipeline expansions and LNG terminals. You know, natural gas is pretty abundant and affordable. And the heat pump is got to compete directly against that. And so what's going to bring demand for heat pumps? I think there's a

little bit of awareness. There's a little bit of flexibility where maybe you can't always get natural gas to where you are. You're only on propane or something like that, and propane is expensive. And then there's policy, right? Like there's a lot of policies, I mean, California is leading the way where kind of natural gas is not going to be attached to new homes or allowed in new homes, and eventually that'll come back to existing homes and retrofits. But heat pumps are also going to continue to improve, I would say, and they are continuing to improve. The most recent heat pumps from Bosch, for example, have brought a 10% coefficient of performance improvement, and that basically means a 10% electricity consumption improvement. And I think there's more to go there. They're still really

far away from Carnot ideal. In fact, there's some folks out there that are saying they can get another 30% improvement in heat pump efficiency, which now you're really talking about being competitive on a total cost of ownership against natural gas, even in places where natural gas is the most inexpensive. Yeah, and also some novel designs for heat pumps that have a higher coefficient of performance, specifically in high-lift situations in, like, low-temperature regions, which is where they've performed the worst historically. I mean, what you're describing, though, the fundamental dynamic of like, it's really difficult to do something with electricity that competes straight up with natural gas is a massive constraint in the United States, right? And then you look at Europe and it's much easier. But in the United States, it's a huge problem. And you're describing it in the residential context or maybe small commercial and industrial for heat pumps. But, you know, the master plan also electrifies or it assumes electrification of like industrial process heat with thermal storage.

It includes some green hydrogen production for steel and fertilizer. There's synthetic fuels for jet, whatever it might be. And all those things are like in the realm of competing against really cheap hydrocarbons today. Yeah, the way we ran, we built a model that is trying to find the lowest cost total investment to solve the balance of demand and supply. And it turns out that with the current known technology costs, And we put the technology costs in the paper in so you can obviously do this

with different technology costs, but with the current known technology costs for different types of storage and different types of generation, what the model does is it really largely overbuilds renewables to solve the winter scenario, right? And now you have this large, overbuilt renewable base. Well, what does that mean?

That means that in many times of the year, and even in the winter on a sunny day, you will be curtailing that those renewables. if you're not otherwise using it in something like charging an EV or charging a thermal battery. And so that's how you get the affordability of these technologies to be realized because you have overbuilt renewables to that extent. So yeah, I definitely wanted to talk about that. So just to put a number to it, in this scenario in the paper,

you build so much wind and solar that you end up with about 32% curtailment across all the generation of those two. So that's a scenario lots of people have described. And, you know, the like 100% windwater solar world ends up with this massive amount of curtailment as well. I have always had two questions about that. One is a sort of practical economic question. Do we think that it is going to be economic to be building wind and solar projects that are going to be assuming 30 some, and maybe 32% is an average, right? so some of them are going to be more than this.

Like, is there a financial scenario wherein that actually is a plausible, like, infrastructure investment at scale? Obviously, there are wind projects getting curtailed a lot today, but I don't think that is viewed as an acceptable outcome for those owners, right? So, like, question one for me is just, are we going to hit a brick wall in terms of, we start to see these levels of curtailment and the economics of wind and solar just becomes really challenged. And then sort of related question two is, look, if we have 30 some percent curtailment of just an absolutely enormous amount of wind and solar, that curtailment is going to come at

different times. It'll be seasonal for solar. We're going to have a ton of curtailment in the spring, et cetera. But is there nothing that we can find that can be a beneficial use of that curtailed power, even though it is available sort of intermittently on those schedules? Like, can we not find something to soak up a couple terawatts, terawatt hours of, like, really cheap but intermittently available power? Yeah, for sure. I mean, when we were putting this paper together, we were trying to find the, let's say, the most straightforward kill on the,

is this feasible path? And there's so many alternatives, right? Like, we didn't really include long-duration energy storage in this paper at all because we don't have access to any third-party costs, numbers that we can could really depend on or performance. But, you know, there are a lot of companies that are working on that at the moment. And that would change probably the amount of curtailment proposed you'd end up with less renewables, but more LDES. And that would be a different, you know, techno-economic outcome with a similar results in terms of supply demand balance. The other thing that, you know, we've discussed and others have discussed is, isn't there some useful thing that you can be doing? on an intermittent basis or an alternative to long-duration energy storage, but it kind of operates in a similar way where you're doing some chemical process on one side and another chemical process on another.

