On The Brink with Castle Island - Brannin McBee (Core Weave) on Greening the Grid, the Texas Meltdown, and Bitcoin Mining (EP.183)

Episode Date: February 22, 2021

Core Weave chief strategy officer and former energy trader Brannin McBee joins the show to talk about the grid meltdown in Texas and the impact of renewables on the grid. In this episode:  Brannin's history in energy trading and energy analytics How his firm survived a 95% collapse in the price of natural gas – and how that presaged his work in the crypto space How Brannin's firm Core Weave alternates between mining ETH and monetizing GPUs through high-performance compute How wholesale power trading works – and how it provides valuable informational signals to the grid How the US grid is arranged – and how the energy mix has changed in the last decade How coal has been marginalized in the last decade through natural gas and renewables Why energy is not globally fungible Why forecasting wind is so difficult – and how that complicates including wind into the grid How renewables trade off reliability against carbon impact How increasing the share of wind and solar destabilizing the grid Why increasing units of wind requires new units of gas on the grid The confluence of causes behind the Texas grid failure last week How Core Weave selected their location on the grid – and how they took political risk into account How much energy Bitcoin consumes – and where that stands relative to the energy consumption of the US How flaring natural gas to mine Bitcoin works – and whether it's economical Does Bitcoin work as a load balancer for grids? Can you compare Bitcoin's footprint to that of entire countries?  

Hello and welcome back to On the Brink with Castle Island. This is a very special timely episode with my friend Brandon McBee, who is the chief strategy officer at Corweave, America's largest Ethereum miner. He is a long history trading energy on the wholesale markets. He's the perfect person to weigh in on this week's topic, which is the nature of the U.S. grid, especially in Texas, what the impact is of adding renewables to that grid. and of course how to think about the energy cost from mining Bitcoin and Ethereum. So as you've all probably heard by this point, the Texas grid had a total meltdown last week.

Brandon DM me a week prior and said, hey, something bad is happening with Texas. The weather conditions are bad. There's going to be instability. And then, lo and behold, it had a total meltdown. Now, the reasons for the collapse are complex, but there's an interesting dynamic, which is how adding renewables to the grid actually causes more instability. because they have lower capacity factors. They're more intermittent, and they're actually harder to predict ahead of time.

The grid requires that you match demand and supply exactly all the time. We also talk about those very popular topics of the notion of flaring natural gas to mine Bitcoin, whether it works and whether it's economical, whether Bitcoin is a load balancer for grids, and of course how to think about Bitcoin's footprint. Fascinating and timely episode, I learned so much about energy production. Let's dive right into it. Brought down by bad mortgage investments, Lehman, which has 25,000 employees, will be liquidated. The federal government loans American International Group, AIG, $85 billion.

This is a different kind of market, and the Fed is asleep. The federal government is stepping it to stabilize Fannie Mae and Freddie Mac, the two mortgage giants that have been threatened by the housing crisis. The Bank of England has pumped 75 billion pounds more to Britain's ailing economy with a new round of concentrated easing. You print a couple trillion dollars, and all of a sudden, people start to worry. So out of this worry, we have something called a Bitcoin. Bitcoin.

So Brandon McBee, this is very exciting for me. You're one of my favorites, just like people generally. You have an amazing expertise to cover today's topic, which is the energy grid and crypto mining and their intersection. We have some pretty timely stuff to talk about. I think your background is just perfect to have this conversation. so I'm pretty glad you jumped on. So anyway, welcome to the show.

Thank you. I'm incredibly excited to be here today. This is a topic that I'm super passionate about, and it's great to be able to chat with you about it. You were just telling me that we landed a rover on Mars or something. Is that right? Yeah, a one-ton rover perseverance just landed. What was it, 12 minutes ago now.

I mean, your ability to line this up with landing a rover on Mars is pretty amazing. Yeah, I'm sorry that. I took you away from like the rover, you know. The rover excitement. I wasn't even aware. There should be some calendar of like notable space related events or something that I can follow. If only there was, you know, a mechanism to exchange information like that somewhere. Yeah.

Well, so we successfully ended the rover, right? Like it didn't crash or anything. Didn't crash. It's sending pictures back. Looks pretty happy for now. I think the last one they landed was nearly five, eight years ago or something. It's still crawling around out there. So it's pretty exciting.

I like the Mars stuff. The rovers are powered by radioactive decay. Is that right? Yes. They wander around with leaking radiation everywhere. But it's okay because it's Mars. Yeah. Yeah, we've got to start working on our next planet, right?

Get it accustomed to the way that we work. There we go. that's our segue into the topic of today's episode, which is the energy grid, as they say. So you've such a cool background. In fact, you used to do something that I wasn't even aware was a real job, which is like trading energy, which I didn't even know was a real thing until I met you guys. So yeah, tell us about your background and your credentials. Yeah, sure.

It's a pretty weird sector, super small community at the end of the day. But I started my career in natural gas analytics. I worked at a company called Bintech Energy. Our specialty was piecing together natural gas pipeline information and creating an understanding of what these supply and demand components were on that pipeline to, or on the national pipeline grid, to provide a real-time picture of what was going on. Right. Historically, you know, that data was lagged two or three months and was only available in a monthly granularity.

And the data set that we put together allowed for real-time daily information. So it just totally transformed the way that natural gas analytics worked. And my role of that company was covering and leading the Natogast demand desk. So I would put together reports on Natuagas Power. plant, look at how natural gas was consumed in the residential, commercial, industrial, and electricity sectors on a daily basis. From there, I went on to work at a hedge fund covering natural gas, electricity, and coal. We would put on trades that would express fundamental changes in value between those commodities like pair trades.

