00:12:39 · Scott Nolan
[00:12:39] S2: I think to your point on the carbon emissions piece of burning oil, I think people don't understand the... Like, it's it's underappreciated what the physics is, and it's... If you have a very long chain hydrocarbon, the way you can envision it, just to elaborate on this, you've got a long string of carbons down the middle, and you've got a pair of hydrogens for every carbon. In the limit case, it's two to one. And then you look at natural gas, methane, and it's four to one. And so that's how you get four times the energy... Well, four four carbon hydrogen bonds versus two, and therefore, half the carbon emissions per unit of energy. So that's how The US has actually brought down carbon emissions for people who care about carbon emissions over the last few years by expanding natural gas production. And so that has been the story, I think. Compute has ramped, as electricity production requirements have ramped, natural gas has been the story. It's been fueled by fracking and by turbines. And so now I think the big question is, where do we get all the turbines over the next few years? Will there be shortages? What happens if there are? Do we cannibalize some turbines that are currently at power plants? Do we swap in solar batteries? Are there things we can do that create portfolios of energy? And then what's the transition to five years out? And as we do these upgrades on nuclear, as we keep the plants on, as we do restarts, as we upgrade other things, okay. Can we do that all in time fast enough, and can we bring the nuclear on fast enough, new nuclear, to power the growth? You look at the grid today, and I think it's about 4,000 terawatt hours per year of consumption in The US. The growth projected by end of decade for data centers alone is about 1,000 terawatt hours. So we've got to grow 25% plus by end of decade. That's going to be really hard. The only way you do it is what we have available today, and you just go into as aggressive production as you possibly can, in most cases behind the meter, is what I think. But a lot to unpack there, but I think it's gonna be a complicated story the next few years, and it's gonna take everything. And to put additional numbers to some of Secretary Wright's bounding on, like, okay, we had hundreds of gigawatts free, and data centers are expensive.
00:14:52 · Scott Nolan, Ty Morse
The rough rule of thumb is for a one gigawatt data center, just for a very, very traditional data center, it'd be about $10,000,000,000 so about $10 per watt. For the new advanced ones, it's $40 $50 $60 sometimes. And so you're talking about, to even harness 100 gigs, you're talking about trillions of dollars. And so at some point, do run into capital constraints of how much we can actually deploy. Rewind a couple years, you look at a company like Anthropic and their growth and how much money they can make on a gigawatt, it's maybe $25,000,000,000 I think, something in that ballpark, just for round numbers. And we're talking, Okay, we want to go from tens of billions of revenue. Now we want to go to hundreds of billions at these growth rates, which are demand driven. And you're now talking about, Okay, we need to go find tens of gigawatts. Where are we going to find them? And I think the story over the last five to ten years has been people finding those isolated pockets of supply of energy. Lots of companies have been doing this. First, they were doing it in Bitcoin, then they're doing it in data centers. And that's mostly been gobbled up. And so question is, where do we get it? It's gonna be all new production. And I think that's the challenge for for the industry over the next five years. [00:16:07] S0: For just turning on enough power over the next five years, what does that actually require? Like, is that just gonna be us turning on a whole bunch of nat gas plants? Does the next five years actually look like?