And that might also help with transmission constraints. I think there's some interesting ideas to look at there. But the other thing I want to say that is already happening in this paper is there is a lot of use of this intermittent resource already. So we're doing hydrogen and storing hydrogen in the summer and then using that hydrogen to produce clean fuels on an annualized basis and also using that hydrogen for ammonia production and steel and other things. And so we are trying to do some things

to leverage the intermency or to manage the intermidency, including when the EVs charge. And I think there's another sort of feasibility question about that, which is even though it's so logical for EVs to charge when renewable resources will otherwise be curtailed, will the pricing signals

and the behavior change actually occur. Now, this is just a general question. I know you've done a number of recent podcasts, actually, on tariffs and tariff design. I think that that is, those experiments need to accelerate and propagate into more markets, and we need to drive more consumer behavior change and more charging, you know, more plentiful charging.

The interesting thing about both China and the Netherlands, which are my most personal recent, experiences with, you know, truly prevalent charging everywhere, they're really not doing this, even though they're charging everywhere. And so, and the Netherlands is also sort of struggling with their grid is kind of grappling with the challenges of their policy changes and they're behind. And so they actually need to, and they're starting to develop a standard to think about how to accomplish this. And a global standard to accomplish this would be super useful to kind of codify this availability of otherwise curtailed renewables that can be offered on the super

cheap, let's say, and getting the demand side to respond. Is your view, I mean, maybe you're saying the experiments need to accelerate because we don't know yet, but is your best guess that pricing signals to consumers is ultimately the path? Like, if everybody, if we get introduced real-time pricing to everybody overnight, does that mostly solve the problem? Or is it more a challenge of, like, like consumer behavior and, you know, maybe that introduces a scenario where the utility should control the charger within some bounds or something like that. I think it's much more about what is the consumer product that drives the consumer behavioral

change. Like, is it folks are going to respond better to $100 a month? Hey, I'm going to pay you $100 a month, but you have to charge when I tell you, or are they going to prefer the direct pricing signal? And actually, if you look at what Tesla's doing now with Tesla Electric in Texas, they're actually almost providing the choice. Like, you can get, you get a fixed rate to charge your vehicle, like, at night, or there's more of, like, get exposure to the tariffs, pay-as-you-go kind of thing. So I think we don't really know. It's kind of like insurance, you know, car insurance. There's pay-as-you-go insurance where you take a little bit more risk, and then there's just flat-rate insurance. And I think we're going to need to,

I think the techno-economic, like, global, cost-optimal thing is going to be, is going to require that we leverage this curtailment for better purposes and minimize their curtailment. And the question is, will the policy and behaviors enable that to happen? Yeah. I want to get back to the distributed energy resource thing, but actually, since we've talked about wind and solar, one other component here, so it's heavy electrification, lots of new electricity generation, but it is very wind and solar focused. I'm curious, it obviously does not include then the suite of other zero-carbon electricity generation approaches, which includes

nuclear of various stripes, includes, you know, hydrogen for power, includes carbon capture on fossil plants, includes geothermal for that matter. What, you know, was it a modeling decision not to incorporate a lot of that stuff, or is it a view that you have? It was a little bit of a complexity and also if you took the best known cost of all those technologies and just stuck them into the optimizer, the optimizer wouldn't pick them is

really what it comes down to. Because of cost, basically. Yeah, just based on like the investment, the total investment required and it was trying to minimize total investment. Now, things could change. There are people out there, like Fervo as a company as an example that are really trying to reduce

the capital costs per kilowatt of of geothermal, which would be a major breakthrough. You know, firm renewables. Firm anything reduces the overbuild of the intermittent resources, right? So we didn't include nuclear. I personally have no objection to nuclear. It's just, again, when you look at the capital cost per kilowatt of nuclear, with everything

that is known today, it's much more expensive than even highly curtailed renewables. And there are a lot of people out there trying to change those things. And I think the incentive is obvious, right? Like, you can either have 30% curtailed solar and wind or something that's better, right? Is that geothermal at $3,000 a kilowatt? I don't know. Like, right now, geothermal is at $8,000 to $10,000 a kilowatt. If you can get to $3,000 and you compare that to solar, it's a firm $3,000, you know.

Right, or solar plus Eldaz or whatever. Right. You know, it's starting to look good. So I think there's definitely room to improve. And actually, in some ways, the curtailment. reality is driving all of these innovations because everybody can see the future and be like, well, you know, when you look at how much solar is curtailed and what that effective investment cost per kilowatt is, okay, now all these other technologies are in the mix and could be interesting.

Virtual power plants are becoming a reliable way for utilities to manage capacity, but enrolling devices is just the start. What really matters is confidence, knowing those resources will perform when dispatched and being able to prove it from the control room to the living room. Energy Hub's platform handles the full picture, from near real-time forecasting, locational dispatch, and the kind of rigorous verification that holds up when regulators, grid operators, or leadership ask, did it deliver? Easy enrollment creates momentum, proven performance builds trust. That's why more than 170 utilities rely on Energy Hub to manage over 2.5 million devices

delivering 3.4 gigawatts of flexible capacity. See what that looks like at Energyhub.com. We're living through a profound economic shift, and energy sits at the center of all of it. Trillions of dollars are flowing into power plants, transmission lines, battery factories, data centers, but the future of energy isn't shaped by technology alone. It's shaped by markets, by policy, by capital, and by the institutions that connect them. I'm Alfred Johnson, CEO of Crux, the capital platform for the clean economy. Join me for my brand new show, Critical Capital, as I talk with people deploying capital, shaping policy and building projects.

