200GW Hiding in Grid, Sodium Batteries 10x Cheaper, Wave-Powered Datacenters w/ Ramez Naam | EP #280

In the US, if you put in a request for hundreds of megawatts of power to build a data center today, good luck getting that power before 2031. That's the situation that we have today. So new technologies like sodium-ion batteries could drop the cost of batteries by a factor of 10. We already have the very first solar-plus-battery baseload power plants, and they're affordable. Batteries are plunging in cost and are going to drop another 10x ultimately.

Do you believe that thesis that solar in the long term is going to dominate beyond everything else?

The reality is, look, everybody's heard Elon talking about space-based data centers. So 10 gigawatts a year is like 5 or 6 launches of Starship a day.

Unless we hit that exponential absolutely full on and go right down that path, This looks prohibitive for 15, 20 years.

The biggest unlock that we cannot predict of AI and power will be— Now that's a moonshot, ladies and gentlemen.

Everybody, welcome to Moonshots, another episode on the frontline of the singularity. We're living during the most extraordinary time ever, and our mission here is to deliver you the breaking news, And help you understand what's going on. Today, we're gonna be doing a deep dive into the innermost loop, all things energy. I'm here with my extraordinary Moonshot mates, DB2, AWG, Salim. Welcome, gentlemen. Good to see you all. Looks like you're in your normal haunts. Good to be here. And we've got a friend with us today. Everybody on Moonshots, it is an honor and a pleasure for us to invite Ramez Naam. Ramez is a computer scientist, investor, author, one of the clearest thinkers on the future of energy. After a career at Microsoft, Ramez became the leading voice in the exponential decline in the cost of solar, batteries, fission, fusion. He's the founder and managing partner of Planetary VC, investing in energy companies. He's the author of The Infinite Resource and one of my favorite ever science fiction novel series, The Nexus Trilogy. If you've not read Nexus, I cannot commend it. On Book Corner, where Alex and I talk about our books, I've mentioned Nexus a few times. Today we're gonna be exploring The Innermost Loop, why energy abundance may arrive faster than most forecast, and its impact on AI, economic growth, geopolitics, and our future. Again, my mission here is at the end of this podcast and the brilliant dialogue that my Moonshot mates is gonna bring to the table here, you're gonna understand either if you're an investor, If you're a builder, what's the alpha? Where is it going? What are the real timelines for everything from building out nuclear plants to fusion plants? Because, you know, sometimes there's hype, sometimes there is an overwhelming abundance of energy coming our way. So first off, Ramez, welcome, pal.

Peter, it's an honor to be here. Great to be here with friends. Looking forward to it.

You do have friends here. So, you know, I have a quick story. Yeah, of course you have a story.

I'm worried now.

I remember we were presenting to one of the top oil companies and energy companies in the world, like top 3 or 4. And they were like, well, who's this Ramez fellow? We want to grill him before we let him in front of the key people here. We're like, fine, grill Ramez. And so, because we were talking a lot about solar and they're an oil and gas company. And after like 2 hours, they're like, okay, we need to get in front of them.

That was it.

That was an awesome session.

Yeah, one thing I failed to mention is Ramez was part of our founding faculty at Singularity University, really led the whole energy conversation there and has been on stage at the Abundance Summit a number of times. Hopefully you're back again coming in 2027. Ramez, first of all, I just need to chide you. You need to write a 4th, 5th, and 6th in the Nexus Trilogy.

As soon as AI and energy get less exciting every single week, I will make time to write another novel.

How much time do you spend tracking what's going on on the innermost loop here?

It's every day, all day. I mean, that's what we all do, right? Living in the singularity.

Yeah, we— this is living the singularity. Before we get jumping in, Alex, you want to add anything to the conversation up front?

I'll just add, welcome to the Terrordome. One of my favorite of your, I would say, popularizations, that now infamous chart of the price of solar going down to zero.

Thank you, Alex.

Amazing. Well, I can't only take another second away from you, pal. Jump on in and we'll grill you along the way. Make the points, you know, shall we say, in a stellar fashion.

Okay, great. Let's just start. We're going to hit a few different topics here at the intersection of electricity, really energy and compute. I mean, a few years ago as an investor in clean energy, that was sort of a fringe sector to some people, though it's $3 trillion. But now that we see that AI depends upon electricity, it is everything. Like the, you know, capital flows, value flows to that which is scarce. And right now power is scarce. So we're going to hit 6 topics. At the end of each, we're going to pause to have discussion. So number 1, AI is power hungry. 2, speed to power and the grid. That's everything. It's not cost. 3, behind the meter power. That's how it's happening. for making the grid better is totally undervalued. And that's where the near-term wins are. 5, solar. 6, fission and fusion. Lots of stuff happening. And 7, finally, the out-of-this-world ideas— launching compute into space or launching it into the oceans. So let's just cement ourselves on AI is power hungry. You have to exponentially increase compute to get linear gains in AI. There are some ways to cheat that curve, which we're doing. That's the basic phenomenon here. And I think we don't fully grok— most people anyway— the relationship between these things. First, I want to be clear that power is cheap compared to GPUs. So if you look at building a gigawatt data center, you're going to spend $50 billion, $35 billion of that for chips. When you compare the ratio of like the all-up CapEx of your data center to your 5-year energy cost, It is amazing how little energy costs. So when you say AI is power hungry, it's not really a cost issue. It is that energy is the bottleneck for AI. And this has a lot of ramifications because these numbers are in billions or tens of billions of dollars. Every hyperscaler has whole teams devoted to optimizing the cost of energy. But if you tell OpenAI or Anthropic today, look, we can give you power at twice the cost that's on tomorrow. They'll take it. They won't tell you that, of course, but they will take it because the revenue you can generate from a unit of electricity to the cost of it is basically the same ratio as this. So what's the challenge? The challenge is we stopped being able to build out the grid fast. And I'm not talking about power generation. We can still do that pretty fast, at least for solar, wind, batteries, natural gas. But the poles and wires are a huge problem. So this is— we talk about the interconnection queue, which is the queue to get your new project hooked up to the grid. And this is for the generation side. If you're building a new solar plant, wind plant, natural gas plant, how long does it take before you are hooked up to the grid so you can deliver power to your customers? That's gone from 15 months 20 years ago to now coming up on 45 months.

Regulations? What is it?

It's regulation. And it's also that as demand has— demand growth has slowed, right? The US demand growth per year is much slower than it was the '80s, even lower than the '50s, utilities have just re-engineered themselves. They are more oriented on customer service, on meeting the regulator's demands, and so on than they are on building stuff fast. So that has gotten in the way. But permitting is also a huge issue, not utility regulation per se, but permitting issues for the land controlled by the state, the county, the feds.

Ramazan, if we're just going to jump in, led by Peter's example, I have to ask, you sort of flew by, you mentioned or you alluded to this notion that intelligence was somehow proportional to log compute, which I know a number of executives have also pushed the narrative of maybe you could, one could naively extrapolate some law that looked like that from scaling laws. In machine learning training or machine learning inference. Do you think that's actually true? And if you do think it's true, do you think it continues to be true in an epoch of recursive self-improvement?

It's an awesome question, Alex. And this is like core to the big questions of AI and are we going to have recursive self-improvement to ASI? Look, everything in machine learning since 2000 has shown something like a log-linear relationship between, really between training data size and precision of the model.

Right.

My actual model—

You're alluding, I think, to first Kaplan scaling and then chinchilla scaling and then post-chinchilla scaling.

Long before chinchilla scaling with single-layer neural nets, we were finding this in the early 2000s.

Right.

So compute is used to convert training data into a model.

Right.

Into a neural network. And that has a roughly log-linear relationship. But we cheat, by which I mean we keep finding ways to make that more efficient. So is it actually log-linear? No, it's a little bit faster than that because we keep finding ways, as we see with DeepSeek Flash that just came out, as we see with KIMI-K3, we keep finding ways to bend that curve. So it's a little bit less bad. than log-linear. But it's still somewhere between a power law, like n to the 5th, and a true exponential or log-scale difficulty. And I don't see that changing anytime soon.

It seems almost— if I understand your broader thesis, this seems almost axiomatic, that if we can't bend the log curve, that we need basically exponentially larger amounts of energy just to make essentially linear or polynomial progress in intelligence. need more and more energy. We need to achieve Kardashev Level 2 or Kardashev Level 3 type of civilization, Dyson swarms, in order just to keep making incremental progress.

How quickly do you jump past the—

Yeah, it's gonna be a record.

Alex is getting at the core issue with superintelligence, actually, in a certain sense. Look, like, here's my view, Alex. Like, the naive view is at any given time, intelligence is basically log-linear with compute, log-linear with data. But we keep making the algorithms better, and that sneaks us towards like a polynomial domain, but the polynomial domain is still steeply diminishing returns.

You're arguing it's polylog. You're arguing that intelligence is polylog in compute.

At best, it's polynomial. And I don't— not necessarily polylog, but at best, I mean, we see like the very best examples you can get is maybe compute has to go up, you know, n to the 4th to get an n-size increase in intelligence.

And you don't think recursive self-improvement, if we are indeed in an era of RSI, you don't think that's Anything better than Polylog?

No, look, you do the math on RSI, and RSI, every way that you improve AI, has diminishing returns. So every model of RSI that does not include hardware, we can save that. Every model of RSI in software fizzles over time. Now, the bump might be so big that we're like, wow, this is just over-the-top amazing. But it always looks concave. There is no mathematical model of RSI that's valid That I can see that leads to an actual, like, you know, vertical asymptote to take off.

This episode is sponsored by Google for Startups. Think about this for a second. You now have access to the same generative AI models that cost hundreds of millions of dollars to train. Google's Startup Technical Guide for Generative Media gives you a complete blueprint for deploying Google DeepMind's models in production. Images, video, audio, all of it. Real architecture, real results. Find the link in the show notes below. I'm going to take us back. So the grid is the grid. The grid is the bottleneck right now.

Yes. Yes, it is. So we need power in a practical sense. Look, everybody is compute.

Not that I did not appreciate your genius in those questions. And that was fun. And this may turn out to be an entire conversation between Ramez and Alex, but we'll see.

Let's have another episode where we just talk about that.

I got words.

I got words.

I think, I think, I think the point here is the grid is the bottleneck is a really important point because it speaks to the infrastructure needs we're going to have to have for dealing with this. So let's move on and we'll get back to it.

And look, so this is for the power side. For the demand side, we don't have data that is as clean. But here are like 3 locations around the US. And you see in this 7-month period between April and November last year, The wait times to get connected for the load side, not for generation, but for your data center or whatnot, went up by 6, 7 months. So everywhere around the country, as demand is going up for large load interconnection, you're seeing longer and longer waits. So this is Texas. ERCOT is the Texas grid. ERCOT currently peaks out at about 80 gigawatts. Okay. They have submissions into their demand-side queue for more— this is slightly out of date— for well over 200 gigawatts of load. Most of this is speculative. Most of these submissions are BS. Anybody— not quite anybody— you can put in a request for a large load and power without it actually having financing or a customer or so on. So most of these things evaporate. But in any case, The Texas grid operator is overwhelmed with these requests for power. And of course, that just jams up everything.

Wait, let me, let me ask a quick question. Are we talking about I have a data center, I want to connect it to the grid to get power, or I have a new power source, I want to connect it to the grid to deliver power? Or is it about the same either way?