00:16:18 · Chris Wright
[00:16:18] S1: So gas will certainly be the dominant source. I mean it's just by far the cheapest, fastest way to produce new firm, meaning 20 fourseven generation. As Scott said, we've lowered greenhouse gas emissions in this country not because of subsidies or government policies. Dominant amount of it has just been natural gas has outcompeted coal in the power stack. It's just gas plants are cheaper to build and they're cheaper to operate They're more flexible. They can turn up and down faster. Now don't get me wrong, love coal. It has been the largest source of global electricity since the data started in 1900 and it will be the largest source of electricity globally when I die. That's a long ways away. Natural gas is number two and in rich countries it's usually the reverse. Natural gas is number one and then coal or nuclear will be number two and the other will be number three and those three things are what run modern power grids. But on the natural gas side in The United States, gas turbines as Scott mentioned, they've gotten tight supply. So people say, Oh, if you sign up now you can't get one for five or six years. Well that's true if you get in line now but the deliveries in the next five or six years of turbines that are already ordered and already under construction is probably close to if not over a 100 gigawatts. So there's a massive amount in construction. And of course it's not just turbines, there's gas reciprocating engines that are used to produce, think diesel generators that now can run on natural gas. We are going to see gigawatts and then tens of gigawatts of reciprocating engines that burn natural gas to generate electricity. I used to have those at my former company. We're also seeing fuel cells, which is another way to take natural gas and turn it into electricity. Fuel cells, something that's been talked about for years, that's finally scaling up and hitting scale. I think we will certainly in the next five years we will see multiple gigawatts of new electric generation from fuel cells which also take natural gas as their input fuel. The one thing I should say just so people cannot worry about this one is this supply of natural gas, the fuel itself is no problem whatsoever.
00:18:32 · Chris Wright
Today in The United States' biggest oil field, the Permian Basin in West Texas and Southeast New Mexico, oil production is curtailed a little bit because we don't have enough capacity to take all the associated gas. When you produce oil, you always produce natural gas too. Just the shortage of pipeline capacity to move natural gas out of the Permian Basin, so it's like 2%, 2.5% of US natural gas production, that shortage is enough to produce about 15 to 20 gigawatts of electricity. Just the gas we can't find a home for. So gas is cheap. It's wildly abundant. If you want to go to a different location you may need to build a pipeline or some transmission thing. But The United States has an amazing abundance of natural gas. People often think that the cost of electric generation, it's about the fuel. We want fusion because it's water and it's really cheap fuel. Well, there's not a lot of tritium in water. But Scott knows in the nuclear space you've got to have that affordable reliable fuel but mostly the price of electricity is dictated by the price of the machine. It's not the price of gas that determines gas generated electricity. It's the price of the machine that turns that gas into electricity or the machine that turns that coal into electricity or turns that uranium into electricity or turns that deuterium and tritium into electricity or turns that wind and sun into electricity. It's the cost of the machines that matter. The problem, again, with the intermittence is they don't produce a product anyone will buy. They produce electricity. Some of the time, we're not sure when. So to make them real machines, it's just enormously expensive to build giant amounts of storage to last a few days where you can have cold spells or hot, you know, lit darkness or high pressure masses where the wind doesn't blow.
00:20:22 · Scott Nolan
[00:20:22] S2: Yep. Yeah. And it depends on how many 9s you're going for. So what sort of uptime do you need? If you're trying to do a high end reliable data center, you're looking for more than 99% uptime, 99.999 often. And so you can't afford to have the power source go down. That's why we think a lot about baseload, then baseload being really the dominant thing that you have to solve. You can get there with enough of a portfolio mix in certain geographies, but it's often getting to those same nines through a mix of different things with all the storage you might need or all the peaker turbines you might need, that ends up being just as expensive. So I think, yeah, it's exactly right. If you're trying to achieve baseload, it becomes all about the upfront CapEx, not as much about the OpEx, even for nuclear. The bigger the nuclear plant, the lower the cost of the fuel relative to the cost of energy. If you look at the classic huge gigawatt scale nuclear plants, the fuel is in the single digit percentages of total cost. The smaller you go, though, the more you solve the machine problem of actually building the plan and bringing that cost down, then the fuel becomes more expensive, which is why we're not just focused on schedule and being online this decade. It's how do we also bring the cost down over time just so The US can compete on baseload, just across the board?
00:21:40 · Ty Morse
[00:21:40] S0: There's this interesting idea that people are not very good at understanding exponentials. And right now, we can pretty well extrapolate the growth of energy demand for the next few years. But I think it's interesting to think about, like, what if we're massively underestimating what is required and what the demand might be? What are the biggest, like, constraints and things that are stopping us from moving faster on just turning on more generation?