Together, we unpack how risk is priced, how incentives are structured, and how progress is actually made. Listen to critical capital on Spotify, Apple, or wherever you get your podcasts. Are you tired of overpaying for big-name PR firms, but not really knowing what they're delivering? Is your comms team wasting time reviewing lengthy messaging briefs and decks, instead of engaging journalists or producing content? Are you wondering why your competitors are getting press and you aren't? Fish Tank PR is an award-winning climate and energy tech, renewables, and sustainability-focused PR firm dedicated to elevating the work of both early stage and established companies. Whether you need to position yourself as a thought leader in between project announcements

or translate complex ideas and technologies into tangible, compelling stories that resonate with the media, fish tank can help. Check out fish tankpr.com. That's f-i-s-c-h-fishtankpr.com. Okay, let's talk about the elephant, two elephants in the room at the moment as it pertains to any, like, heavy electrification strategy, right? And those two elephants are, one on the demand side, the rise of electricity demand that is independent of decarbonization. In other words, data centers basically manufacturing to a lesser extent as well. Right. So, like, how big a challenge is it going to be to, like, you're modeling a tripling of electricity demand.

presumably not including any of that, right? Yeah, there's no growth included, which you could definitely say is unfair in the analysis. But I think we tried to avoid stating a growth rate, a global growth rate, because there are people that are super worried about population collapse, and there are some real population collapses that are going to happen in Italy, South Korea, Japan, and other highly industrialized nations. that would tend to send the command the other way. But then at the same time,

there's people are always coming up with great new ways to use energy and some not so great ways, like Bitcoin. Yeah, I was going to say, one thing I didn't say about curtailment is like, I do know what is going to soak up all that curtailed power. If nothing else does, it's going to be Bitcoin mining. 100%. Oh, my goodness, yes.

I've been describing once every six months on this podcast, my idea of putting a Bitcoin mine on a barge and then putting it in the northern hemisphere for six months out of the year and the southern hemisphere for the other six months to manage seasonal variations and energy generation. Anyway, yeah. Yeah, I mean, I think, but I am curious how you think about that. It's a real challenge now, which is all of a sudden,

load growth looks really dramatic relative to recent history, at least in some regionally clustered areas where data center regions are going in. Like, from a decarbonization perspective, Do you view that as, is that a headwind? It's definitely a change. I mean, there's so many changes in the electricity sector. If you just go from the 90s to today, in the 90s, the electricity sector was flat to very minimal growth,

certainly growth below the rate of GDP. And now there's some potential, well, let's not say it's potential hopeful, reality that by electrifying everything, ignoring the demand growth, we will see the electricity sector grow at higher than the rate of global GDP growth. It kind of needs to to achieve this objective. And then there's even additional demand gen in the form of data centers. I'm not so sure that it is going to be as dramatic as people think, both because I think there's a little bit of the toilet paper problem going on here. And if you don't know what I mean by the toilet paper problem. It's like in COVID, COVID happens and all of a sudden there's

no toilet paper anywhere. And it's like for whatever reason, everybody was like, oh, I know I'm going to have to use the bathroom. And so I, you know, and I don't know the next time I'm able to go to the store, so I got to buy a lot of toilet paper. It's a run-on-the-bank situation. Yeah. Although I like describing it as the toilet paper problem better, so go on. Yeah, sure. So the toilet paper problem applied to data centers is happening, right? You've got all of these companies that think AI is going to be massive and want AI to be massive. And they're like, well, the real bottleneck on whether I'm going to be a winner in AI or not is GPUs and, you know,

megawatts of transformers and cooling towers and, you know, diesel gen sets, all the things that make it possible to have a data center. So I'm going to go and like order a whole bunch and enter a whole bunch of interconnection cues and like blah, blah, blah, blah, just so that I'm not the loser of the AI race. And I don't think all of it is going to be built. I really don't. And then on the other side, you have, Nvidia is doing a good job and other companies to reduce the watts per flop of useful compute. And I think the most recent, I actually don't know the percentage reduction, but the most recent chip that they announced was a significant reduction in. Yeah, I've been actually spending a bunch of time trying to make sense of these metrics, note topic for a future episode, because they

announced the 25x improvement in energy efficiency, which we then dug into a little bit, and it looks like it's like a 20% in terms of watts per flop. So the metrics people are using, I will note, there was another, I think it was Google or somebody, one of the other hyperscalers has their own chip that they're building, and they said an 8x improvement. And like similarly, it's not exactly what they're saying. So there's, I can't, as far as I can tell, there's not consistent metrics here. But point taken, right? Like the history of compute is one in which we expand compute dramatically while keeping overall energy consumption basically flat. Like, that's been true for a couple decades. So people think it's different this time in the

