Unlocking stranded renewables with Bitcoin and data centers with Brock Petersen (Satoshi Energy)

Welcome back to Transmission Today, Q is joined by Brock from Satoshi Energy a company pioneering the co location of data centers with renewable energy assets to unlock otherwise uneconomic energy In their conversation, they explore two major themes First, how Bitcoin miners and AI data centers differ dramatically in their requirements, from their flexibility to their vastly different footprints and cooling needs. And second, the massive load growth hitting the grid, with ERCOT seeing one hundred and thirty five gigawatts of new load requests compared to its eighty gigawatt peak demand. One thing I found particularly interesting was Brock's contrarian view, but I'm not gonna tell you what it was. You're gonna have to listen to find out. And as always, if you're enjoying transmission, please like and follow wherever you get your podcasts. Now, a reminder, ninety percent of our audience is already subscribed. So, if you want to join the crew, go ahead and hit that subscribe button.

Now, let's jump in.

Brock, welcome to the podcast.

Yeah. Awesome. Thanks for having me. Super excited to be part of the conversation.

Alright. Let's jump into Bitcoin miners and what their power profile looks like. And if you can set the scene, when you started Satoshi Energy, so six years ago, two thousand and nineteen, what was the landscape for Bitcoin miners then?

Yeah. So this is actually super interesting. So we started the company Satoshi really as Bitcoin mining was going from a hobby of scale. So that's you know, I have a mining machine in my basement. And peep the returns were at a level to where you could be on a retail electricity rate and still be making money.

I remember people having miners in their sheds, making a lot of noise Yep. And miners in their sheds. Of course, you can't do that.

No. That energy I think there's, like, one misnomer, and we'll kind of dive into this is right now, if you look at what the revenue for a miner is, it's in the kind of hundred and thirty five dollars a megawatt hours. So, really, you had need to have pretty cheap electricity to be making money as a as a minor and, you know, paying off that CapEx. But so back then, you know, it was hobby of scale. You know, maybe you started getting, you know, kind of, like, one to five megawatt is it started growing a little bit.

And then really in that half twenty nine twenty nineteen, twenty twenty time periods where we started seeing it go more industrial scale, which at the time, our first project, was a hundred and forty megawatt Bitcoin mine collocated with a hundred and seventy megawatt wind farm, and this energized Wow.

That's still pretty big.

It was at the time, it was massive. Right? I mean, a hyperscale data center, it's like a Google data center at that time, it wouldn't even be a hundred megawatts. So it's for to do a hundred megawatt load is was huge and really kind of showed that, you know, Bitcoin mining very quickly went from the hobby scale to the more industrial scale.

And Where where was that project?

That project was in West Texas, so a little bit south of Midland.

Okay. And that was Did it did it get built?

It was built. So it energized in twenty twenty two and has been operating ever since. So the Awesome. Congrats.

Yeah. So the fascinating thing about that project, it it has a lot of history to it. So it originally was an Enron wind farm. So one of the first wind farms with land royalties in the state of Texas.

It was also on a piece of land. If you go back even further, it was, like, where so the original shale was found in that in that region of Texas. So you had to you know, this land was some of the first, oil and natural gas, one of the first wind farms, and then kind of this first co located Bitcoin mine. And, really, the key kind of insight for us coming from the, you know, clean energy space, renewable energy space is you have wind solar assets are dealing you know, so in California, it's mainly solar dealing with, you know, do the duck curve and negative prices in the middle of the day.

For Texas, where we just built more wind, you have also negative pricing, congestion, curtailment issues, also basis. So we were able to come in and create this win win opportunity by having the data center take on some of that otherwise uneconomic energy and provide more economic uplift for the wind asset. And this is actually a recently repowered wind asset. So they were still able to keep their production tax credits, keep their racks, and then have this data center come in and create accretive revenue on a nodal basis for the wind farm.

So just to clarify, so there's there's a how many turbines is that?

So the number of turbines on that project, repowered in twenty nineteen, they're all about a a little over a a megawatt wind turbine. So it was, like, you know, right around a hundred and ten, ten, a hundred and twelve.

Hundred and ten turbine. Hundred and twelve turbines. So this is a big project. Yeah. Big project. You're saying is that by putting a Bitcoin miner on that site or next to that site, it added additional value to that wind project.

Yes. So what we do this is where we kind of get into the insight that we had, and this is one of my favorite things about our job just from a creativity standpoint is by doing this and and the way that we do it, we actually have net metering, meaning the wind power that's produced and consumed by the data center, that never hits the ERCOT, hold settlement meter, what it's called. But think of it as, like, the meter to the market. So because it's it's being net metered behind that, we effectively get to create our own energy market.

And ERCOT allows the rules where we can do that any way that we can imagine. Generally, for a lot of our projects, the wind asset or solar asset has VPPA or hedge already on it. So because of that, it kind of limits some of our creativity, but we're our own little mini market behind the meter between the buyer, the, you know, Bitcoin miner, and the seller, the energy company or the the wind asset, solar asset owner. Generally, what we've seen to work the best for both sides has been a floor price PPA concept.

So let's say the nodal price, you set that floor price at ten dollars a megawatt hour. So the nodal price goes to negative twenty. Well, that never hit the market, so we're gonna clear it at ten dollars. That gives that nodal uplift to the wind farm.