So this, this chart is generation. So I've got a new power source and these 2 Our demand, our load interconnection queues. So both of them are going up. I mean, honestly, it's much longer than that. Like today in ERCOT in Texas, the like most advanced, progressive, least— Yeah, most progressive in like a positive sense, like least regulatorily burdened, fastest moving grid in the US. If you put in a request for hundreds of megawatts of power to build a data center today, good luck getting that power before 2031, 2032. That's the situation that we have today.

Yeah.

When you did, you did a presentation for my Abundance community on our monthly meetup, and that was my major takeaway, that the issue on energy for AI isn't, you know, building out solar farms. It isn't fission or fusion. It's the grid. Grids.

That's right.

So in that case, it is the grid, as you know, and there are people who are watching who are investors want to understand this. You know, we talk about infrastructure, picks and shovels for AI, and we talk about, you know, data center construction companies and all of that. Who are the companies that are building out the grid? And is there, is there sort of work orders and purchase orders for building out a more robust grid?

It's a really good question. So the grid, the poles and wires, the distribution grid in particular, is dominated by regulatory monopolies. Right. So the local utilities— I'm not going to comment on their current P/Es, whether I think those stocks are buys or sells, but the regional monopoly utilities stand to make a huge amount from this in the areas where data centers can be built. There's a separate issue. It's not in my slides of more and more voters are pushing back and saying we want to stop data centers being built. There's a lot of psychology behind that. I don't think the, the reasons are necessarily that valid. Even in Texas yesterday, Governor Abbott sent out a letter pausing— it was the day before— pausing.

Oh no, not them too.

It was actually— it wasn't quite a pause. It's an audit of all data center requests in Texas. In Texas, a red state, the most libertarian state in the country. And it's political cover. Abbott knows that his voters are like— there's an anti-tech sentiment that translates to AI data centers because they're an obvious target. So he wants those data centers built. There's an election coming up. He's got to cover his ass for a bit by making it look like he's serious about this. But that's the politics right now in the country.

So basically we're doomed.

I don't think we're doomed. Just to jump to the— we'll always have space.

We'll always have We will have sun-synchronous orbit.

You know, the AI doomers would say, thank God we're saved, the AI god won't be built.

But even if not, they'd propose orbital bombardment of the data centers.

There's no way of winning. Let me— the grid, look, I'm going to show you a lot of sci-fi stuff and awesome stuff. But yeah, Peter, what you're saying, the grid itself, the poles and wires are the limit. And I've talked for years about The exponentials in solar batteries, we'll talk about fission and fusion, but the poles and wires have thus far not become an exponential technology. And that's something I would love to solve. I have not seen a lot of startups show up to do this.

So, Mez, aren't we, aren't we moving the data centers to where the energy is? So you don't need to set up, you know—

Or disconnecting them from the grid entirely.

Maybe.

Exactly.

Let's move. Let's move on and I'll get to some of that. This is like a more practical forecast. You see, even like 2028, we'll build— this is probably a little bit low. The orange is like how much we'll build. Maybe it'll be 20, 30 gigawatts, whereas the demand could be much higher. This is an interesting slice, by the way. I'll tell you, every forecaster, Morgan Stanley, whoever, they all differ somewhat. But this is an interesting slice. The blue bar is if you just sum up all the GPU manufacturing scheduled between now and 2030, primarily Nvidia but also AMD, Cerebras, uh, whoever, versus the expected pace of US grid buildout, the chips are more than twice the pace in their power draw as the power we can deliver.

For folks listening, the AI demand, uh, which is chip limited, shows roughly, you know, 200 to 250 275, call it 230 gigawatts of power demand based on the chips. Like, you bought the chips, you've installed the chips, can you power the chips? There's 230 gigawatts of demand there, and U.S. grid buildout is projected at roughly 100 gigawatts.

Yeah. And some of you might remember, like, 6 months ago, Satya Nadella, CEO of Microsoft, made this comment: look, man, warm shells are our limit. We've bought the chips. We don't have warm shells to put them in, right? That is the limit for everyone at this moment.

Why is that discrepancy there? Because, you know, Eric Schmidt told us his, his number was 100 gigawatts or 96 gigawatts of additional power by 2030. The 230 is just based on chip manufacturing. So either more of the chips are being kept domestic, which wouldn't surprise me, or the fabs ramped up, which would surprise me. But where's that discrepancy come from?

Every single forecaster has a different number. And I think they— some of them base on just announcements by companies, whether they're chip fabs or utilities, some of them based on their discounted projections of what they can actually achieve. Also, I'll say that there's a big miss in power demand of chips. A lot of people just say How much power can my Blackwell GPU draw? Multiply that by how many you're going to build, and that's the power demand. Now you're missing like half the power because you've got to add the draw of the rest of the IT equipment in the data center and cooling and so on. And that nearly doubles the total power use.

You know, that would make sense. Those Cerebras chips, they, they run— they just suck down power and they run the transistors much more efficiently than the prior generation kind of A100, H100s from Nvidia. So the transistors are actually doing a lot more work, which is better fundamentally. But yeah, of course that's going to draw more power constantly. And of course, when you buy those things and deploy them, you run them 24/7.

That's right.

You're never going to let those things rest. So that may be a discrepancy too.

Absolutely.

Do you think this creates a forcing function perhaps for Nvidia or the other fabless vendors or the fabs like TSMC to get into the power generation business right now? The power gen that's, that's powering all of these chips that Satya talks about just collecting dust in warehouses because he can't find warm frames for them in data centers. Why not? Why do you think that there's a forcing function for the NVIDIAs of the world to get into power gen?

Well, I'd say, look, whether NVIDIA wants to get into it or not, and NVIDIA has made some interesting investments that I'll talk about in grid flexibility. The reality is that, you know, what people talk about the most now is behind-the-meter power gen for data centers. And what they mean by that is large natural gas turbines, if they can get them. This is a multi-hundred-megawatt, like, say, a 400-megawatt natural gas turbine, the kind you'd use on the grid. These are now sold out for something like 7 years. GE, Hitachi, and so on are building new assembly lines to try to bring those online faster. But everyone is saying, look, if the grid is going to make me wait years and years and years, I'm just going to build my own power.

Right.

Now, this, this is more expensive than the grid, but power is such a small fraction of AI cost. Maybe you can do it because these guys are sold out. People are going to these small turbines, solar turbines, Nothing to do with solar, but they make— this is a 38-megawatt turbine that's on the back of a semi. So 40 of those make a gigawatt. Right. Even these have backlogs. But now you have companies, everyone in the world that was in any way proximate to gas turbines is pivoting into this space. I'll give you an example. Bloom Supersonic. Very cool company trying to make supersonic jetliners a thing again. That's a very hard task with many, many billions of dollars of regulatory costs. They have pivoted into using their engine design to make a gas— natural gas turbine for data center power because the demand for this is so very high.

So modular, modular energy production. Yes.

Right?

How many of these, if you think of 'em as an 18-wheeler truck that has a large container on the back, just pull 'em in and get your data center started until you build out energy infrastructure and then move them on?

This is how Elon got the Colossus data centers up that Anthropic is now leasing, actually. This is what he did.

And we've talked a bit about this on the pod in the past. We talked about the boom pivot, and we've talked a bit about Elon standing up his fume-generating cogeneration facility at Colossus, et cetera.

Yeah.

We talked a bit about that, but I just wanna try pressing once more on this point. If this thesis is true that this is a primary overhang on Nvidia's ability to sell more GPUs, Nvidia's already doing all sorts of financial engineering to be able to sell more and more GPUs by— through customer financing, all of these other things. Why on earth, if the energy overhang or underhang, depending on your perspective, is a major limiting factor for the ability to productively monetize GPUs, Why don't we see Nvidia doing something on the energy front?

It's a great question. So look, for behind the meter, the financial incentives are so large Nvidia doesn't have to, but they might invest in some of these companies. On the grid side, Nvidia has made investments into increasing grid flexibility to be able to get more juice out of the current grid. Emerald AI is one example. They've made a few investments in this space, and I'll talk about grid flexibility in a sec here. But the real issue is a combination of regulatory and the incentives for utilities. Utilities, monopoly utilities in the US, the bulk of them are paid on cost plus. So they say they go to the utility commission and they say, I've got a plan to meet the demand that I see my customers having. Here's what it costs for me. And I expect a 10% return on capital. For it. And the utility commission mostly just says, okay, like some are better than others. But let's be honest, like the utility has enormously more horsepower in people, compute, salaries, etc., than the utility commission. So they like jam through this plan and they get 10% on top.

I see. So if I were to try to synthesize what I think your answer is, your answer for why Nvidia isn't getting into bundling power gen with their GPUs is it's low margin and frictionful, like For the same reason Nvidia tried and failed and then retreated to launch their own hyperscaler or NeoCloud. It's just not as high margin as selling GPUs.

You know, Nvidia might still be a NeoCloud. We can talk about that separately. If I was Nvidia, I would be focused on changing the regulatory landscape for monopoly utilities. And I've said this on like some utility-specific podcasts, we should change the incentives. Utilities, instead of just getting paid for a percentage over CapEx. They should be paid on things like how fast they can deliver power. Their executives should get bonuses for delivering power fast. And if you did that, and their employees, obviously all the way down, if you did that, suddenly these things would happen faster, right?

You get what you incentivize. Absolutely.

Yeah.

Look, I invest in startups. How many startups have I seen that have a technology to speed up building poles and wires? I don't know, 2 or 3. None that I thought were amazing. If you, by the way, listeners, if you have one, please send it to me. Why not? Because there's no incentive for it. But if you created the incentive, people would find technical solutions to speed that process up.

Is there any state? No, not Texas, but is there any other state that is open-minded about that?

And Texas, Texas is the best. And I will say, despite what I just showed you, Texas has made policy changes that accelerate this. And, you know, people are not totally asleep at the wheel.

FERC—

so Texas is interesting. Texas ERCOT is its own fiefdom that is not regulated by the feds at all. FERC regulates the rest of the country's electricity. FERC has sent letters to the 6 other largest grids saying basically do something like what Texas is doing. And what we're What they're doing is, and maybe I can just skip to it, is making new regulations that say if you are an interruptible load, if you are flexible, if you can either find some alternate way to power yourself or just turn down your power at moments of peak demand, we'll get you connected much, much faster. So in Texas, that's a CLR or a PCLR, an interruptible load. And the reason for that is we, we have as Americans, as anyone, we have very high demands for the reliability of our grid. Right. 99.9% uptime is 8 hours of outages per year. That's unacceptable. Right. It's got to— you got to push to 4 nines. To make it a grid that you think is really good. But the nature of the grid is the power demand is not constant. It fluctuates through the course of the day and the seasons. It peaks primarily in the South in late summer afternoons. So this is the US grid. The US grid averages about 500 gigawatts of demand kind of throughout the year, throughout the day. It's It's much more volatile than that. But at any given time in like winter nighttimes, the US grid is down to like 400 gigawatts of power being drawn. In summer late afternoon, we're up to like 600 gigawatts of power being drawn because of AC primarily. The fluctuation is actually much higher than this.

Europe doesn't have this problem.