AI world, and there is some possibility it's not. Well, there's also a little bit of a reality check here where, you know, one of the differences between the 8x and 25x and the total power consumption is that's talking about like the main chip, right, the core processing unit. But one of the biggest bottlenecks for everything that's going on with AI is actually the interfaces, the memory interfaces, the data interfaces from chip to chip, all those things. And those have not seen as much innovation, actually. It's almost hilarious how you'll have this incredibly advanced silicon array that's doing all of the neural net training or the inference execution. but then the back end integration into the rest of the world

is through stuff that hasn't innovated very much and uses a lot of the power, the network cards and that sort of stuff, the data arrays. And that's where I think we're going to see a lot of innovation. And there's also a little bit of a self-limitation where eventually the latency does matter within the array. And so as the arrays get bigger,

you start paying more and more of an energy penalty to keep it connected. So there's going to be some self-limiting thing there. that will then drive innovation in, well, how do we improve it? How do we make it? And any improvement in latency is going to be bringing things closer together, which is going to use less power. So I think as long as the industry stays focused on the overall watts per flop or like equivalent flop, because I don't actually think flops really apply to these training computers, but we'll continue to see power reduction. And then there's this other question of like, okay, they need to make money too.

So like there needs to also be a use case that people are going to pay for this technology. And at some point, the investment, people put flooding all of the investment into AI is going to be like, well, where's the return? And so that, you know, until there's applications where there's real money flowing in the other direction, we may see a little bit of a correction as well. Yeah. I mean, the sort of, is there, are there valuable uses aside? The way that I've been thinking about this is that undeniably there is, you know, like GPUs are fairly new, certainly at the scale that we're deploying them today. and energy has emerged as easily the number one problem. Maybe Nvidia's production capacity is up there too,

but otherwise it's energy. And so the combination of those two things makes me think it's highly likely we're going to see some pretty meaningful energy efficiency improvement. So I would bet almost anything that the flops per watt or watts per flop or whatever you want to use instead of flops will get better. But in some ways it's right now, it's like a race between two things, One is the energy consumption and energy efficiency, and the other is how big a training model can you build? Because every next generation of these models is like an order of magnitude bigger than the previous one.

There's some limit to that, right? But at some point, is somebody going to build a $100 billion model to train GPT 7 or whatever it is? And plausibly, if that is true, then it outruns the efficiency improvement. Anyway, it's hard to predict, but this is a factor that is a confounder in my mind to the current state of electrification. I think it is undeniably gumming up the works right now if you are trying to accelerate electrification of other stuff as quick as you possibly can, particularly industrial stuff. Yeah, there's a lot of things that can be electrified with very minimal additional, central investment. Like a lot of end-you, if you actually look at the typical home built after the 70s, you know, or maybe even the 60s in some cases, there's a 200-amp main panel that barely ever peaks above 50 amps. And that's true not just at the main panel, but also like the local transformers have a lot of thermal margin and up and up and up and up.

And that's why EVs have been able to be installed and charged in lots of places without a lot of regional or local, infrastructure upgrades required. The same is true for heat pumps. There's a lot of end use, let's say, fat to be absorbed without major infrastructure investments. And then distributed energy generation can continue to help with that problem statement by generating power at the end that doesn't need to go through the transmission system. But you're absolutely right when you're looking at these large central plants, and the central loads and the central gens are kind of competing for the same resources.

And that's all the resources. That's the EPC resources that build the interconnection of the equipment is the transformers, the switchboards, the GSUs that, you know, connected to the transmission grid, and the policymakers and the like, you know, managers of these networks that have a lot of studies to go through. And, I mean, yeah, both the load and gen side are kind of bottlenecked to the same pipe at the moment. And we'll see who wins. I mean, ultimately, you need them both to go hand in hand. And then one thing, other thing I was going to say is there's also plenty of opportunity to be thoughtful about how you shape the load curve of these specifically inference engines, but also the training compute, right?

If you co-locate with wind or solar, that's especially solar that's going to be curtailed or seeing lower market prices during the middle of the day, like in California, you know, all of a sudden you have a higher value and you used to send that power if you have that data center behind the meter at a solar facilities, let's say in California. And then as the, if you want to sell the AI product right now, now I can be like, well, the AI product is much cheaper if you do it between 10 and 2. And that's when people are working anyway. So maybe it's actually complementary. So I actually think, again, if people are creative, maybe it's maybe it all works. There's definitely a paradigm shift in data center world that it's hard to tell how quickly it's happening. But obviously, like, data centers, cloud data centers in particular are used to this like 99.999% reliability world.