And then whenever it's above that, the data center is gonna pay whatever the nodal price is, which a load in ERCOT normally pays the load zone price. So we focus on finding the lowest price nodes in Texas, and then we try to set that floor price to create a win win for both sides to where the data center is paying slightly less for their electricity, the wind asset's getting paid more or solar asset, And then also the data center gets out of it the speed to market. So there's an existing project substation that they get a leverage to actually connect. So instead of having to connect at, you know, three forty five kV, they're connecting at thirty four point five kV on the low voltage side of the project substation.

So their speed to market's quicker. Their energy cost is quicker. So we really create these nice win win scenarios.

Okay. And let's just take stock what's that project called again? Sorry.

That project's, Stargazer is the data center.

Stargazer. Yeah. It is a beautiful part of the world. I mean, if you the I mean, if you ever get a chance to go to West Texas at night and look up, it is just breathtaking.

It is.

Went with my family to Big Bend, a few months ago, and it was just one of the most magical moments. But, so Stargazer, the project, when a hundred megawatt, I noticed that you're mentioning data centers, and we're talking about Bitcoin miners in the same kind of Venn diagram circle here. So how big is a hundred meg Bitcoin miner? It like, for if you looked at a picture of one, how big is one of these things?

Most of the audience on here is you know, understands inner or energy storage development really well. Bitcoin miners have a super similar footprint to what a battery project would. So you're looking at about ten megawatts of Bitcoin mining for every acre that you're building. So that hundred and forty megawatt project's about fourteen acres.

I mean, we have some auxiliary buildings, etcetera. It's closer to twenty acres, but that's really it's a pretty tight footprint. Two quick things to kinda go back on. Stargazer is actually a type of fish.

So, what we so, again, starting off in the Bitcoin side is Bitcoin miners, you know, come back to my point earlier, when you when your revenue is a hundred and thirty five dollars a megawatt hour, you're, you know, shaving off the top, you know, five percent of hours are profitable for you to be pulling energy because the energy is too expensive. So what we call it is kind of bottom feeders in the energy market. So all of our projects are named off your bottom feeder fish. With that, you know, we kind of we fill the valleys without adding to the peaks, and that's really where kind of Bitcoin mining is this great technology for load customer for a grid, because they're able to sell energy when energy is cheap, and then they're not adding to those, you know, scarcity event times.

That's one of the things that probably doesn't get talked about enough, which is that I know it's different with data centers because you have a amazingly high uptime requirement. But for Bitcoin miners, it's a little bit different in that if the power is too expensive, which tends to be when the system's under stress or on your most carbon intensive periods, you just switch off the Bitcoin miner. Right?

Right. Most Bitcoin miners today are operating on nodal price contracts, so that means they're just a price taker.

They do that, one, because that's you know, their hedge is effectively their ability to shut down. So if you have that ability on, you know, in a five minute notice to be able to ramp from hundred and forty megawatts to zero, that is your hedge. You aren't having to hedge and then sell back into the market or make day ahead decisions. You're decisions. You're able just to leverage your own flexibility.

But what are the ramping characteristics then for a Bitcoin miner? And is it a square wave? Because so, okay. You got five minutes notice. You you've got a lot of power there, a hundred and forty meg of demand that you're gonna switch off in five minutes.

Thinking about spinning plants or anything thermal, that's quite a tricky thing to do. I know peaking plant can turn on and off very quickly, but can you just talk about ramp rates for Bitcoin miners? Yeah.

So the ramping profile of one of these assets looks a lot different than, you know, a a traditional data center. And, really, one of these Bitcoin mines, at the end of the day, I've been using the two terms interchangeably because a Bitcoin mine is a data center. It's just with less redundancy, less resiliency.

And we'll get into some more of those differences as we go into the AI. But at the end of the day, it's a bunch of power electronics. So shutting off those servers, yeah, you wanna shut them down, you know, softly so that there's not a hard stop, but you can do that in seconds. And then, you you know, you'll have some thermal soak back, which to give you kind of former lives, I've been in power electronics.

So that's if you shut something off and the fan also shuts off, you'll actually see a temperature rise, and that's your soak back before it starts falling. So you'll wanna do things like keep those fans running for a little bit longer so that you don't have thermal or temperature kind of shocks to the system. But you can shut down in kind of grid. In power electronics terms, it's a very long time.

But in, you know, grid infrastructure, you know, doing things in under five minutes is very quick. So you're they're able to kind of ramp up and down quickly. This also creates things that we have to work with, you know, ERCOT and others on because in the same way that you don't want, you know, a hundred and forty megawatts to go from zero or a hundred and forty megawatts to zero or zero to a hundred and forty megawatts. So we're starting to work on some of those ramp requirements, but the system itself can can ramp up and down very quickly.

Okay.

So it's more the, ERCOT's problem of managing the system where you wanna you wanna introduce some ramp rates, but the system itself, no issue there turning on and off within minutes.

Yes. Okay. So then let's continue on your story then. So you built this Stargazer site, which is, for a first site, is very large. Yeah. And that was six years ago. Then then what happened?