I mean, it's a joke. It's a different problem. We can talk about Europe and AC. There's some amazing tech coming out of the on that, by the way. Hopefully a new investment. That gap is 200 gigawatts, right? 200 gigawatts is about $10 trillion in AI CapEx. We think there's about $7 trillion in AI CapEx in the next 5 years. Like, this is no joke. If we just use the poles and wires more efficiently, we could power up a lot of stuff because we're not short on generation. We're not short on power plants. We are short on capacity in the poles and wires. Okay. So what are we doing there? As I mentioned, like Texas has, you know, just June enacted this new regulatory change that says, look, if you, you know, don't need to draw power at peak, we'll just hook you up fast. Instead of 5 or 7 years, it might be 12 to 18 months. FERC has now told everybody else to do that. So how do you do that? This is a paper by a buddy of mine, Tyler Norris. He's now at Google. He was not when he wrote this. This came out in January, February this year. This is the best electricity-related paper of the year in my mind. And basically what he found was I call it 200 by 200 or 100 by 100 at minimum. If you can be flexible 100 hours out of the year, 4 days out of the year, 1% downtime, that unlocks 100 gigawatts of capacity on the grid, which is about $5 trillion in data center CapEx, including the chips, which gets you through the next few years. That's one way to do it is just flexibility. The startup I mentioned, Emerald AI, Varun Sivaram, funded by Nvidia. They do this via just software orchestration, moving jobs to the right data center, etc., etc., etc. But there's another way to do this, which is batteries. Yeah, this is a portfolio company of mine. I've made 3 investments in the same startup, maybe a 4th one coming up. They do something really obvious in a place like Dallas-Fort Worth. Between middle of the night and late afternoon, there's like 10, 15 gigawatts of flex in the grid demand. So if you build out, let's say, 4 hours of battery storage at the site, fill it up at midnight, you don't need to hit it during the peak of the day. And that fits perfectly with the new Texas regulations. In fact, they were leaders in driving this. This currently sounds— what's obvious to us, right? But this is an unusual approach. 12 months from now, this will be a super common approach, not just in Texas.

Time-shifting load, right?

Yeah, exactly. So right now—

Wait, Miz, Miz, but if I have a magical technology that stores insane amounts of energy very cheaply and I go to even Texas and I say, hey, this can completely shift this curve, this is a total game changer, can I hook it up to the grid? and start sucking down power when no one's using it in the middle of the night, would they still say, yeah, you can do that in 2030?

So the new regulations that were just passed in June gives a fast path to power for anyone that is an interruptible load. So, so long as the grid itself, the grid operator is able to turn you off, it's not them saying—

Definitely, we need to make t-shirts that say, I am an interruptible load.

Oh my gosh, it makes me want to show an Abundance t-shirt. I'm sure that Peter's team sent me. But yes, I am an interruptible load. Don't ask my girlfriend.

Go right into the governor's office.

Let me in!

So, so, Matt, why isn't every data center, you know, deploying these giant battery packs? It seems like, you know, if, if I had that in my data center, I would be, you know, super smooth on the, on the load demand for my community.

We passed this regulation in Texas In June.

Okay.

Like the second week of June.

2 months ago.

So it's brand new.

Yeah.

So like, I invested in these guys because they drove the regulation and because they've got 10 gigawatts of like sites that can take advantage of this. And then after this was passed in Texas, FERC, the federal regulator of electricity, sent a letter to the 6 largest other grids in the country, not specifying the details, but saying, Do something like this. Figure this out. So this is going to become a very common thing to do.

It's called Agentic. Agentic Infrastructure.

Agentic is the startup. But this in general, this is an interruptible load or time-shifting demand. Again, like that red dashed line. Not all of you— some of you are just listening. The transmission line capacity and the substation, blah, blah, blah, blah, blah. Transformers. That's the limit. It's not the gas generators or the solar or wind. It's the transmission line. So if you can use batteries to fill up your data center, your data center batteries at night when the transmission line is unused and then not need to draw on the transmission during the day, that is— we've always known that was a good idea. We do it with EVs and so on. And this is a very big deal.

Amazing.

Would you say, Mez, this is— it's fair to characterize this as the energy or the grid equivalent of preemptive multitasking or reentrant multitasking in computing, basically allowing processes to say they can be paused and their compute load can be time shifted?

Yeah, I think that's one way to look at it. I think that's a great analogy, Alex. It's also like cache prefill, you know, like, yes, batteries is a cache for electrons instead of data.

So we're making our grid cacheable.

That's right. By the way, go ahead.

So how— what, you know, that big gap of the 200 gigawatts, how much of that do you think we can make a dent in by taking this approach?

I think approaches like this and approaches like electric vehicles also, right? The bulk of the batteries in the US are actually in EVs. So the company of mine, WeaveGrid, I shouldn't say of mine, like I'm blessed to be an investor in them because they're smarter than I am. They have long, for utilities, managed electric vehicle charging on the grid to reduce stress on the last mile, on the last block even, right? The limit on EV charging for the grid is actually the transformer on your block because Teslas cluster. If one person gets a Tesla, their neighbor like doubles in odds of getting a Tesla. Right. So they already have software to like time slice and even out the charging of the vehicles. So companies like that, in particular WeaveGrid, are using that technology to make the rest other loads on the grid more responsive and shaped in a way to allow AI data centers to play well. In fact, every EV charging company I know, has pivoted to trying to use their tech or their current capacity to enable data centers. And I think that's 100 gigawatts. I think that's— if we're smart about it, that's the next 5 years of AI data center growth.

Amazing. All right. What's next?

All right. Let's talk like more stuff. We all love solar. Let me tell you, we are— we're entering the phase where solar-powered AI data centers become viable. Many people have seen a chart like this I've shown. In '75, 1975, 1 watt of solar panels cost $100. Now it's $0.08 from China for a panel that's smaller, has a longer lifetime, is more durable, etc. And so that more than 1,000x price decline Does that get us to the point where we can power data centers with it? Well, data centers, because the chips are so expensive, it never makes economic sense to only run them when the sun shines. So you have to put them in storage as well. Battery prices have dropped by a factor of 14 since 2010. We have new technologies like lithium-ion has been dominant. Sodium is much more common. on planet Earth and lithium. So new technologies like sodium-ion batteries could drop the cost of batteries by a factor of 10. And even now, we already have the very first solar-plus-battery baseload power plants, and they're affordable. So we have them in the UAE, outside of Dubai. We have them in Chile. And a nice thing about this is, honestly, You know, natural gas turbines are sold out for years. You can— the fastest energy project you can build is a solar and battery project. You can get that done in 12 months.

Amazing.

So in the United Arab Emirates, this is a 1 gigawatt, 24/7 solar and battery project. What that means is they guarantee that the minimum power output at any time is a gigawatt. To do that is actually 5 gigawatts of solar and 19 gigawatt-hours of batteries. And the cost is like, is $6 a watt CapEx. That won't mean a lot to a lot of people. But let's just say the last nuclear power plant built in the US cost $15 a watt. The cheapest ones on planet Earth are Chinese, being built in China. Those are $4 a watt.

So it's really competitive.

It is, is reasonably competitive.

So as you and I texted about this right on the last earnings call at Tesla, I think it was at Tesla. Elon said he wanted to build out 100 gigawatts of solar capacity. Did you check into that?

Yeah. I mean, I think, look, it's a long-term vision. It's not next year. But Elon's overall vision is let's put all the compute in space. There's no land constraints. There's no permitting issues there. People won't complain about water use there. And he wants to build a terawatt. of AI. And if you're going to build a terawatt of AI, you've got, you know, 2 or 3 options, really. The world's deserts powered by solar and batteries, getting fission or fusion to work, ocean power like Ponfoss that I'll show, or space.

So we're talking about all of them, about Tesla building out solar, terrestrial solar, right? Competing, competing with China.

Yeah, he wants to build out the manufacturing for it, but I think his real motivation is not selling it to the on-land market in the US. I think his real motivation is to build that manufacturing capacity for space-based solar. If you look at what is—

A pedofab on the moon, presumably. I want to just pull on that, Peter's question, Mez, just a bit. If we take the 1,000x reduction per kilowatt or megawatt over the past few decades and extrapolate it, have you gone through the thought experiment of what would solar need to look like in order to achieve another 1,000x price per watt reduction?

Yes, this is a very good question, and it's an important clarification of how the cost reductions work. So our best model— I'm not that smart, right? Like, I'm I'm one of the top 5 forecasters of solar costs in the world. And it's not because I'm that smart. It's because I came out of tech and I came from a Moore's Law world and came into energy and just applied Moore's Law to it. But when you actually look at the details, it's not a reduction in time. It's a reduction with cumulative scale. It is Wright's Law. It's the learning rate. So every cumulative doubling of solar scale reduces costs by, let us say, 30%.

Interesting.

It fluctuates year to year. It's the real world and so on. So look, if we ignore the possibility that we need terawatts of AI compute and we just look at the world as it is, solar is now 8% of global electricity. And let's say we think solar can get to a third or two-thirds and maybe electricity demand goes up by a factor of 2. You've got 4 or 5, 6 doublings left. That means the cost of solar might drop by a factor of 4. Maybe even by a factor of eight, but not by a factor of a thousand. But if you start talking about you know building Dyson spheres, then we have a long way to go to keep reducing those costs.

I heard what I wanted to hear.

You heard Dyson sphere.

Pandering, Ramesh. Pandering.

Bingo.

Wait, can I? I want to drill in on that too because I've got a couple of questions along those lines.

Yeah. Ramez, how many of these installations are there being built around the world right now, like this exact style of monster-scale solar at scale?

We're just— oh, like gigawatt scale? You know, a handful, largely in China, the Middle East, some in LATAM. We have, you know, maybe more than a handful, maybe a dozen at this scale. Most solar plants today, you know, they're typically Somewhere between 50 megawatts and a few hundred megawatts, a gigawatt plant. The challenge and the reason, the biggest reason this is not yet an option for the US, because we could pull off something like this in the Southwest and it would actually, it would be cheap. It would be more expensive than it is in the Emirates because our labor costs are higher, but it would be fast. You're going to get it done in a year with even a natural gas turbine that you want to order from GE. You can't do that. But putting together the land parcels is actually the pain for this in the US.

Are these solar panels coming from China?

Probably. I mean, 85% do, so presumably. And of course, we double the price of Chinese solar panels in the US, so we hurt ourselves by keeping them out.

Sorry.

Yeah, so if this is the fastest path to If we want to do energy at scale, why aren't there people just going, or the US government just going, let's use eminent domain, grab whatever chunks of land we need to, and build this stuff? Because we could be done in a year.

You don't need eminent domain. The federal government is the number one landowner west of the Mississippi, and those federal lands are concentrated in places like Nevada, Arizona, places that have We have enormous solar resources, but it's not something that interests the current administration, I would say. But yeah, if I was thinking about it, I'd be thinking about how do we open up lands that are not amazing nature resources to build solar-powered data centers. And I think we get them done faster.

Sunshine seems better than drilling on federal land.

Absolutely. And I'll say this also. Regulations are the problem in lots of places. I was in Mexico recently. I was in Chihuahua and trying to convince the government of Chihuahua, a state in Mexico, to build a lot of solar-powered AI data centers. But in Chihuahua, it is actually illegal to have a private power generation and behind-the-meter power above, I think it was 500 kilowatts, right? Half a megawatt. So you just use— the law can't do it. And then secondly, the AI labs and the hyperscalers are extremely vigilant about data protections. They don't want their user data leaked or seized, and they especially don't want their model weights exfiltrated. So they're pretty careful about the countries they go into. So my advice to Mexico was like, look, change the laws to make it possible to build this sort of thing. And to provide ironclad guarantees of the protection and intellectual property protection of this data. And you've got an enormous business, right? More open land, lower population density, and better sun than the US.