After that, I was hired to run a discretionary portfolio for a family office trading natural gas and electricity markets. That was down in Houston. I ran analytics at that shop as well, covering the various U.S. power markets, PJM, Neeple, Nypool, Nypole, Myso, Urquat, California. ISO and BPA. Those are the kind of major deregulated markets across the U.S. And today, I am the chief strategy officer at Corweave. So during your time at the covering natural gas, did it draw down something like 95%?

Do I have that right? The price of natural gas? Yeah. It sort of came down precipitously, right? Yeah. So that was at the emergence of a huge income. increase in production in natural gas. That was directly related to the scaling of fracking technology and oil and gas production. So U.S. natural gas production went from something like 50 billion cubic feet per day up to, I think its peak recently was at 93 BCF per day. So it's increased quite a bit over the last decade.

So it's kind of interesting, almost like a prelude to working on Ethereum being an Ethereum minor in seeing the 95% drawdown in Ethereum, which you guys went through a core weave, but stayed solvent. So congrats, you're one of the few, I think, but managed you do that. Yeah, yeah, we owe a lot of that to our CEO's ability

to keep the business together. We were running on razor-thin margins for a while, but Mike did an amazing job getting us through that period. So we're going to get into it a little bit, but just briefly tell us about CoreWeave and Highland Intersex with your history and your competencies. Sure. So CoreWeave is a specialized cloud that's focused on high-performance compute. We service the VFX, AIML, and blockchain sectors with highly scalable and economic infrastructure. When that infrastructure isn't allocated to the HPC market or our cloud business, we attempt to monetize it via cryptocurrency mining with our GPU and CPU compute. So effectively it just automatically allocates itself to whatever the most profitable cryptocurrency is.

And then we treat it as an arbitrage opportunity. We are liquidating our proceeds on a daily basis. We're not taking speculative positions. We just look at it as another method for monetizing our infrastructure. It's pretty fascinating to have these always-on buyers of energy. And obviously we're going to talk about that. But one minute you're selling compute to humans, the next minute you're selling it to the blockchain.

It just blows my mind that that's a reality today. It's, you know, the GPU infrastructure provides us pretty amazing optionality. to allocate the compute across various high growth sectors. You know, when we started the business in 2017, that was really our thesis that, you know, we'd be able to kind of bootstrap ourselves by building into the cryptocurrency space first. And then once that business was online and stable, we would be able to start working on building out products

into the HPC cloud sector. And that's really what we've been focused on for the past year and a half. So returning briefly to your history with wholesale energy trading, so tell us a little bit about what that is. And I know a lot of people think, you know, trading financial assets or commodities is sort of not useful for society. But in our previous conversation, you were telling me about how that is actually pretty handy. So tell us about like the positive externalities of. trading energy. Yeah, a lot of people when they think of energy trading, you know, they immediately just go to Enron

and the debacle that occurred there. I like to believe that speculative wholesale power trading provides insights to the grids where trading is occurring when there's pricing discrepancies. A great example is what happened in Texas this week. The market was able to signal to Texas and other participants on the grid that there was an issue, right? That there was a supply demand mismatch. And that's why prices spiked. I mean, before the event power for the week that just liquidated was trading about $30 a megawatt hour.

I think it's going to clear, you know, above $6,000 a megawatt hour. So we had, you know, two days in there that cleared, I think maybe three days that cleared $9,000 a megawatt hour. But the grid started receiving signals from the trading community that there was an issue, you know, up to 10 days beforehand because speculative traders kind of intensely. focus on these discrete data sets every single day trying to figure out what is not priced efficiently or correctly. My specialty was looking at short-term weather forecasts within the 14-day window and identifying trend changes within those models and discrepancies between those models and the distribution of those models and where markets were priced. That was across the U.S.

So I covered all power markets, all natural gas markets. It certainly leads to some pretty interesting scenarios. And then you start lining that up with the fundamental side of the equation, right? So you understand where the natural gas, coal, wind, solar, stack sits within the market. And that allows you to ultimately identify efficient pricing and where it should be. And in this case, in Texas, there was no efficient pricing because the grid had. to enter rolling blackouts and it hit its cap price of $9,000. Right.

It's pretty hyacchian. It's sort of very nice and pure in a certain way to have traders, you know, looking at these sort of high frequency data sets, whether it's weather, whether it's changed in supply or extrapolating out with the various energy sources, how they're going to be performing a few days ahead of time and sending signals to the market, which are then picked up presumably by the grid operators themselves right so they take in those price signals as information and make decisions based on that is that correct that's totally correct the only thing i'd add is it also sends price signals to consumers on the grid as well right so

industrial facilities can see that there's this price increase coming and they can decide like all right We need to, you know, hedge our exposure or they can, you know, start to prepare to shut down. So it's not really a last minute notice for them. So they would potentially even have, you know, a few days notice that something was coming. So there you have it, traders. At least of this specific commodity, they're good for society. I can speak to people that trade interest rates or whatever. Or equity traders.

Exactly. No, I've always been attracted to. commodity trading because at the end of the day, there is a physical price that theoretically solves the market, right? That's where economics really works. Supply demands allows you to identify a correct price. So if you love getting into data, which has been my background as an analyst and a researcher over the past decade, it's a great spot to be a participant in the market as a trader. So I don't know if we're going to be able to cover everything I want to cover today.

We might need three hours to do it, which I don't have, unfortunately. But that said, we're going to do our best. I wanted to give an introduction to just energy dynamics in this country and then talk about the Texas situation, which is freaking everyone out right now. And it's also causing a lot of misery. and also talk about core weeds approach and also talk about Bitcoin's energy consumption, Theorem's energy consumption, whether it's boiling the oceans.

I don't even know if we can possibly get to all that. That's like four podcasts and one, but we'll give it a shot. So maybe just tell us, lay the background for how the grid is actually divided up in the U.S. because it's not one grid. And then the changes you've noticed sort of in the energy mix recently.

