{ "episodeId": "SLP487", "speakers": { "stephan": { "name": "Stephan Livera", "role": "host", "tag": "STEPHAN" }, "bitcoin_mining_energy_debate": { "name": "Bitcoin Mining Energy Debate", "role": "guest", "tag": "BITCOIN" }, "guest_2": { "name": "Guest 2", "role": "guest", "tag": "GUEST" } }, "segments": [ { "speaker": "stephan", "time": "00:09", "start": 8.55, "text": "Hi, you're listening to Stephan Livera podcast, a show about Bitcoin and Austrian economics brought to you by Swann dot com. So for episode four hundred and eighty-seven, we have a debate on Bitcoin mining energy and the impact of it with Digiconomist Alex and debating him is Ben Gagnon, chief mining officer of BitFarms, who joined me on the show recently and re-joins for this debate. It's more of a casual style, as that was what was agreed. So we talk about Bitcoin mining models, when do they break down, Bitcoin mining machine efficiency assumptions, the media referring to scary numbers in the bull cycle and then forgetting about it later on, as well as these transaction per second metrics which can be very misleading. Now, before we begin, Plan B Forum Lugano is coming October twentieth and twenty-first. I had a great time there last year, and I'll be there again this year, and I'm looking forward to seeing you there if you're interested in sh- showing up. Lugano is a city in Switzerland taking on Bitcoin adoption. They have hundreds of places you can spend satoshis over the Lightning Network. There are cafes, restaurants, bars, clothes stores, barbershops, hairdressers, watch stores, and more. So it's really real world adoption. This forum or conference will host awesome speakers like Nick Szabo, Adam Back, Paolo Ardoino, Prince Philip, Jukka Muzuka, Mike Peterson from Bitcoin Beach, and many more. There's going to be a main stage, a Lightning and peer to peer stage, as Relating to self custody, multi signature, or setting up your lightning or Bitcoin node. So check your calendars, look up flights and hotels, and go to planb.luano.ch for Plan B forum on October twentieth and twenty-first. Now, when it comes to sending Bitcoin on chain transactions, I like to check mempool.space so that I can target my fee appropriately. Bitcoin is a multi-layer ecosystem, and mempool.space is a comprehensive Bitcoin explorer covering this ecosystem. You can see the mempool, the blockchain, as well as second- Layer networks like the Lightning Network, and with mempool dot space, you don't have to trust a third party. It's free and open source software, you can host it yourself with full node distributions like Umbrel and Raspablitze and others. Mempool dot space have also announced a transaction accelerator product which is coming soon, so keep an eye out for that, and otherwise go and check it out over at mempool dot space. And now on to the debate. Alex and Ben, welcome to the show. Thank you for having me. Glad to be here." }, { "speaker": "bitcoin_mining_energy_debate", "time": "02:28", "start": 147.92, "text": "Same for me." }, { "speaker": "stephan", "time": "02:29", "start": 148.7, "text": "So we're gonna have a, let's say, a casual debate. There's been a lot of back and forth about Bitcoin mining. What are the actual impacts of Bitcoin mining? What are the costs and what are the benefits of Bitcoin mining? And so let's, I guess, we can probably set some of the context. Alex, aka Digiconomist, you're a data scientist and you've been putting out some statistics on what you believe Bitcoin energy consumption is and what those impacts are Ben is the Chief Mining Officer at BitFarms, has some disagreements around methodology and perhaps other questions around the impacts of that. So I think it's probably best to start with Ben if you wanna start with any of your, concerns around Alex's modeling and then we can take it further from there." }, { "speaker": "guest_2", "time": "03:13", "start": 193.08, "text": "Yeah, sure. And, and thank you, Alex, for agreeing to, come out here and talk with Stephan and I, you know, very happy to be here, and go over this issue. You know, I've been following you for several years and, have been pulling your data and have it integrated into my model for, for about the last three and a half years. You know, b-basically, you know, we as a network are consuming a very, very trivial amount of energy. I think it's very clear when you look at all the different metrics and This were a fraction of a percent, and there's no disagreement from you or from, you know, from anybody else that Bitcoin mining's electricity consumption is a fraction of a percent worldwide. Where we get into problems here is people's understanding and people's kind of, distortion of the real world via these metrics. And that, I think there's a couple of things that are going on there. One, I think there's, there's a number of, of assumptions in your model which are flawed. Secondly, I think a lot- A lot of the KPIs are designed to mislead or, or over exaggerate Bitcoin's impact. And if I can just start maybe on, on the model design. So on, on your assumptions, the way that you calculate energy assumption is you assume that the miner's electricity costs are fixed at sixty percent of the mining revenue. And while this is a, you know, a logical assumption I think to make from an outside perspective, you know, because you have to make some sort of a, you know, a margin there for, for the electricity. The reality is, is the margin is, is incredibly volatile. For us as a company to operate, we have fixed costs, but the revenue in terms of Bitcoin price, bitcoins that we're mining on a daily basis, the transaction fees are incredibly volatile. And so we see margins that can grow, go anywhere from high eighties, low nineties, to down to low, you know, low teens, low single digits, or even zero, you know, for, for periods of time. And when we assume that the electricity consumption is sixty percent year round What that means is that when we're in a rising price environment, and when we're mining economics are bullish, that means that we keep inflating