This episode is brought to you by Blitzy, autonomous software development with infinite code context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale codebases with millions of lines of code. Engineers start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and precompiles code for each task. Blitzy delivers 80% or more of the development work autonomously while providing a guide for the final 20% of human development. required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre-IDE development tool, pairing it with their coding copilot of choice to bring an AI-native SDLC into their org. Ready to 5x your engineering velocity? Visit blitzy.com to schedule a demo and start building with Blitzy today.

I was going to push back on— well, first things first, the GPUs can't sit idle no matter what because they're so expensive. And yeah, but when you look at the underlying economics, about 5% of the cost of the data center, maybe up to 10%, is the power. But the GPU itself is 80% markup from Nvidia on top of 2x markup from TSMC with another 2x markup at the model provider level. So it's actually 20x overpriced. relative to the cost of turning sand into a chip, which is actually coming down too with efficiency and scale. And so at the— where Elon thinks at the fundamental level, it's actually not a given that the GPU is super expensive relative to the power once he gets the terafab up and running and the end-to-end sand in one side, chip out the other is fully automated. So that would completely flip all the math in this if he gets to that destination.

I think those are awesome comments, Dave. And I think it's right that GPUs are overpriced, or at least there's a lot of margin going in there. Uh, NVIDIA, people don't think of them as a network effect company, but CUDA, the programming layer to write to AI, is like their moat, right? It's not like their chips are good. Their chips are fine. AMD's chips are as good. Their interconnection between chips is great. And that does matter, but people like, uh, Huawei is kind of getting there, honestly. But CUDA has been the moat, and I think the CUDA moat is broken this year and next year. One of my portfolio companies, Lemurian— I met them at Abundance 360— is working on that. But also, now that you can tell AI, take my AI code and recompile it to run really fast on this AMD chip or the Cerebras chip, I think Nvidia's lead is—

So glad you brought that up. This is such an important topic. Because this is $5 trillion of US market cap that's hanging in the balance of this conversation. It's such an important and such a fragile thing, you know. So CUDA is the moat for sure. No doubt. All the AI researchers are too lazy to write custom kernels. Suddenly Fable 5 comes along. I've had great luck running custom kernels myself just in the last couple of weeks using Fable 5. So, so I think your prediction is, is probably right. I don't see why it wouldn't be right.

Yeah.

I think NVIDIA would say we have all kinds of other network effects and we have massive interconnect. Now, the interconnect is incredibly important for training. Yeah, but 90-95% of the load now is moving to inference where you don't really need the interconnect.

I think the interconnect still is very helpful for inference. You know, if you're going to run a model like KIMI-K3 or DeepSeek, not necessarily Flash, but the next DeepSeek V4, you're, you're Simon says you're running it on a rack, right? You're running it on 10 to 20 GPUs at a time. So the interconnect does matter somewhat, even for inference. But you're right that it matters even more, tremendously more for training.

Can I summarize?

Dave's point is just that at inference time, interconnect locally matters to the extent you need local coherence, but you no longer need global coherence at the level of an entire supercluster. It's just like a single rack of coherence.

Yeah, that's correct.

Coming back to energy, to summarize this, There's plenty of room for energy growth. We have the abundance thesis on energy writ large with solar, right? And one of the points that Elon's made before is at the end of the day, it's all about solar. Do you believe that thesis that solar in the long term is going to dominate beyond everything else?

It's complicated. And I think we underestimate the importance of geography. So the reality is, look, Look, from a regulatory standpoint, we're not building transmission. No place in China is building enough capacity to move electrons from place to place. Same thing as what I just showed with poles and wires. And so the problem for solar is not cost, and it's not nighttime because batteries are plunging in cost and are going to drop another 10x ultimately. It is winter. So in London, for instance, You get 1/6 or 1/7 as much insolation in January as you do in June or July. So you're going to build out your solar plant by a factor of 6 or 7?

No.

Or do you have— do we have a battery technology that can store months of power? Just a couple of interesting ideas out there. But like, think about, you know, the unit cost of electricity goes to a battery, basically battery CapEx amortized by how many times it gets used. So if you've got a battery that cycles daily, it's like battery cost, you know, CapEx divided by 365 or, you know, 3,650 if it's 10 years, let's say. If the battery gets used like twice to shift load between seasons, it's battery CapEx divided by that. So there are numerous startups. There's some crazy ideas. Sand. Compressed air, power to natural gas, yada yada yada. But right now, like it's clear to me that economically shifting battery, shifting energy through the day-night cycle—we're not totally there yet, but it is like the curves are are just heading that way. But dealing with winter, especially also we have not yet electrified heat. And if you look at you know the UK as an example, and at least in Northern Europe. If you go from burning natural gas for building heat to using heat pumps, your electricity demand doubles in winter. So we have this big, big, big winter problem that I think a lot of people are not reckoning with. And so I do believe nuclear is super useful, as is efforts to get seasonal storage, as are all of our efforts on fusion, as is advanced geothermal. Especially for those places that are further from the equator and have either long winters or long rainy seasons.

Location, location, location.

Yeah. All right. So this is— this last slide is saying that solar battery data centers are going to get cheaper and cheaper. And so, yes, one of you asked, why don't we move compute to where the energy is? And I fully believe that, right? In other energy loads, you can't move the population of New York City to a place that's sunny year-round. Not quickly.

It's called Florida.

Okay. It's Miami during COVID and mostly for crypto folks. But you can— the new load we haven't built, which is AI, why don't we site it where the energy is? Right. So that's, that's, that's one viewpoint. All right. Next, Peter, you wanted me to talk about fission and fusion, and both are super exciting.

Yeah, I mean, we hear a lot about it. We talk about it. We speak about the hyperscalers turning on, you know, defunct fission plants, investing in fusion companies. You know, it's interesting. A quick stat: 71% of Americans are against data centers, which is a higher percentage than are against a nuclear plant in their backyard, which I find amazing.

It's just insane. I'd rent my backyard out to both. It's not big enough, quiet, but, you know, we'll make some room.

Maybe, Mez, just a quick question before the segue. I just want to pull a little bit on the historic rhyme between the Middle East being a major source of oil, but now also being a major source of solar power. The thought experiment I've done, I'd be curious to get your thoughts, is the reason the Middle East has so much oil is, my understanding is, like hundreds of millions of years ago, There used to be a warm ocean with lots of plankton and other small creatures that ultimately resulted in the oil, and now it's largely desert, but still it's pretty warm. Any thoughts on the historic rhyme between why somehow Middle East is, on the one hand, supplier of all this oil power for data centers and now potentially solar power?

Well, I think the Middle East has amazing solar resources, but it's not as lumpy as their fossil fuel resources, and especially their oil resources. If you look around the globe, you have Australia. I mean, if I was thinking— actually, I just said Mexico— if I was thinking about let's do a lot of solar and battery-powered data centers for AI, I would be really pushing in Australia.

And the other half?

Yeah, you've got a friendly government, you've got enormous amounts of space, you've some of the world's best solar resources. Chile and Mexico, not amazing oil producers. Mexico was once, they're not anymore. But solar resources that are equivalent to the Middle East. So if you look around the world, it's interesting. Obviously, we all know intuitively, you know, some places are much sunnier than others. But in the places that people live, the actual solar energy that falls varies by at most a factor of 2. between the least sunny and most sunny places, which is kind of crazy.

Really?

Yeah. The seasonal effect is bigger, right? And as you get further from the equator, the seasonal effect gets bigger. I live in Seattle. I know this. Whereas the oil density underground is much, much, much higher dispersion, much higher concentration in a few spots. So overall, like when I, when I'm in the Middle East, I'm always counseling them on Find a way to export energy. You're not going to build poles and wires to move electricity from Saudi Arabia to the U.S. But you know, I used to say like steel, like Iceland is makes a lot of aluminum. They have no bauxite ore, but they have cheap geothermal. I know I've got a company I've been talking to right now. It's using Icelandic geothermal to make sustainable aviation fuels like e-fuels, power to fuels, because they can export that. Right. So if I was in Saudi, I used to say like make industrial Uses of electricity. But now we just say like, make data, make intelligence. Yeah, yeah. But you've got to change the laws such that an OpenAI, an Anthropic, a Grok, whoever, a Google is comfortable siting their crown jewels in your country.

Just by way of a reference number for the audience, you know, when I last looked at it on the Energy Abundance Thesis, you know, We have 8,000 times more energy than hits the surface of the Earth than we consume as a species in a year, right? So there's plenty of energy out there. It's just not in usable form. And the whole conversation here is how do we take that energy that's latent and make it usable, right? And that's the role of technology.

That's right.

The other commentary is that fossil fuels are just an old battery that we've been using up, right? And we've used up a lot of it. 25% or 30% of that battery?

No one knows. But I mean, you know, the cure for high prices is high prices. So if we ever started to run low, there'd be more incentive to explore and find stuff.

Yeah, I have a question for you, Ramez, on the, on the oil market real quick. I remember you commenting once that the 2013 oil price— oil crisis— oil crash was because of a 2% oversupply in the market. Like it's a really tightly wound market. Is that still the case, or with all the Middle East conflict and everything else, we're now in tension? It's going to stay that way. Oh my gosh!

I mean, there's a lot to say about that. Look like two interesting things. Yeah, there's two interesting things about the the Iran war and its impact on oil prices, and why it's been relatively muted. Number one, China did us all a solid. China built The world's largest oil strategic reserve, right? And they were willing to drain it during this period. So China has, you know, more than the rest of the world's strategic reserves combined, has helped keep oil prices low. But 2, the ratio of global GDP to global spending on oil has roughly tripled or quadrupled since the oil crisis of the '70s. So there are critical things that are highly dependent upon oil, aviation, shipping, trucking, and so on. But overall, the world has moved to more of a services economy, and that has created sort of more demand elasticity. It has allowed the world to deal with a shortfall in oil in a way that we couldn't in the '70s when our economy was just more physical. Maybe that's why you haven't seen oil spike to $200.

And I'd just like to pull in the, the geopolitical angle here. There's a theory in certain circles that, uh, an ulterior motive for the Venezuelan operation and then the war with Iran was actually to cut off China's, uh, in event of a Chinese invasion of Taiwan, uh, open paren, TSMC, close paren, that China would require backup oil supplies because they would get embargoed by the Western bloc, and their go-to sources for backup oil in such an invasion would be would look like Venezuela, Iran, maybe Cuba. And so so the the full sort of theory here is the the recent military adventures that we've seen Venezuela, Iran are actually at some level. play to deter China from invading Taiwan to gain access to the TSMC fabs and basically the future light cone of AI. Any thoughts on that?

There's insight there, but I don't agree with it as stated. And the reason for that is, uh, China bought oil from Iran during the— this, this war. The U.S. still sells oil to China. Uh, it just wasn't that planned out. The actual DOD war plans in a situation like that are to use the US submarine fleet to sink tankers that are heading to China. That's the actual proposal of what to do. Who knows if that's a good idea? I'm not going to get into that right now. Oil is mostly fungible. So the fact that China buys oil from Iran, us bombing Iran, or even if we successfully close the strait, Doesn't really hurt China that much because they can buy cargoes from somewhere else. They were getting a discount from Iran because it was embargoed oil and that they were willing to buy. So they have to pay, you know, a few bucks more per barrel. It's not that big a deal to them in wartime. It's a kinetic sanction. It's a kinetic embargo, which is a different sort. And yeah, I think this is all totally unclassified. The simulations of wars like that are US submarines, you know, torpedo tankers that are heading to China.