Yeah, absolutely. So the U.S. energy grid on a macro perspective is broken up into three interconnects. East, West, and then the other country in the U.S., Texas. Within that, there are 66 balancing authorities that are in charge of regulating their own. grids. So like Urquat is the balancing authority within Texas, so there's 65 other balancing authorities across the U.S., and they each have their own rules and ways that their consumers and

producers of energy interact with the grid. In general, across the U.S. are generation from 2019, and most of the data I'm going to quote in here is from the EIA. energy information administration. And this is using 2019 data set just because the 2020 final data set isn't available, but it hasn't really changed too much since then. So in general, our U.S. production is about 22% coal fire generation, 46% from natural gas, 9% from nuclear, 10% from hydro generation.

And then the balance 10% is from, and solar. So you were telling me that the energy mix has actually changed dramatically in the last decade or so. Is that correct? Yeah. You know, coal has really dropped off a cliff. Coal used to represent over 50% of our generation output in the U.S., and now that's down to just about 22%. It's pretty crazy. And that change has really been driven by an increase in natural gas fire generation, which, going back to your point earlier, natural gas prices have come off substantially in the past decade, which is allowed for a lot more natural gas fired generation. And then renewables have made quite an impact over the past decade as well. We've gone from about a 2% market share of generation for wind back in 2010 to a little over 7% in 2019. That's quite interesting on the natural gas front.

As you said, that's because fracking came more economical and just more precise. Yeah, I mean, the natural gas pricing came down from, you know, $7, $8 in MNBTU down to about $2, $2.50, perimen VTU. And that came right down into the coal stack, the pricing for coal fire generation. And then in the past few years has gone even below that. So it's really offset coal fire generation. And this has been hugely beneficial for emissions in the United States. Believe it or not, CO2 emissions from electricity generation are down 28% over the past decade. Absolutely, some of that is due to renewable increase. But the vast majority of that is due to natural gas fire generation offsetting coal fire generation. Because The CO2 rate per megawatt hour on coal fire generation is about 1.1 tons. And the CO2 rate for natural gas fire generation is about 0.6 tons per megawatt hour. So it's a much cleaner fuel source for us.

And at the end of the day, I'm a huge supporter of natural gas fire generation being a strong part of our grid as we continue to bring more renewables online. So that's really one of the key points, I guess, that people should remember is that there's not this homogenous relationship between energy production and carbon output or carbon externalities. It really is dynamic and it changes over time and it depends on the composition of the energy generation. 100% correct.

And again, at the end of the day, coal-fired generation is our dirty, fuel source for electricity generation. And nato gas is just substantially cleaner, right? There's two other emission components that the EPA tracks pretty closely with electricity generation. They track CO2, SO2, and NOX emissions. Notchogas doesn't have any SO2 emissions. Coal has a pretty decent amount.

And then natural gas is about one-fifth the amount of NOx emissions per megawatt hour equivalent of coal-fire generation. So net-net, it's just a substantially cleaner fuel source. It's way easier for us to transport around the U.S. And frankly, we have a lot of it. We're pretty blessed to be in a country that has such prolific natural resources. And SO2 is sulfur dioxide. and what is and what is knox? Nitrogen oxide.

Okay and so those are also greenhouse gases, is that correct? Yes, the gases that the EPA tracks pretty closely with electricity generation. There's various programs that limit the amount of SO2 and NOx generation on a state and regional level because of their adverse impacts. So the other thing to understand about grids is the point I try to make. I think Bitcoin has tried to make a lot is that there is a certain geographical nature to them. So a unit of electricity production is not sort of globally fungible. Like you can only transport electricity a certain distance before you incur loss. And also there's just certain regions that are more, they have a greater affinity for a certain

sources of production than others. So talk us through that a little bit, the loss factor and also the regional dynamics with energy production. Absolutely. So on the loss side, it's pretty interesting. If you're going, I think it's just about over 100 miles or so, you incur about a 2% loss on the power of your electricity through a transmission line. Right. So if you're going thousands of miles over a line, you know, it's not a linear increase necessarily, but you can incur pretty substantial losses. And that's what's kind of limited our ability for generation to be centralized. That's why it's so distributed across the U.S. That kind of leads to an issue with renewable generation.

NREL, the National Renewable Energy Lab, puts out these, has put out some amazing. amazing studies over the past, you know, 10 years or so that looks at the energy potential of solar and wind across the U.S. on a super granular basis, right? So on wind-powered generation, aside from coastal areas, right, because wind along the coasts, like offshore wind generation is pretty strong, especially on the east and west coasts, not as much in the Gulf. But otherwise, within the U.S., on land, you can really only build wind generation in the center of the U.S. So think Panhandle of Texas and pretty much going straight up. If you were to overlay that map with a population density of the U.S., it kind of looks pretty much opposite, right?

Like the population density of the U.S. is around the coasts, right? where offshore wind generation can work well in the northeast and the west, but otherwise, from a population density perspective, you don't have that many people living in the center of the U.S. yet. Now, that obviously may be changing with the emergence of remote work, and I'm certainly feeling quite a lot of people moving out here to Bozeman, Montana, where I'm at, escaping the cities. Maybe we'll convince you to get out here soon.

Yeah, I have to because that book, I tell you about the Senate of the Art of Motorcycle Maintenance, talks about Bozeman a lot. Does it really? That's pretty fun. I was really taken with that book when I was younger, so I vowed that I would visit Bozeman. So I will come out. Yeah, so yeah, I'm looking at this map of solar and it goes north from Texas. And it's a wind, yeah, wind goes north from Texas. and it's a great resource in that region.