It's been super fun, you know, building the company. So we you know, really, at the end of the day, you know, we're the leading data center developer of colocated assets. So our, you know, niche in the world has been doing everything where we partner with IPPs to create these projects. So we have almost five hundred megawatts of data centers that we've developed that are online and constructed and operating.

We have another seven hundred megawatts that are gonna be starting, construction here pretty soon and then a little over a couple gigawatts of just earlier stage development assets. I will say the best way to think about our company is, you know, the data center term for it is powered land. But what we're doing is is the way to think about us, I think, for your audience is we're this we're look very similar to a greenfield battery developer, a greenfield solar developer, where we come in, we do, you know, the entitlements, land, we get, you know, go through the load interconnection process, which looks a lot like, you know, the generation interconnection process.

And then once we get that project to, you know, notice to proceed, construction ready, that's when we kind of flip out of the project and bring in that data center company who ultimately builds, owns, and operates it. But because we create these very unique energy contract structures, we do we are the the broker for those energy contracts and then stay in, kind of, as I was talking about, some of our software earlier. So that's how we maintain the relationship with the project. So we don't it's not a true pure play developer where we just flip out. We actually maintain, kind of continuation in the project, but we are that kind of greenfield developer that is kind of our niche is this colocated world.

So these projects are your babies, and you're not giving them up, basically.

So alright. So tell me about the evolution of your business model from Bitcoin miners to I know, data centers is part part of the same thing, but particularly for AI applications.

What, what's happening in the market? And can you talk about the scale the difference in scale there?

Yeah. And I think too is it's helpful to kind of go through, like, what are just some of the differences because that's really the thing that we've been on the forefront of is you have Bitcoin mines that are flexible. You know, ninety five percent uptime is acceptable for them versus traditional data centers and all these AI data centers. They're looking at, you know, five nines of reliability or six nines where it's you know, they wanna be up ninety nine point nine nine nine nine percent of the time.

So Wow. And what is that in practice? How many minutes a year is that?

Oh, it's it's it's not minutes. It's it's it's seconds.

Wow. And that and really what they're looking at is, at the end of the day, you kind of have to think, what is your user? Right? So on Bitcoin mining, you know, a new block is added to the the blockchain.

Just think of it like you have a big Excel sheet, and every ten minutes, a new row is added to that Excel sheet that says, hey. Where did money move? That is largely what kind of the Bitcoin network, and that's what the Bitcoin miners are doing is they're all trying to add that next block. And if you let's say you're down for ten minutes, you miss out on that block, but you can come back up and you can immediately start working towards the next block.

So every block is a complete, independent, and unique. So everything is can be done in a way to where you don't have to start back from the beginning.

If you are let's say that you're from another kind of just perspective, you're a cloud, you know, provider and you are, you know, helping run Google Maps. You know, you don't want Google Maps in New York City to go down. Otherwise, everyone just should be walking, have no idea where to go, where to get on the train, etcetera.

It's hard enough getting around the city with Google Maps.

That's true.

Let alone without it. Yeah. K.

Yeah. We've all lost that part of our brain to be able to look at a physical map. So that's, you know, that's an example of a compute load that you needs that kind of high uptime.

Really, AI falls a little bit kind of in between because, generally, that's also a compute process that's parallelized.

You know, it can withstand a little bit more up and down, but their revenue is such that you don't really wanna be down, because their revenue per megawatt hours, you know, is in the thousands of dollars. So they're gonna build Wow.

Is it?

Yeah. So they're gonna build infrastructure to so, like, whether that's, you know, diesel backup generators where if something happens with your grid connection, you'll fall back to a UPS that will then take you to your diesel generators to you know, your cooling system is gonna be built with redundancy. Your feeds your electrical feeds are built with redundancy. It's just so much money goes into redundancy, and it's almost like the grid.

Right? If you when we do, work with the transmission system provider to do load studies for one of our projects. They're, you know, looking at n minus one, n minus two, n minus one, minus one, all these contingency scenario planning to make sure the grid can keep the load online. That's what you're doing when you're designing a traditional data center.

So just they are two very different designs. But at end of the day, they're from, like, your original question was a scale perspective. You know, what we've done with Bitcoin mining, which, you know, everything we've done there has been between a hundred and two hundred and fifty megawatts. You know, now the smallest AI project we're working on right now is four hundred megawatts.

And this is it is mind boggling large. Right? Like, in my, you know, fifteen year career in in the renewable energy energy industry, it's been a flat load growth. You've never we've never seen load growth in the way that we have. And I think, you know, right now in, you know, ERCOT, for example, we have over a hundred and thirty five gigawatts of load being requested, which to put that in perspective, ERCOT's peak is about eighty gigawatts.

So you have more than ERCOT's peak is right now in that interconnection queue.

Grand For data centers.

For data centers.

Wow. So, yeah, we should probably just that is huge.

It is.

And so and just a couple of comments here.

Whenever I see these mega data centers, the amount for this redundancy requirement, the backup power they need, In practice, when you look at a satellite image of one of these or a planner's design, you're looking at what looks like a warehouse as a data center, fine, and then you just see rows and rows and rows of of of stacks of, like, basically chimneys, right Yeah.

For backup power gen.

And it really is quite an assault on the senses to look at it and and and realize how much power generation these these things need locally.

And then, thankfully or hopefully, they will hardly ever run those units.