the amount of electricity that we're consuming because we're assuming that the miners are growing at the same rate that the price of Bitcoin is, and it's not. It's not possible because we have real world lag effects with building out energy infrastructure, acquiring new miners, deploying new miners for us to catch up to the incentive. And so, you know, in a situation like last, the last bull market in twenty twenty-one, we had a- situation where the prices is going up and so is your electricity consumption index. But then we have the China mining ban, and the China mining ban had a very real, very measurable, very noticeable drop in hash rate of about sixty percent. And, you know, what happens when your model, assumes fixed costs on electricity on sixty percent of, of the mining reward? That means that that sixty percent of the hash rate dropped off, that means the mining revenue propped up for everybody else who's plugged in and the profitability skyrocketed. But our operations- It didn't change. Our electricity consumption didn't change, it's just the profitability increased. And what your model shows is that we'd skyrocket off to like twenty-three gigawatts, something double, double the electricity consumption that your model shows now, two years ago, and, you know, something that can be checked with the efficiency, by comparing that with network hash rate on a per tera hash basis. And it just, it just goes out the door. And so that's my, my, my big problem is the model isn't designed according to how operators operate. You know, me as chief mining officer of BitFarms, I'm looking at profitability, I'm looking at operating costs, I'm looking at relative competitiveness on my cost efficiencies as a miner. You know, I don't have fixed costs, and, you know, in a situation where the price is going down, I'm underclocking, I'm trying to find ways to cut my costs, I'm trying to curtail more, I'm trying to do all these different things to drive down my energy consumption and drive down my costs because of that energy consumption and improve my, my, my profitability. So First part, and that's a pretty large part, so maybe we just, maybe we just start with that, that single assumption, the sixty percent assumption." }, { "speaker": "bitcoin_mining_energy_debate", "time": "07:19", "start": 439.3, "text": "Yeah. So I don't know, maybe we should first, you know, try to find some common ground. I mean, there might be at least one thing we can maybe agree on, which is, let's say the minimum energy consumption of the network. I, I don't know if you have any problems with that, but that is one other estimate that is featured, well, also by Cambridge, and, Calculation how? I mean, then you just take the network hash rate, you look at what type of machine is available in the market, what is the energy efficiency of that machine, and then you just multiply the hash rate with the energy efficiency of the most efficient machine out there, and you get a certain number which then represents a lower bound for what we think the energy consumption of the network is. And if you look at that today, at least for Cambridge, they put it at about seventy, tera- Kilowatt hours of electrical energy consumption a year, and then the only guarantee we have is that the real energy consumption is probably going to be higher than that lower bound, but that's where this discussion starts, and that's where all the assumptions come in, and that's where all the models come in, because that's where we're trying to figure out what is really going on with the energy consumption of the network. And then there's, different ways to do it. I mentioned Cambridge, well, you specifically mentioned the sixty Percent assumption in my model, maybe just to first of all clear up a misconception about my model is that it assumes that miners spend six per-sixty percent of their revenues on electricity in a long run equilibrium, so it actually varies over time. It's more like the sixty percent in the long run is a point where energy consumption over time will hover to, but it's not going to be sixty percent at all times. And actually, you can find the current percentage typically at the bottom, of my Bitcoin energy consumption index page. at the moment, it is in that direction, simply because, well, the Bitcoin price isn't super high, The hash rate is, super high. Cambridge is actually estimating the network to consume even more energy than I'm, guessing. but I just wanted to highlight that, it's not always sixty percent. It's more like this dot on the horizon where over time the number will hover towards, but it's not always going to be sixty percent. And what you said can be true, it can be much less. it can also be more in the short run, like you said, there are no fixed costs, and what the only thing that matters in the short run is your electricity cost, huh, what, what you need to keep your machines going. And then in, in the, in the long run, everything becomes fluid, everything becomes variable, so even your hardware investments in the long run become variable, but in the short run, that doesn't play a role. In the short run, hardware costs are what we in economics consider to be sunk cost, huh, you made the investment, you can so the only thing that matters going forward, your prospective cost is just the cost you need for keeping your machines running, which is mainly going to be electricity cost. So in the very short run, electricity cost can go to, well, worst case, one hundred percent of your revenues, and that's where you reach the point where you're going to be forced to cut your machines. Hey, I mean, if you start making a loss, you're going to turn them off. So I, I don't know, I just wanted to clear this, this thing up. It's, it Sixty percent, but only in a long run equilibrium. It's not-- it, it varies over time." }, { "speaker": "stephan", "time": "10:59", "start": 659.19, "text": "So can I then ask the question, if it's true that as Ben mentioned, at some points in the Bitcoin cycle, that, yes, it might be, as Ben mentioned, in