And the product that the hyperscalers are selling, AWS is offering you exactly that, low latency and extraordinarily high reliability with flat pricing. So like this idea that you're describing requires this paradigm shift, but it may just get forced into existence by virtue of scarcity. I want to talk about the other elephant. So one elephant is like there's all this other demand growth that we have to contend with. The other elephant you're going to predict, which is transmission, if you are trying to build that much new wind and solar, you've got to get it connected to the grid and then it's got to get load.

All this new load has to get connected to. You and I have talked about this a little bit. I think I'm more pessimistic than you are on our ability to, like, dramatically expand our transmission system in the United States. But what gives you confidence, given recent history, wherein we're building less transmission than we have in the past? Like, what makes you think we can turn that around? Well, not all transmission is interstate.

You know, a good amount of transmission is actually within a state, and it's a really nice infrastructure project to get everybody excited about, like building a highway or something like that. And in some states, that will fly, and in other states it won't. It's the interstate stuff where things get a little bit trickier because you have way more state actors involved. I think a really good example of how this can go badly

is the fight over the Colorado River, which has been going on forever and over and ever. But, you know, we're seeing some good things happen, right? FERC just kind of issued their new guidance, which is intended to make these things easier to, The study is easier to complete, kind of reform the process a little bit about how interconnection is supposed to happen, at least some ideas for how to do that. And of course, Fork doesn't directly do it. They have to go through all the isos and the various bodies.

But at least it's good to see some change from the top on the regulatory side, promoting an easier transmission process. I mean, the primary reason why I would say I'm optimistic about it is it doesn't have to. to be one link. Like, there's lots of different links that make a difference in an interconnected system. And sort of similar to the natural gas pipeline or the, you know, the oil pipelines in the United States, like they've kind of built up over time and they just become more and more efficient and effective with every link that gets interconnected. And while there will be plenty of NIMBYs and, you know, local pockets where we will see problems in getting transmission done, there will be others where we do get transmission done.

And the more interconnected it becomes, the more valuable it becomes. It's like the Internet, right? Initially, it was just like Denver to New York or something, and now it's everywhere. And not everybody wants to have the fiber trench in their backyard, but people were creative about finding highways to get it done. And actually, that's something that I haven't seen enough happening. You know, why isn't a transmission line strung over all of the railroad easements in the United States.

Like, we should be doing that. And we should also be doing it over the highways. And if this becomes enough of a hot-button issue, I think the government will do that. And then the last thing I would say is it doesn't even need to be overhead. Everybody's like, oh, it has to be overhead and all this other stuff. But, you know, we buried all of the fiber along the railways. You know, all of the, like, internet is all basically fiber laid along the railroads in the

United States and also some highways. And somebody paid to bury it. And those fiber bundles are not that much smaller or different in size than what you would bury for a high voltage transmission line. So, you know, there's also a bottleneck on the people who can make that insulated cable, but that's, why does it have to be only one company? It doesn't need to be, you know, some innovators can certainly. Yeah, that part is not why, of all my fears about building enough new transmission, companies to build insulated cable is low on my list. I'm just generally an optimist about humanity. If people really want to get out and solve a problem,

creative solutions will come out to solve that. And there's a huge economic incentive. If you start getting into a world where you have highly connected parts of the country where renewables and storage are playing a complementary role from one side of that grid to the other, and the spot price differential in that like the nodal electricity prices in that market versus another one that's 100 miles away

that is poorly connected. You know, the economic incentive is going to be so obvious. And then now you just have to take that economic incentive and figure out how to get all the stakeholders happy through some set of, you know, payments, subsidies or who knows what. And then you're going to close that nodal price gap, you know. I do appreciate your eternal optimism about humanity. And I mean, look, it is one of these situations in which like, From a first principal standpoint, it is the right thing to do.

We should do it. It should make sense economically. I just, maybe I'm too burned by it. But I will say one thing that gives me a little bit of hope in a really micro sense is Michael Skelly, who tried to build clean line or built clean line, the company that was the one sort of like investor-backed independent transmission line developer years ago famously, failed as a company, generally. speaking. However, those lines are largely still getting built, and he's back at it with a new company called Grid United, which is building transmission lines once again. The price signals are there. That's the interesting thing about it, right? Like, at least in the markets that are deregulated enough where you can see the nodal prices, like you can

have three, four, five, six years of clear apparent, you know, gap. And you're like, you and I have talked about this, right? But like my quintessential example is just like, West Texas to East Texas. It's 300 miles or 350, miles and like the prices could not diverge more, basically, seems so obvious. It is in trust state in that case. There are places where this is, this is blindingly obvious. But maybe somebody's maybe somebody's working on that right now. Who knows? I'm sure somebody's working on that right now. All right. So let's, let's posit that everything else in this scenario that we're describing is plausible. We, we electrify all sorts of things. We build enough genera. We build enough

generation, clean generation to meet that demand, and we're able to connect those two in real time, and we've got a system that works. So then this gets to the other question and the thing that people often complain about if they're thinking about these kind of like deep decarbonization scenarios, which is material requirements. So one thing I liked about Master Plan 3 is that it runs through basically all materials you could possibly need for all this stuff, from like concrete to chromium. So high-level conclusion, unsurprisingly, because I suspect you wouldn't have published it otherwise, is there's enough of everything. But what, if anything, gives you pause?