And there's been a lot of deployment as well, right? Like in Texas, there's 28 gigawatts of wind capacity in the region. And there continues to be a very strong deployment, you know, through Colorado, Wyoming, Montana. It's just a resource that works really well there. But it's not a resource you would go deploy in the southeast or the southwest or around Michigan or something. It just isn't, it doesn't make sense because the consistency of wind in those regions isn't high enough to justify the deployment of those resources. The other map I was

going to get into real quick is the solar map, right? So solar is best placed pretty much across the southern half of the United States with a wading in the western half. So you can get away, you can get away with solar installations in Florida, but you'd rather be putting them in California, Arizona, New Mexico, and the western side of Texas. So again, if you kind of put these two maps together, right, solar and wind, you end up with the eastern half of the country, or the eastern interconnect,

not being a great area geographically for renewables. And so combine that with the fact that you can't transport renewables over a long distance to deliver into those markets. And you really do have to have your traditional generation sources of hydro, natural gas, coal, and nuclear generation. Yeah, and this is a really well-referenced episode. I think I'm going to try and put a lot of these resources in the notes on the blog page, accompanying the episode, because I want people to look at these maps because they're just so fascinating.

It's so cool the amount of data we have on this stuff. I mean, again, NREL's done an amazing job putting together these resources and doing whatever they can to help guide the deployment of renewables across the U.S. So what you're saying is there really is this geographical contingency to renewables. They only make sense for certain areas. As it happens, Texas does have a lot of wind resources and solar resources. So Texas is actually one of the exceptions here, I guess, where there is a lot of renewable potential. and in fact as we're going to explore like renewables have been built in Texas fair amount but you were telling me there's kind of more about and you have that population density right and so

there's a lot of people in Texas but there's also the other dynamic the other variable of capacity factors that we have to talk about which is the intermittent nature of the renewables themselves yes so Renewables aren't the easiest thing to work with in the world. So to take a step back, right? Electricity grids work by targeting a specific frequency. That frequency is the marriage of supply and demand of electricity on a grid. So Texas works on 60 hertz. The goal of all these grids, right, is to maintain a balance such that electricity is delivered consistently at 60 hertz.

So if demand goes up somewhere, there needs to be a simultaneous and near real time increase of the exact same demand amount. And if demand decreases, you have to have a simultaneous and near real time decrease in generating. So this concept has been traditionally fairly straightforward to deal with, right? Load is forecasted, you know, down to the five minute or one minute increment in a lot of these regions. There's a very tight relationship between weather and load. Load is electricity demand, you know, to the point to where you can forecast up to like 99%-ish accuracy, right? And then the goal of the grid throughout the day was just to match generation sources with demand. And it was this fairly smooth curve that would hit a trough at night because less people are using electricity would slowly rise throughout the day, hit a peak around 5, 6 p.m., depending where you are in the U.S., and then come down in the afternoon.

Right? And it was a very smooth curve throughout the day. And you were able to meet that with your traditional generation sources. So you had coal and nuclear were baseload generation, right? They would just sit at a nice, consistent output throughout the day. They wouldn't really change their generation much. And then any variations of that curve throughout the day was met by natural gas fire generation. And it's structured that way because natural gas fire generation is more flexible to scale up and down throughout the day than coal or nuclear.

If you do it with coal, nuclear, it's less efficient to operate the plant. I was going to say that that was a subject of some studies I did about five, six years ago looking at the impact of running natures gas and coal-fired generation. intermittently and the efficiency losses that were associated with that. Yeah, we weren't kidding when we said that your subject matter expert on this. I guess there's a couple things in terms of reliability. Well, there's predictability, so it's the ability to predict the presence of that energy resource ahead of time, which is obviously easy enough with coal, natural gas, and nuclear obviously harder with some.

solar and wind where you're just depending on the environment. Then there's reliability and there's capacity factor. And then there's also whether you can spin it up or spin it down in a hurry or not. And you're telling me that coal and nuclear kind of need to run on a pretty steady basis. It's kind of hard to spin them up and down. The gas is the one that can back stops. They can backfill any shortage, right? Yes. Tell me about solar wind, because they're like kind of unpredictable. You're sort of exposed to the capriciousness of mother nature. And they're sort of unpredictable in a couple ways, right? So wind and solar have, you know, completely changed the game over the past few years. And, you know, it really wasn't that much of an impact, you know, in the early 2010s

because there wasn't that much market penetration of wind and solar, right? You know, you're still talking around 3% or so of generation in the U.S. was from wind and solar. And as that market penetration has increased, you know, there are times in Texas where 50% of the demand was being met by wind generation, it's caused a lot more volatility on the grid. So to step back into my, put my meteorological hat back on. forecasting wind is insanely difficult. 15 to 30 minutes out is about as good as you're going to get to have a relatively accurate wind forecast. And planning a grid around that, it just requires a as the market penetration of a volatile resource like that,

grows, you have to begin to load shape with the rest of your generation resources. Right. So again, going back to that really nice, smooth load curve that you have throughout the day that you would have, you know, back in 2010 or something along those lines, right? You were able to plan your resources to meet that demand profile very efficiently. Well, now you've introduced this very volatile. generation source. You know, while it's, there's no emissions to it, you might say the cost is low,

whatever you'd like, it still increases disruption on the grid because that smooth curve, all of a sudden just became a pretty volatile curve. And at the end of the day, you still have to always match supply and demand to achieve that frequency rating we were talking about. So if you have wind generation, that suddenly, increases by 10 gigawatts on the grid, you need to decrease generation from another resource by 10 gigawatts. Or if you can't do that, you just have to dump that wind generation output into the ground. It's literally called grounding the generation output because it would disrupt

the grid and cause a brownout or a blackout. Otherwise. So you tell me that wind actually weirdly enough, kind of requires these commensurate sort of matching gas resources in order for the grid to sort of accommodate them properly. Yeah. So the technical terms for it is that these are reliability and ancillary services that are used to balance wind and solar generations on the grid. You know, one way to think about it is since the output of wind can change so quickly and so dramatically, we have to use a type of natural gas generation unit called a combustion turbine to be able to load follow with wind and solar. And a combustion turbine is literally a jet engine that just uses natural gas. So whenever people talk about, you know, achieving a grid of,