the high eighties as a percentage, maybe the point is that it only spends a very small amount of time at that level and a lot more time below that. I'm curious, Ben, if you have any thoughts there." }, { "speaker": "guest_2", "time": "11:19", "start": 678.59, "text": "I mean, it, it just depends on the, the direction of the market, right? Because Bitcoin price, when it goes up, goes up faster than we can deploy new miners and grow our hash rate or, or grow our footprint in terms of megawatts. And so there is a very large lag effect between rising prices and faster deployments of, of hardware and increasing amounts of electricity consumption. And so you do get, you do get a gap whenever that happens, but it's, it's really dependent on the Bitcoin price. You know, I, I'm, I'm glad that you brought up the I actually don't, I actually don't disagree too much with your lower bound. It's not your low-- lower bound that I have too many issues with, because your lower bound is at least fluctuating with daily changes. It's your normal bound that I have issues with, because your normal bound doesn't fluctuate daily. there are, there are periods of times where it goes months on end where there is zero percent change, and it's not being impacted at all by any variables. And Bitcoin is very volatile when it- It comes to Bitcoin price, Bitcoin mining economics, Bitcoin network hash rate assumptions, and there, there are periods for months on end where there's zero volatility. There are also periods where there's months on end of a fixed growth rate, zero point three percent for, for months, and these aren't factors that are responding to the changes. I think your lower bound actually does respond to the changes. I think your lower bound is, is following your, your methodology as best as I can tell. It's your normal bound which is feeding your KPIs, which is feeding the numbers that people screenshot. Which is feeding the Argentina and the country comparisons, which is what everybody cites in every newspaper article, which is not following those numbers. And, and that's where, that's where I have the big issue with your model. there is a big difference between the normal bound and the lower bound, and what we saw in the twenty twenty-one bull market was your upper bound went up to twenty-three gigawatts in, in, in a matter of months. I mean, the ability for the industry to, to go up and add ten gigawatts in a period of months is, is, is just Logistically, it, it doesn't exist in terms of the equipment, in terms of the electricity supply, none of these factors exist. And then now what we saw is, is during the bear market, your lower bound crashes back down, or your upper bound crashes back down to your lower bound. And so for several months now, your expected has been at your, at your lower bound, right? But there was, there was a very long period of time there where it's, it's, it's just gapped up, and that's, that's the big problem here, you Look at the energy consumption, and the way I like to verify this is, is take that energy consumption index you have on, on a terawatt hour annual, I like to an- or I like to take that and, and deannualize that to make it kind of a daily- a daily rolling figure in terms of, of megawatts, and then I divide that with the hash rate. And what I get with that is I get a industry overall watt per tera hash assumption based on network hash rates, which are mathematically, you know, statistically verifiable by probabilities of finding so many blocks in a certain day and, and a difficulty period. The network assumption is, is especially if you're looking at a seven-day or longer average, is quite accurate. Like a one- One day average isn't, but, you know, longer periods of time are quite accurate, and your watt per tera hash assumption figures go, go off the chart with that. We go from just before the China mining ban, we had a watt per tera hash assumption on your chart of sixty watts per tera hash according to your data, at the peak of the China mining ban, we had a hundred and sixty-three watts per tera hash. And so this is, this is showing how the model is breaking down here in these, these bullish mining market environments by, by assuming that that electricity cost is rate that the mining economics are going up, we're vastly overestimating the amount of electricity being consumed. And, you know, when you're looking at, okay, how does that work from an operator's perspective? So let's, let's take it from my perspective. I'm chief mine officer at BitFarms. The mining ban happens in China. We've got all of these different miners sitting on our racks in Quebec, powered by hydro, we're ready to go, we're sitting here, we're absorbing these mining economics. What do we do now? Well, according to your model, we We'd have to use S-nineteens or S-nines or even S-sevens to get down to like a hundred and sixty-three watt per tera hash efficiency, drastically reducing our throughput, drastically reducing our revenue, and for, for what purpose? Like, where, you know? There's, there's no way that that math checks out. it, it's not possible to go from a sixty-watt per tera hash to a hundred and sixty-watt per tera hash efficiency across the entire industry In the real world, it defies all the incentives, you know? And the reality is, is that I'm a profit maximalist. I've, I've no, no shame about saying I'm a profit maximalist, and I think there's a lot of negative opinion put out there towards Bitcoin miners as we're greedy miners, we don't care about things, we don't care about the environment, you know, all we care about is the profit of mining Bitcoin. It's like, we care about the profit of mining Bitcoin, doesn't mean we don't care about anything else. Like, you We're a company that has powered almost every single Bitcoin that we've mined with hydropower for almost the entire existence, of our operation. I wouldn't say that's something that necessarily, you know, every investor is, is very focused on, but the reality is, it's like we as a company, we, we focus on, on profit