Like, as you look at the material requirement question, where do you think we actually have any degree of a bottleneck? Well, yeah, it's not going to be, are the resources in the ground? It's going to be, do the geopolitics and the permitting authorities that be mean that those resources are rendered effectively inaccessible, even though they practically should be accessible. That's probably my biggest pause. And so maybe that will be solved through trade agreements or rationalization of resource policy in certain developed economies. That's probably the thing I'm most worried about.

There's a lot of people that just do the straight math and they're like, well, look at all the neodymium in every magnet and like all those magnets, we got to multiply that by a billion or trillion or whatever and there's nowhere near enough neodymium. But the problem with that math is that people are using neodymium because the pricing signals they see in the marketplace make it seem like the best magnet to use. But actually, magnet materials, for example, are incredibly substitutable. And if you think of the design space as not just the magnet, but the magnet plus the, you know, electromagnetic system it's inside of with the steel and the geometry of the rotor and the stator and the whole motor and actually maybe even the power electronics and the mechanical advantage garing system it's attached to, you can dramatically change what magnet material you're using and still have and still achieve the mission objective. So you can't just do the simple math and say there's not going to be enough of something. And in fact, we, you know, we took advantage of that and said, well, all of these things are substitutable when coming up with our resource requirements. And that applies not just to like the motors in cars, but the motors in heat pumps, the motors in wind turbines, the motors and everything.

And something similar like that applies to all of the resources that are in use to make this happen. Like there are a lot of substitutes. And one of the fun things that we put in the paper was a comparison of how much material humans just move out of the ground every year in total. and the amount of material required for just this renewable energy economy. And it's actually like a factor of 10. There's almost really no comparison. We move so much material to do all the activities that we do just to build buildings, agriculture, paper, all of the raw material use that we have.

The small amount of raw materials that are going to go into this renewable energy economy just almost don't even matter in the scheme of things. And we also move a lot of liquids out of the ground. ground right now in the form of hydrocarbon. So when you compare the hydrocarbon movement to the movement required for the renewable energy economy, it's also like very favorable comparison. And then the last thing I would say about the raw material in general is it's going to be recycled, at least in the batteries. It will be recycled. And I think we're going to see it being recycled in a lot of different areas. I've seen, I've seen a rare earth recycling happen on the magnet side.

People are working on that. You know, at some point, the solar panel recycling industry is going be formed because solar panels have a finite life and we'll need to be refreshed. And that's in the paper as well, the fact that they need to be refreshed. So yeah, these materials, once they're kind of deployed, they will be redeployed in some fashion with not perfect recovery, but to the point where the ongoing resource requirement is just is not that challenging. Now, one question would be, will all of these recycling operations be techno economically, you know, like is the economics there? And that is as much a technology question as a logistics question, because, you know, you basically have to take this thing that would otherwise be landfilled and put it together in a concentrated enough forms that the logistics is meaningful. And that may be its own challenge, but actually, you know, people.

People have been doing it with scrap cars for years, and the total scrap recovery of a car is in the hundreds of dollars. But cars go to scrap guards and people strip them down. So I think we'll see all of these things being recycled. You mentioned, like, the thing that gives you pause is not, is there enough of this stuff in the ground, but sort of the combination of geopolitics and permitting and all that. I'm just curious for you to speak to because you were overseeing Tesla's lithium refinery

that Tesla is building. Like, that's a specific case where, you know, The big question with lithium is not, is there enough in the ground? At least currently, it's where and how is it going to be refined. And currently, that's China for the most part. So what learnings have you taken from that as to the question of the big one to me, which is refining and processing of all the minerals, which needs to largely get shifted out of China in pretty much every case?

Yes. I think it's really coming down to capital projects execution. and where is the excellence in capital project execution right now? And it is in China. They're investing billions, I mean, probably trillions, in capital projects across all aspects of the sort of supply chain. And for that reason, they're just really good at building any kind of capital project. It doesn't matter whether it's a chemical plant or a industrial facility or manufacturing facility or power plants or anything.