Right. That's there's a lot of paradigm shifts happening right now in, I think, the traditional data center world, which, again, you know, that I kinda came out of that, you know, early in my career, you know, doing energy efficiency in data centers. It at that point, there was access to power and sites that met every criteria they could think of. So, you know, you wanna be close to labor. You wanna be close to fiber. You wanna be close you know, have access to water for cooling.

You wanna be able to Are they water cooled?

Yeah. Generally, they are. For redundancy and, like, those generators, you know, you need air permits to be able to run all those diesel generators if you ever need to. So they were able to kind of have their pick of sites, and you just you saw concentrations in certain areas because that labor force is so important.

And, really, now in the paradigm of where we are, where access to power is scarce, that's where you're seeing the AI data center say, hey. Do I actually need all of these things, or can I be more creative and have more site optionality? So that's been the exciting part is instead of thinking of your backup as something I never want I actually never want to run and as side of, like, the once a year I test it. And now it's kind of like, okay.

Maybe I can get online quicker if, you know, a couple hours a year, like, I remove my load from the grid when there's a scarcity event. And, like, for me to do that, I'm gonna run my backup, and I'd have, again, this massive amount of power that never gets used. So I think that's one of the exciting things that we're doing right now is actually looking at what are creative ways to leverage all of this infrastructure that's being built, not as just backup, but is also kind of can I run it at strategic times to help the grid? And maybe it's I can get online quicker.

Maybe it's I can reduce some of my energy cost. Maybe I can participate in some of the same, you know, ancillary programs that batteries are participating in. So I think we're gonna see continue evolution of the thinking there. And I think right now, the forcing function is that access to power, which is the scarce thing in the market.

Yeah. I mean, I guess in ERCOT, you only really need to run ten, twenty days a year for a few hours, and you can avoid all the messy peaks or even take advantage of the messy peaks.

Exactly.

Alright. So who's actually this is not a trick question. Who's actually building these things? How does that value chain work?

You know, everybody knows that Google and Meta and OpenAI and whatever are building in quotes data centers. But is it actually, you know, a team at AWS or Google out there building a data center, or is there an intermediary that owns the data center and then they rent it a bit like AWS? Or do they do you find actually that, you know, you hear about Meta signing big private wire deals and Microsoft signing private wire deals? Is that because they're actually now becoming asset owners?

They're entering the hard asset space.

All the above is is the short answer. So you, and then to be honest, that's how these companies have always operated is, you know, Meadow might have a few large, you know, very efficient data centers that they own and operate and run for special use cases. But, you know, if I'm pinging Facebook here in New York, it's unlikely Facebook has a data center here in New York. You know?

If I'm on Instagram, it's likely going to, like, what's called a colocation provider. So you have, like, your Equinix, Digital Realty that they'll have you know, Equinix has a large data center in Chelsea, and, you know, hyperscalers are some of their customers there. So they there's and, really, what your Equinix and digital realties are is their REITs, real estate investment trust, that they're providing space, power, and cooling, and then their tenant is, you know, people who have compute. So the Google's and Facebook will still own the servers in those buildings, but they won't own and operate that building because they don't wanna have a team out here in New York.

So that has, for AI, that's translated as well to where you have your Metas, your Googles, your Amazon's building sites, but you also have them using intermediaries who are building sites who look more like real estate infrastructure companies.

And it's really kind of in all the above.

And it's I think right now, it's just a bandwidth. Right? There is this race to get as many, you know, gigawatts online as quickly as possible. So if you can bring them a site that you can develop, they're you know, let's go forward with it. You know, if you're like, hey. I'm gonna build a site and get you on as, like, that core tenant, they're also doing that. So it's really all the above.

And which areas are hot for data center development? Of course, there's lots of talk about ERCOT and Virginia, for example.

What does certain regions what's so special about Virginia, for example?

What's so special about Virginia?

No. I love Virginia. Don't get me wrong. It's a gorgeous place. But what's so special about building data centers in Virginia?

Yeah. So it literally was just a concentration of talent and fiber. So really where you've seen the historical hot spots, so whether it's been, you know, around Minneapolis, you know, Virginia has always been the the bedrock, San Francisco, Bay Area. It really was kind of fiber plus cost of electricity plus, you know, where the labor force was because these, you know, getting kind of back to a Bitcoin mine employs a very small number of people and takes up a very small footprint, you know, that ten megawatts for every acre. And AI data centers closer to one to one, so it's gonna be an acre for every megawatt. So if you're working on a four hundred megawatt site, it's four hundred acres of of land that you're consuming.

Wow. So it's it's it's ten times bigger in footprint than a Bitcoin miner or a battery.

Yes. Okay. So they're massive. So that labor force matters. I think, again, where you've seen seen things move is access to power's king right now.

So if it's, hey. I don't maybe I don't have the best fiber there. Maybe I don't have the best workforce. It's okay if you you don't wanna have not everything, but, like, one or two things that that you know, it's shown that the data center owner operators are willing to figure out a couple things.

So what we've seen is, is super interesting. So about a year and a half ago, we you know, most of our original sites were in West Texas.

That's just, you know, where the best wind resource is. And you you kind of mentioned how beautiful it is out there with the stars. It's actually super I mean, it used to be the the, you know, floor seabed floor, and you have all these plateaus. And you have all the wind turbines are on top of those plateaus because they get you up an extra three hundred feet. So it's it's actually super cool to go look at.