and we focus on these other areas where we can, we can get better deals, and we primarily invest in areas where the energy is overlooked, it's underutilized, and it's often being wasted. Like, that's, To reduce our costs." }, { "speaker": "bitcoin_mining_energy_debate", "time": "16:58", "start": 1018.38, "text": "Okay, so there's a lot to respond to, and let me first of all say that what PoTera hash isn't an assumption of, my model, and also not of Cambridge. Well, they make a PoTera hash average, which they then apply to, the network to do the calculations, but it's not like they assume any direction on that. I mean, it, it will go down over time in the Cambridge model, in my model, if you calculate it. Well, you can calculate it from the, energy consumption it It's out, but the thing is, the what bitera has, it, it includes energy consumption from all sources. So, you know, it includes that, maybe people are going to go crazy in some kind bull market where they even start cooling their machines with air conditioning or start heavily overclocking their machines. I mean, there's, there's no, there's no assumption on that in my model. There is an assumption on, let's say, the time it takes to get to that long run equilibrium point that I mentioned before, and I think that is an important- Important one, because if that one is very short, what you're going to see is that the Bitcoin price goes up and immediately the estimated energy consumption goes up. And, the, the thing is, if you compare my model to Cambridge, I'm actually the only one that accounts for manufacturing lag. If you look at the Cambridge model in a bull market, their model always goes up right away the moment the price starts going up because they don't have anything to account for that. They don't account Account for manufacturing? I do account for manufacturing, but my lag is set at one year, and that is a fixed amount of time based on calculations made in twenty eighteen. So it is totally possible that that doesn't apply more recently. it's, that is the assumption of my model that the manufacturing lag is going to be one year, and yeah, you can be totally right, it can maybe, especially when there's a ship shortage, it may be longer. But then again, when the market starts going down, maybe it's shorter because you don't need to put produce that much anymore. there is simply not enough new information to really make a new calculation on that. And by the way, the sixty percent is also supported by a, by now pretty old survey by Cambridge several years ago. They did this survey among miners and then they found that close to sixty percent of their costs were going to hardware. But again, this number might be outdated. Also the five cents per kilowatt hour, which is a crucial- Assumption in both my model and that of Cambridge is coming from the same survey, I think it's twenty eighteen, twenty nineteen, that number. You know, we've had an energy crisis recently. We, we, we have no clue how this number has developed. We know that there are some miners in the US that pay less than that. We know, for example, Riot Platforms is paying just three cents per kilowatt hour. We recently saw a nuclear miner in Pennsylvania doing just two cents per kilowatt hour. So there are miners paying less, but we don't- Don't really know how the average is developing over time. And these are known, unknowns, it's, we know that these are assumptions, and these assumptions can be wrong. The thing is we don't have anything better to go on than the information that we're currently applying in these models. so, and then you specifically mentioned China, which is also a very interesting case because my model assumes competitive markets, and the thing is, in China's, situation when half of the network suddenly gets, Well, kicked out, forced to go offline, then it breaks the competitiveness assumption. So my model doesn't work when a ma-a massive player is banning crypto mining because then the competition is, well, reduced significantly, which is great for you if you're located in Canada because you're basically getting free money, your, your income doubles. But yeah, then the model stops, working, It does return back to a competitive market. I mean, we, we do see that in the Cambridge model also how they do, they do, respond more to the hash rate. So in that case, their model probably works better, and you do see a drop in their data at that point, which then you see it quickly recover after just a couple of, months. So, yeah, it, it does kind of indicate that, yeah, these markets are essentially competitive, People are trying to get these machines up and running as fast as possible when it is profitable to do so. It's just that there can be distortions, and if you looked at my website during the time of the China ban, there was actually notification on the website that, had this was impacting, the accuracy of the data. And the reason I didn't adjust it was that, it's not just energy consumption, estimates that I put on my website, but also carbon emission estimates, and it gets really complex located if you wanna adjust for that, especially during the summer months in China, because historically we saw that China during the summer months, miners were using hydropower over there, so the carbon emissions related to that were actually, low. And then of course, I can adjust the energy consumption estimate, but the energy consumption estimate, the way it translates to a carbon estimate is on a fixed factor for the carbon intensity of the energy consumed. So then I have to make two, adjustments. I first of all need to adjust the energy consumption, but then the carbon intensity, well, we later learned was going up because these mines were relocating, to Kazakhstan, they were relocating to the US. And well, at the time, I- I didn't have the information to be making that adjustment, then I can start guessing and I can start putting out random numbers, but what I thought was the most fair would be to just put up a notice that, hey, this is changing. During the