And so how do we kind of bring. that back to other countries in the developed world and countries in the developing world. And I think there's actually a lot of opportunity here because an ecosystem needs to be created around the engineering, procurement, and construction of these large capital projects. And that industry needs to be competitive. And I think there's definitely opportunities for people to start new companies in the United States and other developed countries where there hasn't been a lot of capital project construction over the past many decades to just build like a ruthlessly competitive execution company to go

and get a lot of this stuff done. But yeah, the stark comparison between going and trying to get a project executed. I mean, I saw that within Tesla, not even on lithium refineries, just looking at factories. Just like getting them built in China versus getting them built in Germany or the United States, the ecosystem just isn't really there of talented consultants, contractors and things to work with. And that slows the projects down. Now, like, things like the Chips Act and other stuff that the U.S. industrial policy that's really driving investment in large capital projects will build that ecosystem back, and then it will be easier to get these things done. Because, you know, it's not like the refineries are in China because the technology

to refine lithium is in China. Like, that's definitely not the case at all. It's purely been, like, the capital investment to build the refinery is the lowest in China. And it's not labor either, because labor is not beneficial, because there's, like, no light labor in these refineries.

It is sort of permitting, though. It's sort of permitting. It's not logistics either. Like, logistics are worse. You're taking lithium from Australia, sending them to China, and then from China. There's nothing other than capital project execution,

which certainly relates to permitting. But let me take permitting as an example. So we built the lithium refinery in a part of Texas where it's a permit by rule. So the permitting authority basically, I don't want to say trust, but it is effectively, puts the liability of the construction on the engineers stamping the drawings. The engineers stamp the drawings. If the engineers sign off, the jurisdiction's like, okay, go build it, right? And of course, there's some air permits and other things that they actually will do the verification on.

But they're not doing the structural verification. They're not doing other types of verification that in most jurisdictions they are doing. And so there's actually a permit checker that's going to read through your whole plan set, ask a ton of questions because they feel the liability for your building falling down or lighting on fire or whatever. And that is very unique in the U.S. and areas of Europe that is not true in China and other places. So I do think coming up with almost like regions or zones within these developed countries where the liability is very clearly on the engineers of record and permitting can therefore go faster rather than basically having to take all the work that the engineering firm has done. and convince a frequently like unsophisticated, unskilled in the art of a chemical, a lithium refinery had never been built in the United States of this size.

And if we were to go to a jurisdiction like Austin or Fremont, California, you would spend a lot of time with third-party consultants being employed by the government bringing them up to speed on like the hazard, the has-ops analysis. And because they need to be independently convinced. So I think there's some opportunities to stream. streamline these things, with clear liability being placed on the right accountable parties that could make a big difference. Yeah, that all makes sense to me.

I guess I think about permitting in two ways, though. One is like, how much time does it take to get a permit and do we have the capacity to offer it? And the other is actually, like, what is allowed, what is permissible in a given location? And there are definitely cases where in the process, for example, like if you want to build a new copper smelter, you can build it in China because it is permissible. You cannot build in the United States because it is unpermissible. I don't think it's just a function of how we do permitting.

We don't allow the particular emissions or whatever it's going to be from the process here. And so I've always wondered whether maybe the solution to this, I don't know if this would actually be helpful broadly, but the solution to this is some version of, instead of like a carbon border adjustment tax or something like that, there's like a, I don't know what's called clean air border adjustment tax. Like if you're going to force us to do all of our copper smelting in China because they'll allow the particulate emissions there and we won't, then we should have to pay more for that here if we're going to try to import it. Just to incentivize, like, let's figure out a way to do it here.

Yeah, I think that's an interesting idea. I think the other challenge with permitting that you're not bringing up, but I think is also there is that it changes all the time. and just the fear of regulatory uncertainty tends to delay projects. So coming up with some sort of grandfathering clauses or, you know, more extended building code or other code windows where, you know, your project started being developed against a 2021 building code, as long as you get it built by 2031, you can build it.

These sorts of minor changes could make a lot of, could help a lot. And then one other thing I would say is, you know, our lithium refinery in Texas, we actually are doing a different process specifically to kind of help with some of the permitting aspects like you've described. That's sort of my point, right? Like incentivize that stuff via making it difficult to import the dirty stuff. Yeah.