Excellent for dinosaur bones as well. Exactly. These are dinosaur bones or a place that looks like Mars. Yeah. That's your place.

It is. So the on those plateaus, now we have these large data center campuses that are going up as well. But, really, Oracle was kind of the first one with this, you know, star gate project that Trump announced. Like, that is all happening around, Abilene.

So you've seen now it's like Abilene's kind of like another new hotbed. Once one company plants a flag, you just see a bunch of others come filling because they're like, okay. The fiber is coming. The workforce is coming.

You know, just takes that one company. So we saw year and a half ago where an Abilene project did not look super attractive to today. You know, every Abilene project you can get your hands on, the AI data centers are are willing to to kinda come in and build them out.

And what's the time to build look like for one of these data centers? I know Elon famously I can't remember the the time. He did the battery in a hundred days. Right? And they did that huge XAI project in, say, like, forty five days. Right?

Right.

Which was huge. How fast can you build these things if you're not Elon Musk?

So it it, even if you're not Elon, it is surprisingly fast. So especially for the projects that we bring to market and we develop, because the substation's already there, now all I have to do is the median voltage work and down. So we've the per the quickest project that we worked on from breaking ground to energization was about nine months, and then they ramped basically from zero to two hundred megawatts in about a month. So it is wild how fast these are moving. You know, it's not the hundred days, but still for to for, you know, nine months to be able to energize two hundred megawatts of infrastructure, it is people are definitely motivated in in figuring out how to move quickly.

Zero to two hundred megawatts of complicated data infrastructure. Well, complicated computational infrastructure online. It's just incredible, isn't it?

It is.

We should start a cable company.

We should. That's what, my brother-in-law actually works at a company that they sell metal boxes, metal enclosures for electrical, you know, gear, and they're, absolutely killing me.

They could have a good Christmas party.

But it is I mean, if you look at the areas that we're in, you know, in Texas, like, if there's any place in the US that's used to booms and bust and be able to mobilize a large group of people quickly, that's your area. Right?

And The Permian Basin.

Exactly. And looking for us right now, oil prices are are, you know, suppressed. Otherwise, we'd be competing with that market from a labor perspective. But, yeah, in the Permian, you can, you know, quickly get, you know, a a highly skilled group of people working on your projects, which are you know, data centers are definitely taking advantage of that.

So I wanna ask you about your special source then, which is the colocation element, which is what we're really here to talk about. Data centers colocated with generation.

How does the Excel spreadsheet work?

Yeah. So it's surprisingly obvious. It is as I think most things, it's the idea is easy. Execution is the hard part. So from, you know, the spreadsheet, I I kind of mentioned it earlier, but you have, again, wind farms either whether it's curtailment, congestion basis.

Is this existing wind farms?

I think it's That is most every project that we've energized so far has been with existing generation assets.

And this is actually one thing that's not super intuitive.

But the way that we structure our deals, they work whether there's an existing PPA on a project or not because we'll we can set that, again, that behind the meter where we're making our own energy market. We can price that data in a way to allow it to work with an existing contract. So we don't generally work on merchant assets, actually. We actually work on already contracted assets.

Just to reiterate, if I think I understand you, what you're saying is that for wind farms that are already existing and have already got a buyer of all their power, so all the power they generate or most of the power, they've already got a PPA with someone who's buying it.

Correct.

And you can come in alongside that even though there's an existing buyer. And then what?

Yeah. So this is the you get to have your cake and eat it too. So that existing contract that's in there, really, the purpose it serves is to finance the project. Right?

If developers do not have attractive balance sheet for the New York bankers to finance us. Right? We have to go sign those agreements, you know, with your Amazons, your, you know, Pepsi Cola, whoever, to be able to then go to the bank and say, hey. Instead of me, developer, with the, you know, bad balance sheet, don't you're you're never gonna finance a project.

But now I have this twenty year contract with a Fortune one hundred. Now I can go get it financed. However, that contract that I signed, you know, with Amazon to buy that power at, you know, the hub is a purely financial agreement. So it has nothing to do with the actual power.

And a lot of times, those PPAs don't even it's not even an as gen PPA. There's, like, a anemometer that's reading the wind speed. And based on that, I now have to sell a certain amount of electricity as the wind farm to that entity. So it is a purely financial, like, out of the actual energy market, the real time energy market transaction.

So what we're able to come into is structure the contract with between the data center and, you know, let's in this case, a wind farm. So the way that we'll structure the contracts is so that they're still based on the nodal price. Because if I have a contract for differences that I currently have with Amazon, I can't sign a a fixed price behind the meter. Otherwise, I can end up with wrong way risk where that hub is settling at, you know, a thousand dollars a megawatt hour, but I'm on for thirty dollars. So that basis can exist. So if we but if we structure the contract to where the wind farms already is always being paid with a market, AKA ERCOT would have paid them or more, then now they're in a better position for that basis than they were without the data center.

So you basically say, owner of win owner of Windfarm x, you're still gonna get paid what you thought you would get paid, and we're gonna do stuff around the fringes.

Exactly.

And we're gonna net it off, and we'll only play when it's worth it.