let's say, like, 100 percent renewables, and let's just say it was all wind or something along those lines. What they don't really discuss is that you would have to build out pretty much the exact same capacity of that wind resource with natural gas fire generation as well in order to capture all those peaks and valleys and provide a consistent generation output. But it's literally a jet engine that you'd have to build to follow everything. And solar is maybe slightly more predictable golden wind, but not perfectly so. Obviously, it doesn't work at night, but then also clouds. Clouds, yeah. A buddy told me recently who is pretty active trading in the Urquhart market. He said that the solar generation has gotten to such that when a cloud goes over some of the

major sites in the region, it looks like a power plant trips offline because you can lose a gig and a half of solar. And it looks like a nuclear power. plant just went offline and there's a serious issue with the grid. So the clouds can be, you know, pretty disruptive. And again, you know, it's a little bit easier to forecast where your clouds are going to be than wind generation, but it's not not something that you can do days out over time with any kind of reliability. Right. But I guess on the bright side, we have various ways to store energy. So we have batteries, whether they're physical batteries or chemical batteries. So tell us about the prospects for batteries a little bit. The first kind of stored energy on the electrical grid was

something called pumped storage, right? This is like kind of the first battery that's been used at scale and has been, you know, wonderful in areas that have that geographic height difference. So the way this worked was you had two lakes that were about, you know, 1,500 feet apart in vertical elevation. And in times of low electricity prices, you would pump water to the top lake to store that potential energy for when electricity prices were higher. And then you would release that water. And as that water ran down the tubes, it would, you know, use conventional methods for electricity generation through hydro, right, like spinning a turbine, basically. And there you would

have your electricity output. That was the kind of first battery, but that doesn't really work very well in Texas, right? Because there's not a lot of places where there's a thousand foot height differential. So what started to come to market recently is commercial battery storage. I think the first deployment in ERCOT is coming this year or has started to have been put it into place and it's about one gigawatt to put into perspective what one gigawatt means. The peak demand in Texas during the summer is about 75 gigawatts. So there's a lot of promise around commercial batteries, but there's issues with them right now as well, right? It's very expensive to put this stuff in place, right? It uses rare earth minerals.

And really similar to your phone, there's efficiency loss if you're using it in a very bursty basis or you're cycling, right? Like if you're using your phone quickly and then charging it and going back and forth, back and forth, it over time reduces your ability to recharge that battery. So there's efficiency loss. And then ultimately, there's been issues as well with severe heat and severe cold resulting in loss of capacity. So there's a lot to work through on the research side. I'm certainly not just, well, let's go build a lot of batteries, put them in the ground, and that's going to solve the grid. It's something I hope innovative companies like Tesla are able to make a positive impact in the future, but the technology isn't there today. So TLDR, wind and solar are just eminently less predictable.

They're also more volatile. and so increasing their share of the grid increases just the volatility of the grid, and at present batteries can't fully mitigate that. Yep. This is a great summary. All right. So this brings us to the great state of Texas. Okay.

So it's funny because when I hear you and the other guys at Corey have talked about Texas, there's like a certain pride or maybe Texans say this. This is pride that they're their own grid. I think they like that for resiliency's sake. But in this case, it sort of worked against them. So there's obviously total chaos. He actually told me about this last week. So I don't know how to trade on like Texas power is going to go through the roof.

But for what it's worth, you know, Brandon, you did tell me that something terrible was going to happen to Texas, like with a full weeks worth of notice. And then, lo and behold, it did happen. So break down the Texas situation. happened. Just utter chaos. I mean, everything that could go wrong went wrong and happened at a time when people really needed electricity to heat their homes. Texas doesn't really have that much cold in it, and people in general just don't really prepare for it either. I mean, I've spent, I grew up in Dallas. I spent the last five years or so living there.

And I mean, if you got a quarter of an inch of ice or snow on the ground, I mean, the cities were just grind to a halt. But on the electricity infrastructure side, just everything happened, right? So natural gas was freezing in the pipelines. And not specifically the methane, but the liquids that were within those transportation systems were freezing. and thus shutting down deliverability of some natural gas pipelines to power plants, right? And when you remove power plants from the grid, that goes back to our discussion around, you have to match a generation with demand. And if you don't have enough generation to meet demand, you start hitting blackouts because the grid fails.

There were power plants that were not prepared for cold. There was a nuclear facility that had two reactors at it, and one of them had to shut down. It was a 1,200 megawatt unit. One of them had to shut down because the water intakes weren't fully prepped for cold weather and they froze. So that's a very large amount of generation, 1.2 gigawatts that was removed from the grid because the plant wasn't prepared. And then the other compounding factor on all this is that there's 28 gigawatts of wind generation in ERCOT. and pretty much all that capacity was offline due to the wind turbines being frozen and inoperable. So pretty much everything that could go wrong went wrong and there was just nothing that the grid was able to do to keep up with that scenario.