And for sure you can do that. All the precursor almost, I want to say all, over 90% of the precursor in the world is, for battery cathodes is produced in facilities where the wastewater, the like sulfates that are in the wastewater of that process kind of can just go direct. into the waterways because the local jurisdictions are okay with that. And that's also why most of the soap manufacturers and things like this are located on the oceans because they're also just kind of dumping. Now, it's not like the sulfates are not, they're not toxic and it's not like those companies are destroying the oceans. I don't want to make people think that. But the way the Clean

Water Act is written, they're out of concentration that is above some total minimum daily level, and so you can't do it. So you're effectively not allowed to do it in the U.S. So you have to come up with ways to do precursor that don't have any. They basically are zero-discharge wastewater. Now, there are technologies out there that can do that. They have to go through the commercialization pathway, and some of them actually have lower total cost once they achieve their end objective,

and Tesla invested in a couple of those, and there's other companies that have tried it. So I think, again, if the rules just stay the same, then innovation will happen and companies can go out and solve these problems and go through the technology development cycle to get them done. I also was involved in building a battery factory in Fremont, California. We built the megapack and lay through up. You can actually build things kind of anywhere, even in California, if you put the time into understanding how to comply with the rules, and the rules are the Clean Water Act, the Clean Air Act, and the building codes. And those, other than the building codes, those other areas haven't really changed very much

the Air Act and the Water Act, and it's actually the building codes that change more than frequently than anything else. I will say right now we have a bunch of portfolio companies that are like building first of a kind things somewhere and many of them are based in California, so it would make a lot of sense for him to put it in California. I would say that the limiting factor

that I've seen over and over again these days is not as much permitting as it is the cost of power, which is becoming insanely high in California, and that ends up pushing people out of state more than anything else? Yeah, yeah.

And that came back to our supply-demand-balance problem. Yeah, exactly, right? If that gets a lot worse, it's tough. All right, so you have all the power that you want with Master Plan 3. Like, what do you want it to accomplish for the world, obviously? Like, it's Tesla's Master Plan, but, like, what's the intent from your perspective? Like, what needs to happen tomorrow? Yeah.

I would hope that the takeaway is that we should redirect the resources that are going into, let's say, fighting sustainable energy technologies into finding even better sustainable energy pathways. I think the point of putting together all of the arguments in this paper was to say there is a feasible path, and that feasible path actually looks pretty attractive when you look at investment per year, resource use. total electricity production. I mean, one of the interesting stats in the paper, which honestly is almost staggering to me that this is the case. But 1.5 terawatts is the claimed, you know, the paper claims that 1.5 terawatts is the total amount of renewable energy capacity that will need to be deployed on an annual basis

to maintain the sustainable energy economy. And that's basically keeping up with plant retirements. So on a steady-state basis, 1.5 terawatts is how much you need to deploy. Now, last year, the world globally deployed almost 500 gigawatts, which is unbelievable. Yeah. You know how much of that was like solar in China, though? I don't remember the exact number. It's an unbelievable amount.

Yeah. Oh, no, it is. I mean, it's because their economy is still growing rapidly and they're, they've had this central decision-making pathway of like, we will go renewable, right? they can just command, it shall be so. But yeah, 500 gigawatts, right? And that's only one, that's only, that's, you know, much closer than an order of magnitude away from where we need to be. You know, you only need a 3x from here, and the growth rate has just been staggering,

you know, more than 100 gigawatts a year of growth. Yeah, we might have a terawatt year this decade. It's not crazy. Wild. So, that, that, you know, putting these numbers together in these terms, you know, that there's, I'm just going to read them because the numbers are simple to read, right? It's something like 240 terawatt hours of storage, 300 terawatts of renewable power, you know, 1.5 terawatts a year, that's simple math, 20-year project lifetime.

You know, $10 trillion of manufacturing investment, one-half the energy required, less than 2 tenths of a percent of the land area required, 10 percent of the 2022 world GDP in investment, total investment, and the resources are there. Those are the numbers. We just wanted to put those numbers out there so that, again, people kind of redirect their brain space from fighting this concept to finding the best way to enable it. And again, I'm not stating that the master plan part three is the best way. It is a way. There are many ways to do it.

And the carrots that are shown, like it's lower total investment, the resources required, all of these things are care. And, of course, you have less air pollutants, climate change. goes away, or at least is moderated, you know, there's so many reasons to do it. And now we should just all collaborate on finding more paths, not just this path, but other paths, rather than continue to fight this concept. I mean, that's probably the primary point of the paper, is to get people excited about working together to find the best path for it. Well, Drew, this was exactly as much fun as I was hoping it would be. I appreciate you taking the time and glad we got an opportunity, finally a year plus after Master Plan Part 3 was actually published to talk about it.

Well, thanks, Shail, for having me on. I also really enjoyed it. Thanks again. Drew Baglino was most recently the Senior Vice President of Power Train and Energy at Tesla until April of this year. This show is a production of Latitude Media. You can head over to Latitude Media.com for links to today's topics. Latitude is supported by Prelude Ventures. Prelude Backs Visionaries, Accelerating Climate Innovation that will reshape global economy for the betterment of people and planet. Learn more at preludeventures.com. This episode was produced by Daniel Waldorf, mixing by Roy Campanella and Sean Marquan, theme song by Sean Markwan. Stephen Lacey is our executive editor. I'm Shail Khan, and this is Catalyst.