Exactly. And the rest of the time, you'll take you'll take the market price. Exactly.

Okay. Okay. So why so why isn't everyone doing this then?

It should be.

I think it's, it has been when we started the company, that was kind of, like, really the light I mean, like, everyone knows that projection or curtailment exists. Everyone would love to have someone come, you know, have a product to reduce that. But, really, as we're calling up, you know, of all of our friends that work at IPPs and kinda pitching this, having worked with, you know, a lot of these energy companies in the past, it is hard to get them to move quickly. It it was amazing to see.

It's like, oh, here's, like, you know, this project is now on this person. We wanna move forward. So, like, that part of the project is not the hard part. So, again, it's the execution piece is the hard part.

So how do you thread the needle to where you can sell that bilateral electricity that's behind the meter again? So we're making our own little energy market, but, you know, the wind asset, solar asset, they wanna keep their exempt wholesale generator status. So we have to structure this in a way to where we're not impacting that and putting more regulatory burden on them. If it's a wind asset that's, you know, built in the last ten years, they wanna keep their production tax credit.

So, again, how do we structure it and do that if they, you know, their recs, etcetera? And then from a credit and collateral standpoint, so I kinda mentioned, you know, we've also developed, you know, software around this to help from a credit standpoint, energy management standpoint. That was all bred out of doing these deals. And, again, if I'm I'm used to doing as the generation asset owner, I'm used to doing deals with Fortune one hundreds.

You know, now you're bringing me, you know, entities that haven't been in business for ten years, and I'm doing deals with them where they're consuming five million dollars of power a month.

So, really, we've we've had a there's been a ton of thought from a structuring standpoint in how to get, one, get the IPPs comfortable with, hey. This is this works. It's not exposing you to unnecessary risk, but it's giving you that accretive revenue that you're looking for. So it's been super exciting kind of you know, it's a puzzle that you have to figure out, and that's, you know, really what what we've been doing for the last six years.

It's extremely cool. A few questions about data centers then. So these data centers, what's the design life? Because if you think about a, you know, thermal power station is maybe has a twenty five, thirty year design life and then will go on longer than that, a battery, most of the equipment there is you know, your lease is probably twenty five, thirty years, but you might change out your battery cells after ten.

What's the what does it look like for a data center?

Yeah. And this is most things will respond with you. You one thing looks different for a Bitcoin mine than for an AI data center. And for both, though, the the commonality between them is you have your infrastructure that you're gonna put in the ground, and then you have your servers.

For in both instances, your servers are the bulk of your cost. Those, you know, have a functional life of around five four to five years, and then you're gonna refresh all of those servers. However, that, you know, infrastructure that you're putting in the ground, so the power, you know, cooling, building shell, that is all you know, you're underwriting that on that similar time horizon. So that's, you know, fifteen, twenty, twenty five years. Again, probably on the lower side for a Bitcoin miner on the higher side for an AI site, but then you're refreshing your servers.

So it looks a bit like a battery project. But rather than battery cells, you've got GPUs. Exactly.

But I imagine lots of the the the the balance of plan is what it is. You know, it's kinda legacy infrastructure. It's gonna be alright as long as you don't as long as you look after it. Yeah.

There's a massive cooling load. What's what's the cooling load like then? How much of that say your four hundred megawatt project. How much of the four hundred megawatt is doing maths Yeah.

As a computer, and how much of it is cooling?

So this is a fun, the data center industry has a very unique metric that they look at to say, what is my efficiency? And and that metric kinda gets to the question you're asking. So it's called PUE, power usage efficiency, and that is your total site load divided by the amount of load that goes into your servers.

So, ideally, it's one to one. Right? So a hundred percent of the power you're consuming goes into the suit the servers, the compute.

Obviously, that's not reality because you have things like cooling that in there. For a super efficient plant, you might have a PUE of about one point one. So that means for every one point one megawatt, you know, point one megawatt goes to cooling and other, and one megawatt goes into the servers themselves.

So it's a bit like efficiency, but with the denominator on the wrong side. Right?

So first first, factory people who are thinking about round trip efficiency and percentages, we've gotta kinda sit upside down on this one.

Exactly.

Okay. Alright.

Everyone's gotta make it, you know, the the to create a barrier to entry for new people. You gotta make things a little confusing.

Okay.

But, you know, typical PUE might be between, one point four and one point two would be a pretty standard one. And then for a Bitcoin mine so Bitcoin mines are called in very unique ways because they don't have the same thermal envelope that they need to stay within. So they're the servers that you use in Bitcoin mining have a much wider temperature range and humidity range.

Could you give some examples on both of those then?

Yeah. So in those, as opposed to needing, you know, thirty percent relative humidity and seventy five degrees might be a traditional data center. Generally, they're cooling it. You you know, if you ever walk into a data a traditional data center, you need a parka on.

It is cold. Yeah. For a Bitcoin mine, they generally just have, what's called a curtain wall. So it'll be a membrane that has water on it.

So just water's constantly dripping on it. You'll pull air through that for evaporation, so evaporative cooling. Obviously, that increases the humidity well beyond what you would put in a normal data center, and then that gives you the cooler air to go through. But not having to mechanically cool, so use a refrigeration loop, the air, that allows you to have a PUE that's really close to one versus a traditional data center.