summertime, miners in China use hydropower mostly, and I think the carbon estimate that I'm putting out is actually the most important one because that's, you know, you can use a lot of energy, but ultimately environmental impact matters, and likely the carbon emissions at that time weren't going down, so the energy consumption was Probably going down when China, banned these mines because we saw half of the network hash rate forced to relocate, but the carbon emissions related to that, well, because they were going to be using hydropower in the summertime, probably didn't drop as much. There's just no way you can do an estimate for that, so I did put a note saying that, Cambridge actually had more or less the same problem because, you know, they, their data did show a decrease, but the thing is they don't know how the mining landscape looks like locally. In China, they don't know if those machines running over there are relatively, all the devices relatively inefficient, whether the cooling that is being applied is very high. So they probably would have had to do some kind of adjust- adjustment for that anyway, but because in their default model, they just assume an equal distribution of the via, they put a basket of machines in their model and they assume that there is an equal distribution of machines being used. So if you have, for example, ten devices, one of them is an S19, one of them is a I don't know No, it's nineteen Pro. Every machine gets a ten percent share, and that, and that doesn't change. only thing that changes is if there's more machines being added, then the distribution does change, but otherwise you're looking at an equally weighted basket of devices. And we just don't know. we, we, we-- There's no data to say, okay, when China banned these machines, in China they were all using old S-nines, for example, so, maybe it cut a lot more energy used than Cambridge estimated, or maybe it were all newer machines, we don't know. There, there, there is, there is these unknowns, and I don't think it's, it's a big problem as long as we know that these unknowns are there, and we know how this is being used, and we know how this affects, the outputs of these models. So, yeah, these models sometimes just don't work. My model doesn't work when there is a really massive mining ban in China, and, and, and not Going to say it does. What I typically tell people is, hey, if you're going to be looking at energy consumption estimates, it's probably best to look at my model and the Cambridge model combined, because they tell, depending on the situation, very different stories. and one model can be more correct in one situation, one model can be more correct in another situation. So for example, during, the most recent bull market, or even this year, huh, we see the Bitcoin price has Kind of recovered, since, and the low point of last year. Now we see that, in the Cambridge model, the electricity consumption of the Bitcoin network is reaching, record highs, huh, very recently, higher than ever before, whereas in my estimate, the numbers are currently a lot lower, and that's because, like I said, I account for manufacturing, installation, you know, a delay on getting those machines up and running. compared to the price going up, whereas Cambridge, they don't do that, price goes up, energy consumption, estimate goes up. So in that case, maybe you wanna look at my model instead. You can't often verify either model. So, and this is also a bit of a generic problem. Well, that's, that's" }, { "speaker": "guest_2", "time": "26:38", "start": 1598.16, "text": "actually why I brought" }, { "speaker": "bitcoin_mining_energy_debate", "time": "26:39", "start": 1598.94, "text": "up" }, { "speaker": "guest_2", "time": "26:39", "start": 1599.32, "text": "the Whopper TerraHash, because I, I know it's not a model that you have or a metric that you have on your website or in your dataset, but it's, it's how you verify what's rational And, and let me just, let me just share my screen here, just so I can make this clear." }, { "speaker": "stephan", "time": "26:54", "start": 1614.11, "text": "Yeah, sure. And just make sure to explain for the audio only listeners, 'cause most are audio only, but just, yeah, if you could just explain what you're pointing at." }, { "speaker": "guest_2", "time": "27:01", "start": 1620.7, "text": "Yeah. And, and it's just showing up here now?" }, { "speaker": "stephan", "time": "27:03", "start": 1622.52, "text": "Yep." }, { "speaker": "guest_2", "time": "27:03", "start": 1623.3, "text": "Okay, perfect. So what this chart is showing is, is basically it's, Digiconomist's electricity consumption, it's your main estimate, divided by the network hash rate average over thirty days. The same thing is" }, { "speaker": "guest_2", "time": "27:20", "start": 1640.21, "text": "ECI and the red line, and the red line, as well as for your model here in Cam- for Cambridge here, in the yellow. And what we see here is that, you know, here's where the China mining ban, Comes into place here is twenty twenty-one, and we start seeing this skyrocketing watt per tera hash figure o-on your chart. and this is because your energy consumption is going up in a way that just isn't tracking with, with actual operations on the ground. At the peak here, what does this say the peak is? It says the peak is one hundred and ninety-six watt per tera hash. You know, that is, that's not even an S seven efficiency, and what we're talking about here is, is not the, you know, a single- Operators' choice to plug in some older miners and take advantage of some lower cost electricity. This is an industry overall average. So that means that the industry on average is running about an S7. You know, that means that there are people who are running worse than that in order to get that average, because there are certainly people who are running better. And this is how, you know, looking at this, you can see that this model, you know, was within reason for a long period of time, but during bull markets, it breaks down. And then you've