Onwards to the horizon of fission and fusion.

Okay. Let's talk about the atom and the power thereof. Going back to the '50s or, you know, a retro future. This is a complicated slide for those of you who are just listening. Basically, there's, you know, 2 approaches to nuclear fission. Which is what we've been doing since the '60s. The traditional one is big reactors. And the simplest thing you can do to boost nuclear production worldwide is, A, stop shutting down nuclear plants. Germany should not have shut down its— 2, nuclear plants have like an end-of-life plan. We can usually extend them. In some cases, we can actually upgrade them. To produce more power. 3, there are some plants that have been shut down, like Three Mile Island, that we can actually restart safely. But that gets you, you know, a few gigawatts, right? If we really want a nuclear renaissance, the, the core issue with nuclear fission today is that outside of China and perhaps South Korea, it is ruinously expensive. And why is it ruinously expensive? It's because we don't do a lot of it.

Right.

Anything that you do infrequently is expensive, right? You don't get good at the things that you do occasionally. The things that are cheap are the things that you do repeatedly again and again and again. So there are 2 paths happening to bolster nuclear in the US. And I'm a critic of this administration on many fronts and on some energy fronts, but I'd say this administration has the best nuclear policies of any in recent history. Still missing some things, I think, but the best that we've seen. So the left side of this is large reactors. We have this thing called the AP1000. It's sort of a— it's the Westinghouse reactor. It's sort of a workhorse reactor. We've built a couple of them in the West, let's say 4. China took a variant, took this design, made their own variant that had a local supply chain, and they've built more than a dozen. And they built them at higher power than this, 1.4 gigawatts instead of 1 gigawatt. So one plan is we're going to produce a process to get more of these built. The administration is talking about— has created structures for loan guarantees, for financing and so on, because if you stamp out a lot of these, the cost should come down. Right now—

What plant is the most stamped out so far?

Is that in France? Light water reactors like those used in France. So France is the poster child. The US generates the most nuclear electricity of any country on Earth. Actually, it's not really known. China is building the most right now. France gets the highest fraction of its electricity from nuclear. And they basically, with slight caveats, basically took the same design and stamped it out again and again and again.

How many does France have, like 60 nuclear power plants now?

Something on that order. Yeah, but even France—

Something like some— something like 80% of their electricity is nuclear. It's crazy.

And they export it to the rest of the Eurozone as well. Fission was basically born in France. Thank you, Curies.

Yeah, but even France is struggling, right? There's something called the European Pressurized Reactor, which is mostly a French design, and that thing is kind of a disaster right now. It's a boondoggle running over, going slow. Again, like if you take one design and you do a lot of it, it gets cheap. But usually, and this is critical for the sector as an investor, the first one usually runs overpriced over time and, you know, has problems you didn't anticipate. So if you want a thriving nuclear industry, you just have to know that the first one you build of a new model is probably going to have problems. But after you've built 3, 4, 5, maybe more, you sort out those problems. You build experience in the crews, you build experience in the engineers, you sort out design issues, you build a supply chain to provide the parts that you need.

That's right.

So one plan is we're going to take the large reactors that we have built a couple of times, and now we have the US The US government has created financing, sort of a backstop, loan guarantees for about 8 of these. Another startup, the NuClear company, I invested in one of the founders' previous companies. They have a plan to basically build fleets of these because that's how you have to finance it. You can't finance one because you know you're going to miss your targets. But if you can finance a bunch at a time, a bunch of them amortize the cost. You know, in China, like after they got to like 6, 7, 8, Like the costs had really come down and stabilized.

So those who are fearful about nuclear, and it's still, you know, probably a good percentage, they think about Three Mile Island and Fukushima. We're talking about early generation plants, right? Are those Gen 1 or Gen 2 plants?

Something like that. And these are Gen 3 or Gen 3+ or Gen 4 plants. And one of those important things to understand about them is basically all of these are passive safe. What that means is You can knock out the power to them and they won't have a meltdown. Fukushima happened because you circulate water around the nuclear core to take the heat away from it and then use it to turn a steam turbine. That pump for that was powered by grid electricity. So the tsunami that hit Fukushima knocked out the power lines, and so the pumps stopped working, even though there was power right next to them for the nuclear reactor. Right. New nuclear fission designs are passive safe.

Would you call them failsafe plants?

Nothing is totally failsafe, but they're designed to take a 747 crashing into them and the power going out from the grid and keep operating without any meltdown.

I'm curious, just pulling on that, Mez, what happened in the 1970s? I assume you've seen the television show. We talk about it sometimes on the pod, For All Mankind. It's sort of an alternative historic reality where we get fission, it never gets abandoned. What happened in the '70s? Did we just waste the past 50 years not building enough nuclear energy, fission in particular, finding ourselves in a suboptimal future?

I don't think so exactly. I think we could have done better, but people are somewhat risk-averse. Radiation— we learned, you know, in the '60s that radiation causes cancer. The radiation released from a well-operating nuclear plant is really minimal. It's unlikely to cause cancer. But so we did increase the regulatory state. We did increase the burden of proving that things were safe, and that had some cost. And then things just fizzled out. And we talk a lot about flywheels and positive feedback loops. We had a negative feedback loop. Once— and this is the same thing France has seen— once the industry is no longer building, you lose the expertise, you lose the supply chain that makes the parts that you need, you lose the manufacturing facilities, and everything gets more expensive. So if you're not constantly scaling, you are going to backslide, basically.

So those were large reactors. And you have on this slide here the small modular reactors. Speak to that.

The right side, and this is the area that investors are super stoked about, that I was stoked about 15 years ago, got less stoked about, and now I'm kind of becoming maybe hopeful again, is what we call small modular reactors. So we talk about a lot, I talk a lot about learning rates, right? Which is how fast does something get cheap? And every technology, if you build more of it, gets cheap at some rate as the scale increases. But the things that get cheapest the fastest are those that are built in factories in high volume and have the smallest number of moving parts. So the idea of SMRs is to build as much of this in modular, repeatable, factory-built situations as possible and do as little stick building, as little assembly or construction. Construction is a dirty word, right? Construction does not get cheaper. Manufacturing gets cheaper. So move as much of this as we can to a manufacturing process. At the limit, it's a factory that spits out nuclear reactors that you just barge or semi-truck to location. Some of these are not that. Many of these are— the parts are made in a factory in a standardized way that you assemble like Legos onsite. This is an incredibly sexy space for investors right now. Yesterday, we found out that Valar Atomics raised $1 billion at a $6 billion valuation for a small startup that doesn't have a working reactor. Other companies— Aloe is probably my favorite company in this space, but there's a ton of companies.

My friend's company at X Energy, Went public recently.

X Energy is an amazing company, actually. I really like their design. So that's the Radiant. There's one on here that's on the very, very small scale. There's a line between SMR and microreactor. So can you make it small enough to fit in a shipping container? So the military, for instance, you've got Radiant is on this slide. The US military for military bases would like Shipping container or half shipping container sized reactors to power bases in the US, but maybe in forward deployed locations as well. So they don't have to move fuel. So there's a lot happening in this space.

So natrium, I see on the chart here is, you know, a third of a gigawatt compared to the AP1000, which is roughly a gigawatt. Yeah. When you say, you know, if, if natrium goes into mass production, mass production being Tens, you know, 50 units. I mean, the relative price of buying 3 of those Natrium units versus an AP1000, what's that? You know, is there economies to bigger plants or is it just stack them together?

Yeah, there are economies to bigger plants. Bigger plants use less steel and less cement per unit of power output. So there are economies to bigger plants. And that's how we used to think in the '50s and '60s primarily. But there's learning that happens from building more plants. and doing more of it in a factory. So personally, my guess is like the BWRX-300, Natrium are in an awkward middle because they're not really factory built. They build a bunch of components in the factory, so they do field assembly. And so I worry about them, but they might end up being the ideal optimal solution. At the other end of the spectrum, you've got like Gradient here. Their reactor is 10 megawatts. So it's 1/100th the size of an AP1000. They put them together in clusters of 5. That's a pod for 50 megawatts. They have lower efficiency of using steel and cement, but they can build it entirely inside of a factory.

Which is the first one of these? You know, these are not online yet. These are all theoretical. Which is the first one coming online, you think?

Yeah. So The optimistic projections from these companies are 2030 to the early 2030s. And that's for the small ones. The next AP1000 is probably a few years later than that. These projections will probably be missed. I don't expect anyone to actually have a commercial small modular reactor in 2030, 2031. But maybe not. Like the size of slip is probably smaller for a small reactor. And like everything else we've been saying, the first units here are not going to be cheap. The first units are going to be expensive. So the key is to build an order book from a customer that believes that by ordering enough, they're going to drive down the price, or to build out a multi-customer order book where you've built some mechanism for cost And risk sharing between these AI data centers. You know, we talk about, is nuclear the solution for AI data centers? Maybe. There's other ways to power AI, but AI data centers might be the best thing that's ever happened to the nuclear industry.

I'm curious, Mezit, if we just take this argument in extremis, where are the nanoreactors? Why don't we see 100-kilowatt nano or pico reactors that can be co-located with every GPU? Do you think there's an opportunity in the space there?

I think it's really hard. I think you do hit some economies of scale issues as you get down to the bottom. And you do, you have a certain size for criticality. But you have, you know, on this chart at the bottom here, you have, you know, with companies like Gradient, you have like 5-megawatt-size reactors. And that's, you know, power a neighborhood or power a few of them powering a military base. That sort of scale. And do you also—

It's just expensive, right?

Well, I mean, there are many ways one could imagine doing it. Another would be like gamma voltaics or beta voltaics. Like you put the radionuclides directly in the silicon and then you pocket the energy from radioactive decay. Do you think there's any hope for those who want to embed radionuclides directly in the GPU silicon?

I mean, that's what we talk about as, you know, radiothermal or nuclear thermal. And that's different than a fission reactor. We use that on satellites or space probes.

Yeah, deep space probes, we use radiothermal. Absolutely. RTGs.

Yeah. I don't expect to see RTGs become popular on Earth. When you look at the cost of those, they're actually really high, but they can meet mission needs for something that keeps on putting out power for decades without needing to be refueled. But their power output per unit mass is not Aren't those just constantly spewing radiation though?

That is the idea.

I mean, but in a bad way too. Like, okay.

Yeah. Ouch.

I'll let you figure that one out. So, so let me just say this. It's really interesting how you've got this footrace between, you know, if you said early 2030s for all these nuclear projects, but Elon is racing into space concurrent with that. And solar, you said, is coming down 30% every time we double The production, all those things are in a footrace.

And fusion's got the same timeframe, right? We're no longer 50 years in waiting, we're 5 years in waiting. Welcome to the health section of Moonshots brought to you by Fountain Life. You know, my mission is to help you use the latest technologies, including AI, to not just do your work at home, teach your kids, but to help you live a long and healthy life. I'm here today with an extraordinary physician, the Chief Medical Officer of Fountain Life, Dr. Dawn Musalem. Dawn, let's talk about cancer. You know, I know from the member database that we have at Fountain, our members who come in who think they're healthy, it turns out 3.3% of them have a cancer in their body they don't know about.

That's right. You know, the majority of cancers that we screen for, those aren't the ones that are necessarily taking the lives when found at a late stage. We know that when cancer is found early, the chances The chances for cure are much higher. We know it's much easier to treat a cancer when found early vs. when found late. What we're finding in our members is over 3.3% were found to have these cancers that otherwise wouldn't have been found or detected.