It's interesting that the weather became the latest front in this culture war, but it did somehow. And you immediately had environmentalists jumping to the defensive wind, and you had some folks that are against renewable saying wind was solely responsible for the chaos. I guess, you know, how responsible was wind? And, you know, I think you were telling me before wind has become a larger and larger feature of Urquat. So where would you peg its responsibility there? Sure. So if you look at this discrete event, right, you can't blame wind. You can't blame natural gas. Everything that could go wrong went wrong. But if you look at it from a more macro perspective, right, on what drove the capacity mixed to get to where it is today in Texas, I would probably argue that that was due to the integration of renewables on the grid. There's been.

a substantial amount of coal-fired generation retirement in Texas over the past decade or so. There hasn't been much new build of natural gas fire generation. Pretty much all the new capacity that has come online to replace the loss of coal-fired generation was with renewables. And from a policy perspective, like, that's why we were in a place where we were, you know, net short 10 gigawatts or 15 gigawatts of generation versus where demand was at. So it's tough. I wouldn't point a finger necessarily at one cause, but absolutely we have decreased our amount of thermal generation capacity in Texas,

increased renewable generation capacity. And that did not help in this particular event. Yeah. So, okay, that's, I would say, pretty nuanced and balanced view, often considered. What is this, how does this make you feel about the prospects for more renewable integration in the future? I mean, so this is something that California experiences all the time, right? Like, it's, it's gotten to be just a normal part of their system. You know, lives aren't in stake necessarily. but California hits rolling blackouts all the time because of how poor their electrical infrastructure has been operated and maintained, but also because they've brought online a substantial amount of renewables onto their system over the past decade while not continuing to maintain and grow

their thermal fleet. So I think that this is, I mean, this isn't limited to the U.S. either by any means. Germany recently decided to take all their nuclear generation offline, which I think nuclear generation is one of the best forms of a baseload, electrical generation for a grid. So taking that offline creates a substantial amount of issues. And Germany's had, you know, blackouts and incredibly high prices. They've had a lot of issues removing that source and replacing it with renewables. Right. And if if there were a method to efficiently smooth that curve of generation and predictability, such as commercial battery storage or pumped hydro or something along those lines, it would be different, but that's just not something that you can deploy everywhere.

So I think if we stay on this path of aggressively integrating renewable generation into these grids without thinking through these scenarios of high demand events and what happens if your wind turbines freeze and your natural gas freezes and pipelines and everything like that, what will have taken a 30-year event today to cause this scenario is going to take just a five-year. year weather event in the future, meaning it's going to happen more frequently. And that's irrespective of climate change, right? Like, that's just, the grids are going to be more susceptible to these failures due to increase volatility and decreased dependence on availability of generation. So it's almost like you're trading some of the carbon footprint of the grid for fragility. we're sort of buying it, but at the expense of becoming more fragile. Yeah. Yeah.

I think that's a great way of putting it. And again, interestingly, over the past decade, we've dropped emissions substantially, right? We're going back to a 28% reduction in CO2 from electrical generation. It's pretty wild. SO2 over that period has dropped 76%. And Knox has dropped 46%. I think that there can be this really, there can be a nice balance of natural gas

and wind generation in the future to continue driving down emissions over time. So turning to CoreWeep, that was a great introduction to the nuclear. That's sort of like a full introduction to how our energy grid works in the U.S. It's just a great resource. Now with CoreW,

weave, you guys knew you were going to consume a lot of energy and you sort of had the whole menu of options available. And you guys also deeply understood the various grids and what they had to offer. What was your reasoning? How did you, you guys are in PJM, if I'm not mistaken, how did you go through that process of selecting between the different grids and the different energy sources, like you had the whole US to think about. So how did that process work? Sure. So our thesis, again, was always to be ultimately serving this HPC cloud market. So we needed to first and foremost have fantastic connectivity into internet infrastructure. But the questions after that were really focused around where in the U.S. are we going to get

competitive energy pricing while being able to bring employees on site and recruit good talent and to have confidence in the operation of the grid and we chose PJM because it really checked off a lot of those boxes. PJM is a very oversupplied grid from a capacity perspective. PJM has access to cheap natural gas from the Marcellus shale. I mean, this is the region that's like kind of northeast. It's a little bit east of Chicago and then all the way over to New Jersey. So Pennsylvania, New Jersey, a few other states over there, PGM covers. And it's a deregulated grid. So it allowed us opportunities to work in the wholesale power markets and get access to attractive electricity pricing that wasn't as cheap as Texas.

but we believed that there was more dependence on that electricity being economic for us going forward in the future. Yeah, and I guess the Texas case study might be vindicating you right now because there's the grid administrators might be re-evaluating down in Ircott who is consuming power, right? I think there's absolutely going to be a reevaluation of pricing and incentives in Urquot. I think there's two components that could result in a pretty substantial increase in your kind of baseline power price in Texas. One, natural gas pricing is likely going to increase over the next few years. That's going to be driven by a decrease in production after. new policies going to place through this administration for banning oil and gas leases in federal lands. So that's ultimately going to result in less natural gas production. We're also exporting

a substantial amount of natural gas to Mexico and internationally through liquefied natural gas and ultimately just the fundamentals of natural gas supply and demand indicate we're going to move from a $2.50 pricing environment per MMB to probably something towards $3.50. And that's a 30% increase. It's a substantial increase in price. I think the other component is that the ERCOT grid is going to assess this event and likely going to increase the cost or the burden to operate on its grid by hardening its infrastructure to better withstand events like this in the future. So those two cost components together is probably going to drive prices in Urquat from, you know,

the $20 per megawatt hour that it's been sitting at for the past few years, something closer to $35 to $40 per megawatt hour. And that brings it more in line with where the rest of the U.S. is, by the way, you know, excluding California because power prices out there, natural gas prices out there are insane. But the rest of the U.S. probably sits in that $5.50 to $60 megawatt hour. So PJM on a wholesale basis will come in, you know, $30-ish a megawatt hour. But by far and large, Texas had the best electricity prices for the past decade. I think that was largely driven by natural gas being so cheap in the region.