You know, you have traditional AC. So you're using a refrigeration loop to cool air and also to reduce the humidity, and pump that through.

Clinical clinical science. Right?

Yeah. You have your hot air or your cold air. It's a lot goes into that.

Okay. So let me ask you a very basic question then, and this is a bit of a trap. But, surely, we don't have enough power for all these data centers.

Yeah. So I think that's where we're starting to push the envelope on that and also push the thinking on that. So you had a good article come out or study come out from Duke Duke University that was on, hey. If you just added a couple percent of flexibility to these data centers, you can unlock gigawatts and gigawatts of energy. So that's one is I think that we're starting to to look at data centers as more flexible.

How does that how does that work in practice?

In practice, that works. So instead of having my backup generators that I never wanna use, I'm now starting to from a data center's perspective, I'm starting to be saying, okay. If I have to use those backup generators, I mean, we're talking, like, an hour a year, five hours a year to remove my load from the grid to, you know, just free up a scarcity event. I'm willing to do that.

I mean, for Bitcoin miners, they're already gonna be offline in scarcity events because their prior prices are gonna be high. But your traditional data centers are now being willing to use their backup gen some. So now that opens up a whole new world. However, one of the issues there is we now have to get the utilities to study these loads as a set of firm loads as flexible loads.

So that's kind of the next evolution to actually create that benefit that we can see on the grid. And I think the other thing is when you take ERCOT right now, ERCOT has four hundred gigawatts of storage and generation in the interconnection queue. So about half of that is solar, and about half of that is batteries.

A lot of as we know as developers, right, a small percentage of that will actually be built. But as we're building all of these this new load, more and more of those projects are gonna pencil. So as we talk to the IPPs and the, you know, the greenfield generation and storage developers, you're just seeing more projects that are gonna make sense. Or if I'm repowering a wind farm, now maybe it didn't make sense.

But now that, you know, five gigawatts, ten gigawatts of data centers have come online, things start to make sense. So that five hundred the four hundred gigawatts in ERCOT's interconnection queue, more of it's gonna be built. So I don't think that in it's not like we have to start from square one where it takes five years to build an asset. I have so many assets that are in that late stage queue that I can now build that I wasn't gonna build.

But I do and I avoid politics in general as a Brit in America. But so I agree with that thesis. However, everything's changed with the new administration. Right?

Because it seems like the working assumption was being has been let's just build more gas, but there's a five, ten year lead time on gas turbines right now. Everyone around the world wants gas turbines. So that's kind of out the question. And then we have a range of policy changes from tariffs to other stuff, which is pretty combative towards getting battery and solar built.

Yeah.

So, I mean, that could re the the it's like whack a mole, isn't it? The hitting that mole hitting the battery and solar build out in particularly in Texas, mole could end up really constraining AI dominance for the US, which really should be the a should is a strong word. But as as a as an outsider who lives here, it's just so obvious that America could dominate AI for the next century if they have enough power.

Yeah. We can definitely go down the rabbit hole of kind of what they're doing right now in China and, like, how how that's, I think, gonna drive the next few years. But I will say, while you look at the US and, you know, we're definitely making some decisions right now, they're just gonna make green hydrogen highly unlikely.

I think at the end of the day, what it's gonna do to your energy is just gonna make energy more expensive. So it's I don't think that your your Googles of the world, your Amazons of the world, if energy prices go up by ten percent, they're not gonna not build an AI data center in Texas. So, yes, I think we're gonna see energy or we are gonna see the cost of these projects so that the wind solar projects rise as we have the phase out maybe quicker than we were expecting of the ITC and PTC, but I those projects are gonna be built. They're the data centers are gonna be built.

Those gen assets are gonna be built. PPA prices are going up. I think that's the reality of what we're gonna see. So I don't think this is gonna be a they're not gonna be built.

Which is gonna which is, again, you know, anti industry. It's gonna be a political hot potato. You know, the the beauty of Texas is it's cheap cheap power, cheap cheap energy. Right?

It is.

And, if if all this data center pipeline, even some of this data side center pipeline comes on, surely, base load power is gonna become a lot more expensive unless someone can find a magic tree of gas turbines somewhere, which Yeah.

Would be would be good.

Which it'll be interesting to see how is consumers seeing a higher energy cost, how that is dealt with. But we are already seeing in ERCOT changes to how we're gonna distribute transmission costs, and I think we're gonna see more of that burden on the on the industrials from the residential.

And we'll see kind of as residential energy prices rise be as you have the PTC, ITC fall off, we're also probably gonna see some political turmoil there. So I do think what it'll be interesting to see how it all plays out, but I do, when solar batteries at our price point right now, they're gonna be built regardless. It's just how expensive are the energy cost and then how what does that do politically for people?

Yeah. I mean, if I was in the White House, and I do I'm, of course, not allowed there, because I'm not an American citizen.

But I think my on a high level, I would accept that the US has lost the race for battery cells and PV Yes.

And say the competition that we can win is in AI dominance.

Yes.

And you only have to look at DeepSeek and how how fast China's moving on that. America can win that, but they can't win that at the same time as trying to win the the war on PV, which has already been lost, and other technologies.

Yeah. Focus is key, and I think focusing on where the puck is going makes all the sense in the world. I mean, China last year put on four hundred gigawatts of new generation capacity.