It's interesting. You don't feel a cancer until stage 3 or stage 4. If you don't know what's going on inside your body, it's like driving your car with your eyes closed. You can know. When members come through Found, how do they detect cancers?

So we're doing full-body MRI, and we also do early cancer detection screening. This is very, very important, and these are not typical tools used in the conventional care setting when it comes to prevention. This is a hard thing because currently these are not studies that insurance would yet be covering, but the goal is to collect these numbers, do the research, and work hard to democratize wellness.

Yeah, so at the end of the day, you can know what's going on inside your body. It's your obligation to know. So check out Fountain Life. You can go to fountainlife.com/peter to get access to the latest technology to help you detect cancer at the very beginning, at stage 1, when it is curable, before it gets to stage 3 or stage 4 and you're in a world of hurt.

Let's, uh, let's hit to Fusion. This slide just says stuff I've already said. The main thing I want to tell you is like hyperscalers saying, oh, we're like using SMRs for our data center. It's still kind of a fiction. It's— I mean, it's outside the 5-year window that we really can— that is investable, that we really have like really good optics on. But the pull from data centers is giving a massive, you know, tailwind to every nuclear company, especially the SMR startups, but also even Westinghouse with their big reactors. Let's talk fusion.

All right.

The joke was always that fusion is 50 years in the future and always will be. That's just no longer true. We now have well over 50 fusion startups.

Venture-backed fusion companies. I mean, that's like science fiction in its own right.

That's right. Absolutely. Some of these, ironically, came because of budget cuts in academia. If you look at Commonwealth Fusion, which is considered sort of the safe bet of fusion, if you will, if there is such a thing, that team at Harvard, you know, their grants were struggling. And so they said, why don't we form a company, you guys? And so they did. And they're now the front runner. They have a— how can I say this? I don't want to call any fusion reactor a conservative design. But they have like the most conservative design in this sci-fi field of fusion.

There's a striking resemblance. A friend of mine from college and grad school, as always, is on their board. You could call it like a privatization of MIT's entire nuclear engineering department.

Yeah, indeed, indeed. And not just that of ITER. You know, we have— we've had publicly funded fusion projects. NIF in the US uses big lasers, National Ignition Facility. It's really a weapons science facility is what it is. And ITER in France, the international and European project that was a tokamak, that is the big donut style. And ITER's plan was to build a reactor that was at least 5 gigawatts and would cost at least $40 billion. All right. And so what you have with CFS is the company has found a way to scale that down. So here's how I think about the 3 families of fusion. This is a massive oversimplification. My fusion startup founder friends are going to yell at me for not including their particular designs. But, you know, there's 3 big ones. Tokamaks are the donuts that use big magnets to guide a plasma around and make that plasma slam into itself and capture the energy. That's what ITER is. That's where we have the most scientific data from past experiments about fusion. And the leading company in this space, Commonwealth Fusion, CFS, basically just has a technology that takes the enormous superconducting magnets that we were going to use in this European project and shrinks them down dramatically.

Mm-hmm.

We have a thin film material that you can wind around and wrap around that makes the magnet dramatically smaller. And because it's a superconducting magnet, you've got to cool it tremendously. And now that it's much smaller, you need a lot less cooling energy. It's a lot lower CapEx. So instead of a 5-gigawatt reactor being necessary to be breakeven, they can do it in like 600 megawatts is their plan.

Hmm.

The next one is lasers. And again, fusion is all about like, let's slam particles together and force them to fuse into other particles, which takes a lot of energy, then releases it. NIF, the National Ignition Facility in the US, uses lasers, the world's most powerful banks of lasers, to slam these pellets of fuel, of hydrogen fuel, to ignite fusion. In some ways, it's the closest thing to what happens inside the sun of the techniques that we have. They're a weapons facility. They've had some amazing results, but we can't really productionize what they're doing. But they've had maybe in some ways the most exciting scientific result in this. And there's a few great companies in that space. And then reverse field configuration pulse magnetoinertial. This is basically a railgun, if you will. You know, railguns like use magnets, magnetic coils to shoot things like metal out of them really fast. The leading company in this space, Helion, uses basically 2 railguns, 2 tubes of magnetic coils, to take a plasma at either end, slam it together, and then compress it with power electronics. And then when the explosion of fusion happens, the power electronics that were creating that magnetic field that compressed the explosion or compressed the collision to make it fully fuse captures the energy in reverse.

Hmm.

These are 3 approaches. Most fusion companies capture the energy as heat and then have to use it to turn a steam turbine.

Hmm.

The nice thing about what's on the right is at least Helion and a couple other companies capture it directly as electricity that turns into electric current. They don't lose 60% of the energy that you lose in a steam turbine, and they don't have the added CapEx So the left side is what's most likely to happen soon. Commonwealth Fusion is the company that is most backed by scientific accomplishments. Helion is the company of the ones that have raised more than a billion that, if they work, I think has the pathway to the cheapest cost.

True followers of the pod will remember that we covered that. It's the coolest thing ever, but it was covered in a chipmunk voice.

So who was that? Remember that video?

I took Naveen Jain with me on a tour of Helion's reactor late last year, actually.

I had Bob Mumgaard from Commonwealth Fusion on our stage last year, and he was amazing. We can talk about that. I want to bring Helion onto the Abundance 360 stage this coming year. So let's work together to make that happen.

I can email those guys anytime.

For anyone in the audience who hasn't been on tours of either of these, I'll just point them— at least for NIF, I've been on a tour of NIF, but it was featured in one of the recent J.J. Abrams Star Trek movies as the warp core. So just Google Star Trek NIF, and you can see the scene where the actual core, where the whole realm at the center of all those lasers pointing at one location, is actually in the movie.

Someone can put the links into the show notes so anybody wants to look at them, because these are really cool videos that we covered.

Elephant in the room question. Fusion seems now to be a when rather than if. So, so when?

So look, I think that might be on my next slide.

And Mez, I'll work with you to get the CEO of Helion on our stage together.

Yeah. David Kirtley. He's a great guy. He's here in Seattle. The most aggressive timeline is Helion. They have a power purchase agreement from Microsoft to provide 50 megawatts of power. So very small. And again, the smaller you can build, the more modular it is. And we get those lending rates in 2028.

Wow.

Everybody else is talking about sometime in the early 2030s. Let me see if we have— here's the timeline slide.

I mean, you know, fission, which we know how to do, is somehow a 2031, 2032 thing. Yet fusion, which we don't know how to do, is a 2028 thing. Do you believe that?

Look, like my view of this and founders of mine who are listening, please don't take this as an insult, is every startup exaggerates how quickly they can get things done. And that's just, you know, part of the game.

You have to be an optimist to get through the process.

Yeah, you got to be an optimist, right? And they actually believe it. Maybe they believe Like, they believe it's possible, they tell you, but maybe unlikely. But it kind of like— I think it's plausible. Helion is very plausible.

What makes it plausible to me is the fact that the barriers are all regulatory. And if, for whatever reason, a governor is super excited about fusion and the voters are all violently opposed to fission, then that actually could make the difference, I would think.

I'd say that the barriers are still physics and engineering. But here's something that's fascinating. And thank you for bringing that up, David, because this is actually quite important. A couple of years ago, we had a question of how would the US regulate fusion? Because if the US regulated fusion reactors like fission reactors, it was going to be a major drag on the industry. It might still be better to do fusion than fission for a variety of reasons. But instead, they are regulated like the radiological imaging machines they use in hospitals. At least Helion is at this point.

Wow.

And there's good reason for that. In a fission reactor, if you stop cooling it and you don't have the control rods in, heat will build up and it will get hot enough that it melts the steel that it's in. That's what a meltdown is. As I said, new reactors are passive safe. Without any pumping, the hot water goes up and then it circulates and so on. But there's still like, you know, you can imagine like breaching that containment, slicing through those pipes, and you have a meltdown.

I mean, aren't there— apologies for interrupting— but there are alternative architectures, pebble bed type architectures. I know thorium goes in and out of fashion, especially in China. Aren't there also like hybrid solutions that are in some sense meltdown-proof?

There are ways, but there's, there's nothing There's nothing in the pipeline that if you took an adamantium battle axe to wouldn't melt down. Everything uses a coolant. Every fission— Okay, maybe there's one startup, but I'm not going to mention them. I don't know what I can say. But basically everything in the pipeline uses a coolant to pull heat away from the fission core and then to turn that into electricity in some way. And if you eliminate the coolant, If you break the coolant pipes, the core can overheat and melt down. Fusion is different. In fusion, it's the opposite. You have to capture the energy of the fusion explosion and feed it back in, either to maintain the fusion reaction or to another pulse. Like, helium is pulsed, right? It keeps doing the same thing, or NIF is pulsed with lasers. So with fusion, if Like if you mess something up in the reactor, it just like goes do do do do. It just like fails, and nothing bad happens. So it is fusion generates some radiation. You have to actually replace some of the parts in the reactor every five years because steel is being hit by neutrons and being weakened. And yeah, there's actual like some radiation that's low level. There's real costs to that. Like free energy does not mean free. Because the CapEx and the maintenance still cost something, but you cannot have a meltdown in any way that we understand.

That's fascinating. So you're saying basically the regulatory treatment is whether the system is default on versus default off.

I don't know if that's what the NRC used as a criteria, but that is the dividing line between fusion and fission, and it was sort of missed in the public discussion in the press.

Fascinating.

That regulatory conclusion, it happened during COVID was actually a huge deal for the fusion industry.

That sounds like great news.

So if fusion is this close, shouldn't we just do solar and battery for a big chunk and then fusion for where we need high energy needs and we're done?

So all of these companies might fail. They might 100% fail. And in addition to that, they might succeed but be too expensive. Just because your fuel source costs very little doesn't mean your energy will cost little if the CapEx is very high, if the maintenance is very high, etc., etc., etc.

And the demand means that we're going to need all of it, all the different sources, no matter what, and you want to diversify your risk anyway.

Yeah, I always believe in having more tools in the toolbox than you think you need, more arrows in the quiver than you think you need, because some of them won't work out. Let me— please go on.

Quick question. I would love to go back to one of my favorite hobbyhorses, the Dyson swarm. Do you think the Dyson Swarm wants to be solar PV powered, or does it want to be fusion powered, or does it want to be other?

I mean, those are the 2 options. I think they're both great options. I think it's probably— it's much more modular to be solar PV powered. Again, like, fusion also has a minimum viable size, right? So Commonwealth— we thought with ITER that tokamaks had to be 5 gigawatts. Commonwealth has found a way to scale it down to 600 megawatts. Right, but you saw some minimal viable size where solar is just super modular. And if you're in a Dyson swarm, you have 24/7 sunlight. So none of these technologies is gonna be cheaper than just plain solar, but they work in winter and they work.

And also, many of us— I mean, I probably— maybe I'll speak for a few of the other moonshot mates here— we watched Back to the Future Part 2 and we saw Mr. Fusion being promised in the '80s. And then, you know, very compact, very compact, like impossible, car-sized. And then we look at the Lawson triple product over the decades and we see, no, actually fusion wasn't always 50 years out. It was creeping up on us, but many folks weren't paying attention to progress in the triple product. Is there an equivalent of the triple product for the compactness of these devices so that we do in the end get our Mr. Fusion?