created Crash back to reality, and if you see here, you know, your blue line or my-- your blue line and my red line are basically the same and have been the same for all of twenty twenty-three, right? So this is because you've crashed back to your lower estimate line here, back at the beginning of the year. This whole over-- you know, when we switch over to the chart, this is your, your upper bound and your lower bound, and this is your matching down to the lower, lower bound where it stayed at the entire year. And then here's, here's the gap This number that's up here, you know, which one is not changing on a daily basis, zero percent changes, fixed, flat for months on end. Same thing here, same thing here, growing at a fixed rate of zero point three. Like this isn't a model that's responding to changes. You know, we have massive changes that are going on in this network, and what I'm trying to do is, like you said, we've got known knowns and we've got known unknowns, and we're trying to verify this. And I'm looking at your data here, I'm saying, actual, you know, electricity consumption for the entire industry, but I can tell you with certainty that this is wrong. There's no way that a fixed volat- you know, a fixed growth rate of zero percent or zero point three percent per month, is accurate. And there's no way that an, an industry average that shows that the industry is consuming around one hundred and ninety-six watt per tera hash when just prior to this bull run, that same industry was operating about sixty-five watt per tera hash. There's no way say that that data is accurate, and that's the point. you know, i-is there a way to, to track this better? Absolutely. But, you know, there's, there's a lot of different ways that we can argue about this and we can debate the best, the best methodology for trying to calculate the, the electricity consumption. But, you know, we also have to verify. And, you know, what we can see here by trying to do a, a comparison, 'cause we know the network hash rate estimates are pretty accurate, especially on longer time frames, as I ta" }, { "speaker": "guest_2", "time": "30:31", "start": 1831.28, "text": "Known, and when you blend them together, you get that wat per tera hash, and this is a clear indicator that something is very broken in the model or purposefully, manipulated. And, and that's, you know, when I see periods of volatility for months on end, I mean, this, this extends almost two full fiscal quarters, zero volatility in twenty twenty-two, when we had massive volatility in the underlying Bitcoin price. There's no change until we have the, the Bitcoin collapse, you know, breakthrough through thirty- K down to twenty K here, in twenty twenty-two. That's the only time this changed, and, and that's how I know that the data is wrong." }, { "speaker": "bitcoin_mining_energy_debate", "time": "31:09", "start": 1868.73, "text": "So I, I think maybe just to start with, there's roughly three states in which my model can be in that can either be this bull market state where the market or the energy consumption of the market is just going up, as miners expand their business while the price allows them to put more machines into operation so there is a state where it's profitable to add more machines to the market, and that is when, now let's say the costs of, mining are very low, so when the profit margins are very high, you are going to be expecting that miners are going to be using or employing more machines. If you cross the sixty percent, how do you have this interesting zone between sixty and hundred percent where you get a flat energy consumption estimate? Because in that zone, what's going to happen, at least according to the model, is that It doesn't really pay off to invest more in more machines, but at the same time keeping the machines that you have running still generates money. And you need a certain amount of money to, you know, get these new machines, and you need to make the money on those, or you, you want to get, you want to earn those machines back. but at some point the profit margins may become, too thin, too, for that to be an attractive option. but at the same time, you're not shutting down your machines. That's like the, the sixty to hundred, percent zone. And then you, when you hit the hundred percent, that's when, the energy consumption, just dives off a cliff. that's where, you know, miners aren't not just not investing in, new machines, but they are just shutting down their existing devices. Now, of course, there are some nuances in reality because, adding new machines may always be, profitable here and mine as nineteen pro for a long time, and maybe probably even now, adding those machines to network is still going to be profitable. But then, if you turn on those machines, how you're going to also witness some machines dropping off at the tail end. so it can be the case that new, newer, more power efficient machines are still coming off on line while older- Relatively inefficient devices are dropping off at the tail. in any case, if that, that occurs, then, hey, you can be at a state where energy consumption is sort of stable. but that's, that's roughly the, the three states that my model can be in. So you either have growth, you either have stability where, you know, e-energy consumption going up isn't expected, or when you hit one hundred percent, at least based on the cost input of five cents per kilowatt hour, it doesn't pay off Anymore to keep the machines, running. Now, that, that just with the general state of my model, I already mentioned that during the China ban, you probably shouldn't be looking at this number, the carbon emission number was probably still valid during that time, but the energy consumption estimate certainly wasn't. And then when you look at the watt per tera hash hour estimate, that you just showed from starting from 2020 to, recently, one of the issues I've been having with that is that the starting point is probably Too low, because back in twenty twenty, I did manage to do somewhat of a verification. It's not, you know, i-ideally the way you verify this is you go to all miners in the network, you ask their energy consumption, you add it all up, and then, bam, you have your perfect number. Unfortunately, that doesn't work, especially nowadays, I mean, China kicked out these miners, twenty percent of the network, at least according to Cambridge, is still operating in China, which means they're operating illegally. According to their most recent estimate, which is probably outdated by now as well, fifteen percent of the network was in Kazakhstan, the majority of that also taking place illegal. It, it just means that, you know, at least back in the start of twenty twenty, sorry, twenty twenty-two, when Cambridge put out the last update to this data, a third of mining around the globe was happening illegally. So you can't just go to them and ask like, \"Hey, what's your energy consumption?