Right? To put that in perspective, we're at about twelve hundred gigawatts total in the US. So about a third of the US generation capacity, China built in one year. So it is a very much a it is incredible. So anything that we can do to incentivize more generation here and more of these projects here is is gonna go a long way. Otherwise, you are gonna see this go to, you know, places like China.

Interesting. So rather than talking like old men in the pub complaining about things, what about actual solutions? So you you're very active in our COPS, so let's focus on that, but we can talk about any region. What regulatory changes do you want that are gonna open the door for not just battery and generation, but also load at the same time?

Yeah. So I think right now, we, we're still working with general protocols. So that's your you know, what what ISO you're in, kind of their operating rules. Right now, the operating rules are still stuck in the past where it took, you know, ten years to build the coal and natural gas plant.

It, you know, it took ten years to build a new factory. It's just the new reality is data centers take a year. You know, when solar batteries take a year, we need protocols that allow us to take advantage of that. So one I already touched on was the how do you study loads more similar to how we study generation.

So with generation, if I'm a developer and I'm going to build an asset, I accept that there could be some amount of curtailment. So I do my studies, my forecasting, my projections, trying to select locations where I can finance it, and I understand those risks. So I think, obviously, you wanna have certain loads that are critical. You don't want your data center.

You know? I mean, you don't want your hospital on a load schedule that can allow them to be shut off. But maybe if I'm one of these AI data centers and I have that backup, I can be studied in a different way that's more similar to, hey. If there's a grid issue or a grid grid constraint, the same way that we can shut put ramp down generation, well, we can ramp down these loads.

So I think taking advantage of that in a way makes a ton of sense. Also, just from a metering standpoint, right now in ERCOT, it is a nightmare to have a battery, a solar asset, and a load all behind the same meter.

You end up losing the storage asset. We'll lose it exempt wholesale generate or generator status or excuse me, it's exempt wholesale storage load. So it costs you more to charge your battery. That's why it's basically.

Yeah. Exactly. And then it's just it's a nightmare. And that is not there's no technology needs to be invented there.

We just, you know, we just need to change the metering protocols. So I think there's a lot of ways that industry can collaborate right now to create solutions that aren't, you know, I don't need to go do hard sciences. I just need to go convince ERCOT to do things a little bit different, and it'll have a huge impact on reliability, on cost, on how we're utilizing infrastructure in a smarter way. So, yeah, there's a lot that can be done there.

Alright. Now onto the last two questions. So firstly, anything that you want to plug for our listeners?

Yeah. So I think two things. One, just because it's a generally more storage audience. I think that we've kinda come up with a few creative structures to do our deals also with batteries. So if anyone's out there, please, you know, feel free to reach out, brock at satoshi energy dot com. And I think that we can do some pretty interesting things with batteries, in these large data centers and and really create win win scenarios in the same way that we have with solar and wind assets.

And then the other thing is that, you know, I kind of touched a little bit. You know, most of what we've done today has been that kind of greenfield load developer, but also that software solution. So, also, you know, I'd love to have be able to roll that out more to in more different use cases where we really understand, you know, energy management and your energy bills in a very direct way and, like, kind of bringing that transparency to to both the energy bill, but then also the settlement. So those are kind of two things that, yeah, we'd love to hear from the audience.

Alright. So if you want to hear from Brock, you can find him on various social media channels, but you can also check out Satoshi Energy, and we're gonna put a link in the show notes. Go and do that. And now the well, my favorite question. And you must if you're a Bitcoiner, you must have loads of these. By being a Bitcoiner, you are a contrarian in some way. But what's your contrarian view?

Yeah. The contrarianism's definitely there's no void of that in the Bitcoin community. I think one that's will be interesting for your audience is I think that there's the potential and likely outcome that loads are gonna provide more demand response than batteries are in the coming years. I think that with the where protocols are going with where loads are becoming more creative with how they operate, that, I mean, we're seeing massive amounts of battery adoption on the grid.

You know, Texas is pretty close to surpassing California in terms of total adoption, but the amount of gigawatts that we're putting on right now of flexible data center load, and that load can easily ramp down as we kind of talked about earlier. It can ramp down very quickly. And if you add some amount of, you know, UPS to it, it can act, you know, very similar to a battery. Obviously, all we can do is ramp down.

We can't also push energy back to the grid. So there's gonna be places where batteries I mean, you're you're still gonna see proliferation there, but I think from a demand response in participating in those programs, as long as everyone's given a fair shake, you know, that I think you're gonna see more of that market go to loads than you will of batteries. Interesting.

This is the second time we've had someone on the podcast in probably three weeks that's come on and said, DSR, a demand side response of some sort, is gonna eat batteries because the the, the cost profile looks better. Yeah. We shall see.

Wanna say a massive thank you for joining us. Brock, I really enjoyed that that conversation. Thanks for joining us and making the trip to New York. We really appreciate it, and I'm sure our listeners will really enjoy that one. Again, if you wanna find out more from Brock and all the work that Satoshia Energy are doing with collocating data centers and renewables, you can do so on their website. And, no doubt we'll have you on the on the podcast again to hear about some of your gigawatt data centers when they come in.

The next one. Thank you so much for having me.

Thank you, man. Ma'am.