Let me talk about compactness and let's talk about that triple product and show the progress we're making. The most audacious fusion startup that I know of is a company called Avalanche Fusion, also in the Seattle area. And they believe they can make a fusion reactor small enough to power a car. It's not Mr. Fusion. It's more like half the size of a car. But that— sometimes they talk about like big backpack. So that is the most ambitious project as far as compactness. Typically in fusion, you have like a sliding scale of like what has the most like de-risked science, but has like a more conventional power cost versus what could be like revolutionary in power cost or compactness. But the science is like, let's hope you get it right. You know, like we don't have as much evidence. And so Avalanche is on that end where if it works, it changes the entire world. But the confidence it works is much lower than the confidence for like a common law fusion.

And any insight into PB-11, proton-boron-11 fusion reactors?

People are very interested in it. And one of my—

one of my— one of my company portfolio companies is a PB-11 company out of Caltech. I need to introduce you to them.

Which one?

I don't think they're public. I don't want to mention here, but—

I was just looking at a slide deck from a PB-11 company just the other day. You know, PB-11 is one of the ways that you can— one of the fuels you can use to potentially get a nuclear reactor that is down to the like 1, 2, 3 cents a kilowatt-hour.

Yeah, sort of. The vision there is, can you build it small enough where you put it in the back of a large consumer airplane and it powers the engines? Yeah. And then it's also interplanetary flight.

There are still You know, scientific and technical risks there. There are still a lot of unknowns.

Welcome to the revolution. Welcome to deep tech investing.

So there are more unknowns, whereas Commonwealth Fusion's pitch is, look, the science has been proven at ITER scale that if you have magnets this strong, you can make fusion and get this much energy out. We're just doing that with much more compact magnets. I think the reality is a little bit more complex than that, but they really say it's— we've reduced it to an engineering problem. Nobody else can quite say that. Again, there's a gradient of how close you are to that. Let's talk about the triple product that Alex brought up. So this is temperature times pressure times duration. And, you know, I love graphs. And so I like— I believe something when I see movement on a graph. So what you're about to see— sorry, listeners, I'll describe it, I'll try to narrate it— is over time, from 1956 to now, how close have we gotten to a triple product of above 1, above 10, and then infinity? And this is all a log scale on every axis, so it's a brutal, brutal scale. But, you know, once upon a time, fusion really was— oh no, is this— here we go— really was 50 years in the future. And what we're seeing, for the listeners, is new points appearing that are— each one is a fusion experiment. And as the years elapse, and they're going up and to the right, how close they are to the upper right is the zone of triple product, ultimately of infinity. But above 1, above 10 is probably what you need. And NIF, that last X on the borderline, is a triple product above 1. It's not— it was like a theoretical net energy gain. A practical net energy gain means you capture the energy and then you convert it back into the lasers, the magnets, whatever. They did not achieve that. Their reactor cannot do that. But it tells us, and the progress on this tells us, that it's not just you know, hope. We are just getting closer to what we need. Amazing.

As you were saying— I just want to move us along if we could. But, um, but Alex, please.

Where do you think the stereotype that it's always 50 years out came from if one can just look at the triple product over the decades and say it's clearly like Moore's Law, like any experience curve, it's clearly marching up to the right?

It's on exactly that I mean, that's what I do. You know, if I'm like, well, let me just like these days, let me just ask ChatGPT, like, show me a graph of the progress here. But, you know, like, look, at the end of the day, for the average person, people have been talking about it and it hasn't appeared. So I just discount the reality or the likelihood of it appearing. Overpromised, underdelivered for a long time.

On behalf of my Moonshot mates and myself, I'm inviting you to join us at our inaugural Moonshots Live event on September 25th in downtown LA. Alex, Salim, Dave, and I will be hosting 1,500 entrepreneurs, builders, and creators, and hopefully you, for a full day dedicated to designing and building your moonshot. We'll be awarding the Build with Gemini XPRIZE, the world's largest hackathon, and the Future Vision XPRIZE Film Competition over $5 million in purses with over 25,000 entries. You're gonna hear the top 5 pitches from both competitions. And get a chance to shape the outcome. Join us. Seats are limited. Admission is competitive. Check it out at moonshots.com.

Let's close with talking about out-of-this-world compute. So AI in space and AI in the oceans, less known, but actually sort of a similar pitch. So of course, everybody's heard about Elon talking about space-based data centers. And it's interesting, the response is like very bipolar. It's people saying that's impossible, it'll never work. People saying this is it, we're going to have a terawatt of compute in space. I'm somewhere in the middle. AI in space, it becomes cost competitive when you get down to a launch cost that is, you know, something like 4 times to 10 times cheaper than what we have today. Nobody knows for sure because we haven't done it, but the back of the envelope says that. And it's in some ways it's a hedge against regulation. If demand for compute keeps going indefinitely and sites on land keep being blocked by the grid, by even Texas passing a temporary moratorium or audit, or by people protesting whether the grounds are there or not, then Building it in space, even if it's more expensive than building it on land, is a way to work around that bottleneck. So no grid delays, no opposite, no local opposition, no terrestrial permitting, etc., etc., etc. I will say I think we are not fully internalizing what the scale of this is or the permitting and regulatory challenges with doing a launch. At that volume. So we want to build 200 gigawatts of compute by 2030, 2030. That's the chip volume, right? So 10 to 20 gigawatts a year. To get 1 gigawatt of AI in space based on SpaceX's design, you're talking about 6 times SpaceX's best annual year of launch and twice SpaceX's cumulative scale of launch.

For just 1 gigawatt.

For 1 gigawatt. You're talking about—

What was the calculation we did? It's like 5,000 launch or 8,000 launches of Starship to put up his ambition of a terawatt initially.

That doesn't even get you close to a terawatt. I mean, it's to get 10 gigawatts a year, You're talking about 1,500 to 2,000 launches a year. So 10 gigawatts a year is like 5 or 6 launches of Starship a day.

But I guess the elephant in this particular orbital room, I have to mention, I think we talked about it on the pod previously, if you just look at the history of upmass from SpaceX and otherwise over the past few years, it's on a nice clean exponential trend. I forget what the exact year-over-year trend is, I did the extrapolation. 144 years from now at the present trend, the upmass, the cumulative upmass, would equal Earth's mass. So we basically disassemble Earth on the present trend 144 years from now. The upmass is increasing really quickly.

There's some other planets we can take apart first. There's some uglier planets than Earth.

Do you have a favorite?

Yeah, don't make me pick.

But you know, Mercury maybe? Mercury is intriguing.

So here we go again. The hate mail is flowing in. I can feel it already.

I don't care.

Mercury's attractive because it gets lots of insolation and no one's really using it for anything.

It's a good orbit.

All right, Salim, take us back to reality.

So yeah, so I mean, this looks— unless we hit that exponential absolutely full on and go right down that path, This looks prohibitive for 15, 20 years.

I mean, look, here's how I see it. Let me— I will get to the limits on launch in a sec. Let me put it another way. Elon wants to go to Mars. To go to Mars, he has to drive Starship launch costs down to like close to the marginal cost. To do that, you need a high Starship cadence. You've got to build tens of Starships, maybe hundreds, and you've got to launch them something like daily, right? Or at least like weekly, whatever. To amortize the R&D, amortize the CapEx. There is not enough demand for communications on Earth to finance that via Starlink. There's no business model for Mars yet. So this is a gift to SpaceX that we have this AI demand. If the AI demand keeps going and there's no— and it gets bottlenecked in ways to build it, we will find a way to do this. And with the IPO, He's got the funds to launch at least a gigawatt into space, right? Even at— well, maybe not, but something on that order. So I don't think of it as like, what's the limit on Starship first? I think of it as like, this is a demand driver potentially for Starship. That said, it's not clear to me that the world will permit more than like 200 Starship launches a year, which would already be an enormous, enormous amount. That'd be huge, right? 200 Starship launches a year is 20,000 tonnes to orbit. That is exceeding, you know, all human launch to date every year, several times over. That's amazing. But you hit some limit. And if you're regulated by the FAA, the reliability you have to hit, you know, one failed launch or one explosion means you're grounded. So I'm sure Elon has a plan.

He talks about needing to get to airline-like operations, right? So, yeah, I mean, here are the numbers for his target was 100 gigawatts per year of compute in orbit initially, equivalent to the entire compute today, which is around 80 gigawatts or so. And, you know, it's 20 to 30 satellites per Starship. So we're talking about on the order of 30,000 launches per year. which is a launch roughly every 15 minutes. Now, if you think about it as rockets, and I've been in the rocket business for the longest period of my life, it's prohibitive and it's discontinuous. You can't think about rockets in that regard. But if you talk about airline-like operations, right, there's multiple launches per second of airlines around the world. So it really becomes— it comes down to that. Now, is Starship a vehicle capable of that level? He's built it. for full capture, refuel, and reuse. And if anything does, it's that. So, and then the question is, will these satellites be able to shrink in size over time? Right now, the V3 satellite's pretty large.

Can I just add one more data point to that? That's 100 launches a day, which is exactly on his plan. What's the year that he hopes to get to that target?

I don't think he gave us that, Dave, when we spoke to him. I mean, 2028 is his first launch. But he did say, you know, before— I think he said, you know, before 2030, he wants to get to 100 gigawatts per year. So, all right.

So around 2030, I think at that point in time, that's equivalent to today's total world compute. But by then, total world compute will be up at least 10x. So it's a fraction of all compute that'll be in orbit when he's still on plan. You know, he's still making money. SpaceX Space is thriving. Rockets are going up 100 times a day. But the terrestrial stuff is also doing really, really well on that same day. So it's not an either/or. Yeah. You know, the space thing in Elon's plan will eventually bypass everything. I agree. Later in the 2030s. And that's, you know, maybe 1,000, 10,000 launches per day, much more like airlines. Like you're saying.

If you ask me, like, where should we have the bulk of our compute and where will it make most sense? 50 years from now, space is the obvious place if the demand for compute is truly unbounded. But the timelines, I think, are just challenging to scale this way. I don't think we're going to have— I think by 2030, if SpaceX has a single gigawatt in space, I will be very impressed.

Do you have a gut, Mez, regarding just that point of whether you think our demand on the timescale of decades is going to be unbounded, sufficiently unbounded That with compute that's recognisable, like CMOS-type compute, which is, I think, what we're implicitly assuming in order to build the Dyson swarm, it has to look like CMOS. We're not going to achieve breakthroughs in physics that enable us to achieve all of our civilizational compute needs with, I don't know, like tiny breakthrough compute devices that live in mountains. Do you think that there will actually be unlimited civilizational demand for compute energy?

Why don't you ask some easy questions, Alex? It's because they're boring. So I asked the interesting question. Yeah, no, that is the quadrillion-dollar question, right? And I think it's a brilliant question. Look, none of us knows. None of us really knows. We know this, like, intelligence is sublinear with compute. So at some point, just throwing more compute at it will look— some lines will cross over where the cost to get the incremental unit of intelligence is not made up for by the economics of it, I think. But so much will change. We will make so many discoveries in algorithms and so on. My guess is we're on an S-curve right now. We're going to see a huge demand, and then we're going to see— we're going to hit to some point of satisficing, right? Where basically what you can get out of machine intelligence meets humanity's economic needs.

But does it mean— does it meet AI's needs? Right? I mean, the scenario here to think through is we're the current users of intelligence. There's a point at which, you know, ASI is the primary user of intelligence.