\" They, they actually want to stay under the radar, and you can't, you can't really Proper survey, that's problematic. But what I could do back in twenty twenty, when, how, in that time, all these major hardware manufacturers wanted to go, and do an IPO, well, Bitmain was one of them, is still the largest manufacturer, I think, out there today, that also wanted to do what, and they put out a lot of information regarding the machine, the machines they produced in the prior years. If you have that information, you can kind of do a adjusted minimum Calculation, whereas you still calculate a minimum energy consumption estimate for the network, except you correct it for the amount of machines that have been produced according to the IPO filings. And at least following that approach, I established that over twenty twenty, both my own model and the Cambridge model were showing too low results for different reasons. the Cambridge model was severely, overestimating the share of newer machines and underestimating, the share of older devices, especially as nines. and my own model was simply, too pessimistic on the time lag and the time-- the amount of time it takes to get new machines. Up and running. So, and now we're looking at a what Patera hash where we can already establish that they, you know, if you look at the start of that number, it's probably way too low, yeah. So, that, that is a bit of an issue, yeah, because you do, hey, you kind of expect when you look at the number to go down over time, but in this case, and the starting point is already, too low, and then of course you have this weird situation in, twenty twenty-one where China kicks out These miners and then the energy consumption, the watt per tera hash estimate just goes through the roof because my model breaks down during, that specific event. there's no competitive markets at least for a short while. So in terms of verification, what can we do? What can we do? I, I, I do agree that it can help you as a sort of, you know, sensibility check, but at the same time, given these constraints It's, it's also very difficult to put any hard conclusions along with that about whether these models are right or wrong, and I think it's just not a Very solid way of doing verifications, huh?" }, { "speaker": "stephan", "time": "37:52", "start": 2272.4, "text": "Okay, but I-- but I mean, let's take that. I mean, then why are the-- these models are being used to sort of drive fear in the media amongst people who don't, who don't understand this? And if you're, you yourself are saying, \"Well, certain aspects of the models can't easily be verified,\" because, yes, some of the miners, miners would prefer to hide that information or it's competitive information, then are you okay with these numbers being used to basically attack the industry?" }, { "speaker": "bitcoin_mining_energy_debate", "time": "38:16", "start": 2295.68, "text": "Well, you know, the thing The things that I often tell people who are interested in this is just look at the lower bound. You know, there's no discussion about the lower bound, and if you take that number, you have plenty to go on for your policy discussions. We get, I would, you know, if you look at Cambridge today, they're saying, okay, the Bitcoin network is consuming one hundred forty terawatt hours of electrical energy per year. The lower bound is seventy. Well, that factor two isn't really going to make all the difference from the policy perspective, and the factor two is way too low because ultimately Policymakers are concerned about this, you know, they look at Bitcoin and they're looking at the existing financial system and they see that, okay, you know, Bitcoin is consuming somewhere in this range of electrical energy consumption, at the same time the, the amount of electrical energy consumption, related to the, current financial system is estimated to be in the same range as this number. there was a very recent paper in the Journal of Cleaner Production that actually put The total energy consumption for, the f-financial industry, so that means, branches, ATMs, everything, related to banknotes and, cashless payments at one hundred twenty-eight terawatt hours of electrical energy consumption a year. And, you know, their concern is gonna be like, okay, you know, whether Bitcoin is, going to be consuming seventy or one hundred forty kilowatt hours, you know, ultimately Bitcoin is using all that energy while the network is only processing one hundred million transactions a year, while the regular financial system does two point one trillion cashless payments plus an unspecified amount of cash transactions, and that's, and that's where the disproportioneness starts to come in, and a factor two doesn't, you know, doesn't fix that." }, { "speaker": "stephan", "time": "40:05", "start": 2404.62, "text": "Back to the show in a moment. The lead sponsor of Stephan Livera podcast is Swan Bitcoin over at swan dot com. You can sign up for safe and easy Bitcoin buys with recurring purchase plans and one time buys. Now, many people know Swan as an auto stacking application or website or service, but also Swan Private exists. Swan Private