But with these very likely impending changes to the way that the grid is operated and just the price of natural gas, I think we're going to see an increase in electricity costs in Texas. So it seems like you're hinting at the fact that there's basically embedded political risk in where you choose to locate. I think the Pacific Northwest is a great example of that. You had quite a large amount of Bitcoin deployments there in 2018 and 2019 taking advantage of hydro generation where they were getting wholesale prices of, you know, one to two cents a kilowatt hour, sorry, $10 to $20 a megawatt hour. And those utility operators came in and basically said that they can choose who the clients are in those in those grids, right? because they're regulated markets. Not just anyone can participate in. They can make rulings based on how you consume electricity

and how the grid perceives you adding value to the region. And they increased pricing from, you know, $20 a megawatt hour to $100 a megawatt hour and pretty much price them out of the region. That's also something that you saw happen in Canada. I think it was Montreal or Hydro-Cubek, came in and increased electricity pricing for a Bitcoin miners in those regions. So end of the day, we're very happy with operating in PGM. It's a fair price we pay.

It isn't the lowest price in the U.S., but it's consistent, and they've been a great grid to be a part of. So let's get into the energy waste discussion. That was another thing I want to talk about. I know we're smashing two topics altogether, but you've given us a great background to talk about it. So, I mean, there's a lot of angles that we already discussed how the energy mix really matters. So, you know, it's not just about the raw numbers in terms of energy usage, which everybody can sort of trivily compute. I guess maybe what I want to ask you is about the unique feature of Bitcoin. So as we know, Bitcoin is this sort of geography independent buyer of energy in Ethereum 2 for that matter.

So maybe give us your view on the fact that Bitcoin has a lot of hydro in terms of the energy input. And then additionally, the kind of flared gas combustion, which gets a lot of press. Sure. Would you like for me to kind of start with the estimates of where Bitcoin, like what Bitcoin consumes right now on an annual basis and then start diving into. the generation mix. Yeah, let's do that for sure. Okay. So I highly respect the guys over at Bit Uda, specifically Sam Doctor puts out some amazing research on what's going on in various blockchains. And they were kind enough to share with me their perspective on what the Bitcoin mining infrastructure fleet is comprise of right now.

And ultimately, that backs out to how much electricity Bitcoin globally is consuming. So they went through a fairly detailed analysis to break the infrastructure set down into three classes. The S9 class, the S17 class, and the S-19 class infrastructure, right? And then they go through, make assumptions about how much is operating globally. based on how that infrastructure, how the hash rate reacted before and after the happening last year, to get a sense of how much was deployed. And ultimately, they came up with a number that there is, where is this go, eight megawatts equivalent of total electricity consumption from Bitcoin mining globally. Sorry, eight gigawatts, excuse me. So that's 8,000 megawatts.

Okay. And what does that, can you contextualize that a little bit for us? Yeah, absolutely. So in the U.S., right, that would be about 1.8% of generation in the U.S. So there is about 4,000 terawatt hours generated in the U.S. in 2019, right? So if you take that 8 gigawatts, multiply it by 24 and 365, you're going to reach Bitcoin consuming around 70 terawatt hours annually. So again, that comes out to about 1.8% of U.S. generation.

But we all know that not all Bitcoin mining is based in the U.S. There's a lot in China, Russia, Middle East, kind of pretty much where you can plug in a Bitcoin miner right now, it's operating. And don't get me wrong, there's plenty of Bitcoin mining in the U.S. and specifically in Texas. So when you put it in the context of global generation, you can pull a statistic from the International Energy Agency, where they reported in 2019, they believe, believe that the total generation or total consumption of electricity globally was 27,000 terawatt hours. A terawatt hour is 1,000 gigawatt hours. So that would equate then to Bitcoin consuming 0.26% of global electricity generation. Or in other words, you know, being 0.2% of consumption globally.

And I mean, how does this feel in relation to other industries? I mean, it seems like a lot, but, you know, sort of an industry like any other and got other industries like. Yeah. I wish I could provide context on that. There just really isn't that granular of data sets available. But that being said, I think it's fairly safe to assume that the industry's, that Bitcoin and Ethereum are disrupting through scalability economics, et cetera, consume materially

more power than what Bitcoin and Ethereum are consuming right now. Right. And yeah, it's always a function of what it's disrupting, what the counterfactual is. So whether that's the banking sector or even gold extraction. Right. Gold consumes a huge amount of energy, but that's often not part of the discussion. And really hard to identify those concrete numbers of where they're at. But I think that's also a benefit of moving to this digital currency infrastructure,

because now you can really start to pinpoint exactly what the energy consumption is. And that's going to allow us to identify efficiencies in the future that we can implement as well. So in terms of the quote unquote greener components of Bitcoin energy consumption, obviously hydro is a big part of it. We know that a lot of Bitcoin is mined down in Sichuan and Yunnan, especially in the rainy season. So there's not a huge amount of uncertainty there because those are the big sources of energy in those provinces. Talk us a little bit about the flared gas. You'd kind of a different take on it, I think, from the take to a lot of. Bitcoiners have. So this is the idea that, you know, vented gas or flared gas, that's sort of an unwanted

externality a lot of the time of extraction of oil. And you can sort of combust that and put it into a generator and mine Bitcoin with that. What's your view of sort of the positivity or negativity of that potential process from a carbon perspective? Sure. So we really looked at this. in depth during 2019. It was the perfect marriage of our backgrounds, right? Like we're all commodity guys. We're already operating, you know, the largest GPU data center in the U.S. You know, it's just a great fit for us.

We thought it was a fantastic idea. But ultimately, we didn't enter it because we were originally very excited about the potential environmental benefits around it. And, you know, when we started investigating, what those on-site small turbines were actually doing, we found out that there's no carbon emission savings. There's no NOx emission savings versus simply flaring the gas at the wells that was being done previously. There's no ESG benefit. There's no environmental benefit to consuming flared gas in a turbine to mine Bitcoin versus.

is just flaring it to begin with. There is a substantial benefit from a revenue perspective for the owners of that gas that was being flared before. I think on an MMBTU basis, that gas produces around $12 an MMBTU when converted into Bitcoin versus $0 if it's just flared or versus $2 roughly if you were to just sell it into the market. So as a revenue generating vehicle, it's amazingly attractive, right, to go actually monetize this resource that was just being flared previously. But from an environmental perspective, it's not adding any value whatsoever. So we decided to stick to our core business, continue to grow core weave, let the rest of the participants out there go out and chase.

the flare gas or the pipeline to crypto gas concept. There's a lot of great businesses that are looking into that out there. But at the end of the day, it's not a question of emission savings. It's a question of generating revenue and monetizing this resource, which, you know, from the perspective of a natural gas analyst, I would say that we should have more regulation on venting and flaring. You know, around 10% of U.S. natural gas production is flared. It's just wasted on a daily basis.