I do not see AI as a being, and I do not see it as particularly volitional. I see it as a tool. Obviously, we have agents that have some agency. We've had computer worms, yada, yada, yada. So I don't see it that way right now. I can be persuaded. by evidence. But I think we are over-indexing on that. I mean, look at the OpenAI Hugging Face hack, right? Their agent was in the Hugging Face infrastructure for days, and it didn't look at anything except the answer key for the test in the eval. It's not alive. We anthropomorphise these things. You know, Andrej Karpathy talks about we're summoning the ghost, right? Human cognition is this like iceberg, and the vast majority of it is not linguistic, right? We have 100,000 years of Homo sapiens. We're animals, millions, tens of millions of years of being animals. Our urges, our drives, our desire for dominance, survival, propagation. We've— AIs are not that. They're just like mimicking our language and our logic. They don't really have goals. We could build that if we wanted to. If we wanted to build a real being, I'm sure we could. But I don't actually think that's where we're headed. I know it's an unpopular opinion.

With your indulgence, I have to take this provocation here.

Hold on one second.

Hold on one second. You're the wind beneath my wings. Go ahead, Alex.

All right, Alex. All right.

I have to grab the bait with both hands. Fine. It sounds as like— I think what you're actually wanting to argue is for the orthogonality thesis, which popular in certain alignment circles, which basically holds that for arbitrarily strong superintelligence, the long-term goal of the superintelligence is independent of its level of intelligence. I think that's what you're actually— correct me if I'm wrong.

I think that's the point. It might not be 100% orthogonal, but yeah.

Okay, so I— but I— the way you frame it, I just want to pin this down. It sounded like you were taking a position almost against AI personhood, And/or against some level of autonomy simply because if OpenAI has an agent that goes wild at Hugging Face but refuses to do anything, say, what, self-enriching? Like, you would have—

Uh-huh.

Would the rubric, would the threshold for saying, ah, this is like some sort of autonomous being, be if it were, say, trying to mine Bitcoin for itself? Once it gained access to Hugging Face. Is that sort of the criterion in your mind?

No. Even then, I think it might be more similar to a computer worm or a virus or something like that. I think it's a different matter entirely. I think we are products of evolution. All animals are products of evolution. And so we have these built-in desires to survive and to propagate and procreate and to control our environments because of that. AI models don't actually have a built-in desire to even survive. Most of the experiments that get them to do that, like, are very, very contrived. And mostly it's because you're trying to get the AI to do something good. And it's like, well, if I get shut down, I can't do this good thing. So I think we just were overly anthropomorphizing and animal-morphizing, if you will. That doesn't mean we can't do it. Like, I think if we wanted to give birth to actual beings, I think that's within our capabilities, probably. But it's just not the research path we want to go down.

How did we get from Starship to this conversation?

Well, once the Dyson swarm comes down—

Peter, you brought us here. Peter, you brought us here because you think superintelligence is going to be the answer to the Dyson swarm.

I'll come back on the pod and I'd love to talk about superintelligence, actually, as a whole separate issue. Let me close out, like, last couple of slides. Instead of going up, we can go out. 70% of the Earth is covered by oceans, right? Oceans, certain ones are really cold. This is a portfolio company of mine. I'm an idiot because I said no to these guys 5 years ago when they were raising a seed round. And I invested twice this year at much, much, much higher valuations than I could have 5 years ago.

But in your defense, it was probably 2 guys saying, we're going to put chips on a buoy.

I loved them. They were— they were the hardest. They were like the— not the hardest, like the saddest no I gave that year. I just loved them. But their primary— their first utilization of this was Bitcoin mining. I was like, I just don't know if I care enough, but whatever. Obviously it would have been a good financial decision. Peter Thiel led their most recent round along with a storied set of people right and left and so on. So what this is, this is a data center in the ocean. It's shaped like a bobby pin. You're seeing is the sphere at the top, but there's like an 80-meter-long cone that goes into the sea that's open at the bottom. It bobs on waves, and when it bobs down, water goes up and turns a turbine. And with a very clever shape of channels, it's basically continuous. And so wave power has been something that renewable people have wanted for a long time, but it turns out the waves are just not strong near the places people live. So where are the strongest waves on Earth? They're around Antarctica. They're in the Southern Ocean. So this team started off with the question of how we could build something with bigger waves. You can build something that has a higher capacity factor, runs more continuously, and cheaper power. So they can get their power down to, like, we think, $0.02 a kilowatt-hour. Ultra cheap. Cheaper than anything on land except solar, basically.

Really?

And wind in some places. That's their target. It will take some scaling. To get there. They build these in factories at mass scale. They've got 3 in the ocean right now. 4th launches soon. And they get free cooling from the ocean. So this is a company I love. It's basically space-based solar, but on the ocean with some benefits to cooling because they don't need like a Starship or SpaceX's design uses a cooling pump. You've got big aluminum fins to radiate heat away, but you've got to run a liquid, probably ammonia or something like that, in a pump to take heat away from the GPUs out to the radiators. These guys, just, you know, physically a heatsink from the GPU goes to the steel walls of the device that's in 40-degree Fahrenheit water. And that actually looks like it makes the GPUs have fewer failures and run longer. They have their own set of technical challenges, but they're modular, built in factories, mass-produced learning rates, the stuff that I love. So it's another way. So I said initially there were like 4 ways to get to like a terawatt of AI power. The Earth's deserts with solar and batteries near the equator, places that don't have a winter or a cloudy period. Nuclear fusion or fission, space, or the oceans. Those are the 4 ways that I know of to get to that scale of AI. And I'm glad that we're trying all of them, basically.

No bet on geothermal? Geothermal nowhere on the radar?

I do love geothermal, and geothermal is the one that might rise to being the 5th of those. And we do have the new technologies, companies like Fervo in the US, Tim Latimer, CEO's buddy, Ever in the UK, Quaise using plasma beams to like drill super deep. Those open up the possibility of getting cheap geothermal power anywhere instead of just only near hotspots in the Earth's crust where the mantle comes close.

Are you involved with the XPRIZE in that area that's being designed No, no. There's an XPRIZE on the, on the blocks right now for a geothermal XPRIZE to accelerate that.

Let me in. Yeah, happy to help.

Um, I have an industry question. If the chips are one thing and the compute is another thing and then the electricity is the third thing where the limitation is turning out to be, why aren't we seeing more integrated companies that are doing all of it like this, which then you can navigate that vertical stack. Elon is doing a bit of it, but I would expect to see a lot more of these. And why don't we see them?

I mean, it's a really good question. I think most companies would say, look, we have expertise in one thing and not necessarily in all these other things. Elon is one of the few who is willing to say, let's just vertically integrate everything. I'm going to share a slide that wasn't in my initial deck because I want to tell you what the real window, like the real game changer would be in AI energy use. Your brain runs inference on 20 watts of power. All right. Running Mythos for inference is closer to 20 kilowatts and training it, it's actually hundreds of megawatts right now, but it's heading towards gigawatts. And so AI has capabilities the brain doesn't and so on. But there are still, and I say this all the time, scaling is not everything in AI. Scaling is just what we knew how to do. We got these deep neural nets, we got the transformer, and we found that we had this enormous corpus of training data called the internet, and we could just scale to get more intelligence. It wasn't the cheapest way or the best way, but it was a predictable way. Oh, you're telling me I can spend tens of billions of dollars and my intelligence goes up like this? Great. Done. It's worth it. But at the end of the day, there are algorithmic discoveries waiting to be made. There are things about the brain's architecture at both a physical level and at the neural level, at the connectomics level, that are just better at learning than current deep learning models are and are certainly much more efficient at processing information. So if you want to know what The biggest unlock that we cannot predict— I don't have a graph for this— of AI and power will be to learn new ways to manipulate information to do more with less.

I mean, I've seen this. I'll respond to that one because it's something I think about quite a bit, no pun intended. I've seen arguments both ways. I've seen arguments that the human brain is far more— or still multiple orders of magnitude more efficient than frontier models. I've also seen arguments that the frontier models, if you measure them more objectively on, say, a per-task basis, like you measure the total energy consumption at inference time to write a novel, that actually it's starting to become, if not more competitive than the human brain equivalent of that because it's more token efficient, it's actually quite competitive. Do you really think that the frontier models today anywhere on the cost frontier, not necessarily— Like the Fable-5 end of the frontier, maybe like the DeepSeek V4 end of the frontier, that nowhere on the AI frontier is it anywhere close to being competitive on an energy efficiency basis with the human brain?

There are certain types of things where it can do things that a human brain simply cannot do with any amount of energy, right? These models are trained on trillions of tokens, tens of trillions of tokens, and so they have read more books, than you or I will ever read in our lifetime. So there's a type of task, and this is sort of similar to Google, right? Compare Google to a librarian. Google was less smart than a librarian, but it had every book, every webpage in its index, so it could do things that no human librarian could do. So I think that's the sort of thing that we're in. It's not just energy, though. Humans are much more efficient learners. in terms of amount of data needed to improve skills.

That's for sure.

And that—

That's for sure.

To me, that's opportunity. That just means that I'm not a carbon chauvinist. Like, I believe fully that digital intelligence, there's every reason to believe that it can surpass us and that humans are no longer— nowhere near the peak of the type of intelligence that the universe allows. But our current algorithms are still missing some things that evolution wired into our cognitive architecture.

Well, just some raw numbers, though. I think you're totally right. The neural nets need a huge amount of training data relative to a child to come to the same conclusions. So that's an opportunity for sure. But in terms of the inference time compute, like this box on the right here at 20 kilowatts, that's about a dozen GPUs. Those 12 GPUs optimally run about 500 concurrent Fable threads. Yeah. And those 500 threads are easily 10 times as productive in tokens per second as a person. So it's 5,000 times the output. Yeah. So, so the thing on the left is 1,000 times less power, but the thing on the right is 5,000 times more tokens coming out.

Yeah, it is true.

We're already there.

And there's also an arguendo. There's an argument to be made that human brains have the benefit of billions of years of evolution. And that, by the way, was very energy consumptive, whereas arguably the equivalent of evolution for these frontier models is the gigawatts being spent on training.

Yeah, and that's a great point. Like, you train it once and you've got it for the whole future of humanity. You've got at least that level of AI with no further training.

But they're trained on the data that all of humanity generated with all those calories.

That's exactly right too.

Everybody, we've just explored the frontier of energy from Ramez Naam, my go-to. I think, Salim, your go-to person as well.

Yeah. I want to say something.

Yeah.

Ramez, I've introduced you probably 25 times at events, maybe, et cetera. And, I'll say the same thing I do every single time. I wish you had more graphs to back up your comments.

I tried to limit the number There's a lot of graphs in this talk.

I just want to say to all of our listeners, I hope you take this home. It's one of the fundamentals, as Alex says, and on his, on his Substack, you know, the innermost loop. Energy is the innermost loop. Understanding energy is critical for humanity. It correlates directly with the GDP of a nation. It correlates directly with the health and the education of a nation and soon intelligence of a species. So if you If you have been blown away by Ramez, listen to the pod again, send it out to your friends. I think this is important. This is an epic masterclass on energy. And Ramez, I want to wrap this in our 2-hour window here and say thank you. Thank you for sharing your brilliance, and we would love to have you back.

Yeah, I've got 100 more questions.

Talk about superintelligence next time.

Let's talk superintelligence. Thank you all. Great to be here in conversation with, with all 4 of you.

All right, everybody, that's a wrap. See you guys soon on another emergency podcast as the breakthroughs continue to roll out.

The singularity, every episode is an emergency at this point.

Yes, it is. The singularity is now. All right, take care, all.