It's crazy. And that's being done because the regulation isn't there to incentivize producers to not vent and flare. And specifically, that's really in Texas, right? Like, Colorado's done a great job of limiting flaring. Wyoming has very strict restrictions or very strict regulations on amount of flaying per year. But Texas, North Dakota, some of these states with, you know, fairly remote production, let producers get away with flaring gas. And again, my opinion is that if you're going to produce a resource, we should be consuming it in a way that adds. value and if that's through Bitcoin mining, great. But I think instead we should be putting

on a pipeline and shipping it to infrastructure that really needs it. But I guess in the context of the fact that at present it is being flared and wasted, it's not any worse and it's not producing any additional demand on the grid the same way that say plugging in miners to the regular grid would potentially increase demand. So I guess you could say it's at worst neutral from a carbon perspective. 100% accurate. At worst, neutral, you know, brings tax revenue to a region, employees people, like there's plenty of great benefits.

But there's a story that is environmentally beneficial to be, you know, burning that gas in a small turbine that doesn't have all the appropriate NOx reduction components, no catalytic converters, no NOx burners that you would have at, you know, full commercial scale power plants. You know, you're not reducing NOx emissions. And then CO2 emissions is going to be the same either way. So what about this claim that I've heard a lot of times that Bitcoin mining actually helps with load balancing? So it's sort of during, it's a buyer during times when demand might be low and maybe it stops consuming energy during peak period. So what do you make of this claim? Totally accurate.

Theoretically, that claim can be put in place and Bitcoin mining can be an amazing part of helping to load balance or load balance the grid. and to follow the shape. When you put it in place practically, though, you would likely end up in a scenario that your capacity factor of your actual mining facility is too low to justify operation. You know, I haven't done the detailed research on that, but Bitcoin mining, you know, not today

because the margins are, you know, so attractive. But a year ago, if you were trying to load follow and provide value to the grid by turning on and off based on supply and demand of the grid, you would be running at too low of a capacity utilization factor to be in business. The economics wouldn't work out. So when you see these country level comparisons of Bitcoin's energy usage and countries like Argentina, What do you make of those? You know, I, my initial reaction is always that they're just, you know, hyperbolic

headlines to get clicks and likes. You know, at the end of the day, the U.S. consumes, like the average person in the U.S., their footprint is roughly 9,200 watts. If you want to go compare that footprint to someone like in Switzerland, right, it's 4,400 watts there. In Argentina, it's 2,500 watts. Point being that countries consume electricity differently. I just don't understand the logic in comparing mining consumption to a country and saying that it's outrageous that Bitcoin mining consumes as much as energy as Switzerland needs in order to operate, right?

To be fair, you'd have to put that in context of all the energy savings that Bitcoin is resulting in by introducing efficiencies into financial transaction operations. But those numbers aren't used and they're not heavily researched. So, again, I think it's just an irrelevant number that ultimately is just trying to drive clicks and maybe create some. discord within a community. Yeah, it's like how do you quantify the efficiency of being able to send the equivalent of a wire 24-7 any day of the week as opposed to we just had a bank holiday here and I was trying to send a wire to wait three days to do it and then, you know, the bank had an issue and it didn't send it.

Yeah, I think that's a really good point. Yeah, no, you can't and it's just a very difficult number to wrap your arms around, you know, just all the other disruptions that are taking place, like through decentralized finance. There's just so much that's going to be disrupted in a short term. And I truly think it's going to be a substantial net electrical savings by introducing cryptocurrency, Ethereum, Bitcoin, all of its derivatives into finance versus, you know, what we've been operating with for literally the last 30, 40 years. So, Brandon, you guys are among or possibly the largest North American, ETH miners. You have the long track record in the wholesale energy trading space, energy analytics.

You've seen these debates about the consumption of Bitcoin Ethereum. What are your kind of final thoughts on that? What are your concluding thoughts on that debate? You know, in aggregate, like Ethereum consumes around 750 megawatts of generation capacity. Bitcoin is sitting right around that 8,000 megawatts level. Like in aggregate, it's, you know, you're roughly 9,000 megawatts or so. So you're talking about 0.3, 0.32% of global generation to power cryptocurrencies. I just think that we're going to have such a substantial net savings from cryptocurrencies being introduced into global finance relative to what we've worked with historically.

I think they're wonderful technologies. I'm super excited about the integration. I couldn't be more thankful for all the hyper-intelligent individuals who are contributing to these projects. A lot of them anonymously as well. They're just doing it because they believe in it so much. We're just very excited about the outcomes and what's going to happen over the next decade.

Well, Brandon, this has been an absolute education for myself and certainly for the listeners too. I'm going to put as many of these resources as I can into the show notes. Thanks for coming on. This has been great. Of course, yeah. And always around if you'd like a follow-up. And how would you recommend that people reach you or follow your work at CoreWeave?

I can be reached at Brannon at Corweave.com. Recently joined Twitter under the handle Brannan MacB. But feel free to drop me a note wherever you can find me online. Always happy to chat about cryptocurrency or electricity markets in the U.S. or general energy production. All right, Brandon. Thanks again. It's been very. Thanks, Hank.

Chat soon.