is a trusted partner on your Bitcoin journey. This is a service especially for high net worth investors or those people who need support for retirement, trust, or corporate accounts. With Swan Private, you have You have a dedicated Bitcoin expert, you have access to exclusive events, and you also receive original Bitcoin and investment research. So if you are a high net worth investor, or you have friends and family who might fit that category where they need some additional guidance, send them over to swanprivate dot com. Over at swanprivate dot com, you can get a trusted partner on your Bitcoin journey and get additional guidance that you need to help you navigate your Bitcoin journey. And last but not least, coinkite dot com, the creators of my favorite Bitcoin hardware wallet. Or hardware signing device, the coldcard. The Mark IV is the latest edition. It's got two secure elements, it's super secure. You can use it easily when you are directly plugging it to the computer, if you are a beginner, or if you're intermediate or advanced, you can use it in airgap mode, where you get a micro SD card and ferry the information back and forth between your computer and the software wallet, such as Specter Desktop or Sparrow or Electrum that way. And it also has NFC support, so you can use it with, with software such as Nunchuck You can of course disable that NFC support if you don't want that. It's a very reliable performer. There's a range of different configurations you can use the Coldcard in. You can use it in single signature, you can have seed access, you can use bip85, you can use it as part of a multisig, you can have passphrase's, brickme pins, all of these different features, and you can learn more over at coinkite dot com and get a discount on your Coldcards with the code livera. And now back to the show. Okay, so, okay, fine. But even on that point, I think it also depends on what you think are the consequences of the monetary standard that we're operating under, right? And so this is where I would say, and Ben may have his own views on this as well, but the way I see it is you would compare a fiat money standard versus a gold standard versus a Bitcoin standard, and there are certain consequences of living on a fiat standard, and those aren't just the banking industry. You know, you could say, \"Well, what about the military? What about other aspects? What about So at least, that's how I would answer that question. I'm curious how Ben, you think about it." }, { "speaker": "guest_2", "time": "42:34", "start": 2553.64, "text": "Yeah, I, I think about it a little bit more differently, especially when it comes to transactions. Like, as a miner, I think my primary role is providing security to the network, and secondary is processing transactions. It's more important that nobody gets their Bitcoin stolen out of their wallet than somebody gets like the most rapid transaction speed that they can possibly have. And when we look at how Bitcoin works as an actual payment- Payment medium. You know, Bitcoin transactions are comparable to SWIFT and Lightning transactions are comparable to Mastercard. You know, nobody's out there doing hundred million transactions every day, you know, on, on, on the SWIFT network for cans coke, right? There are different payment mediums and different payment settlement networks for different kinds of transactions with different purposes and different values associated with those transactions that give different degrees of certainty and are also associated with Different processing times. So, you know, that's the more comparable, and when we're talking about, you know, just transactions on layer one, that ignores all the transactions that are happening on Lightning Network and that growing ecosystem, which is, which is really where the payments are supposed to be being done, especially for smaller goods and, and services. And the second thing is that-- Sorry, I lost my train of thought there." }, { "speaker": "stephan", "time": "43:51", "start": 2630.55, "text": "So, so not just the, Lightning Network transactions, but-" }, { "speaker": "guest_2", "time": "43:55", "start": 2634.61, "text": "Oh, right. You're, you're, you're ignoring everything that's happening on the exchanges. Right? Which is, which is probably the vast majority of, of actual Bitcoin moving between people's balances is, is it actually just happening on the exchanges, people just trading every day? Like that's, that's the vast majority of transactions, is the buy and sell on the exchanges, and it's all being, you know, ignored. So, I mean, personally, I don't really think about it in terms of the transaction fees. I think it's a pretty-- I don't think it's a very good metric to look at things. I, I think it skews the, the way that Regards to what's actually happening on the ground and how people are using Bitcoin and how people have been using Bitcoin for years. but, but going back to your, your point there, I'm, I'm very glad we can agree, we found some agreement here. Your model breaks down in bull markets, and the-- That's one of my big challenges with this, is because nobody really cares, to be honest with you," }, { "speaker": "bitcoin_mining_energy_debate", "time": "44:51", "start": 2691.09, "text": "when, when there is no competition, that's when it breaks down, not, not necessarily in all bull markets. Okay, well, okay, so" }, { "speaker": "guest_2", "time": "44:57", "start": 2696.89, "text": "we found some The situation is that, like, look, Bitcoin goes down to sixteen K, nobody talks about Bitcoin's electricity consumption. Bitcoin's at sixty-seven K in a bull market, everybody's talking about Bitcoin electricity consumption. And so your model is designed to inflate and overestimate at the time when people have the most attention for this statistic, right? When it comes back to reality, nobody cares. When it's at those inflated stages, that's when everybody's paying attention, that's when everybody's writing the articles, and that's when everyone- Everybody's citing the main estimate. Nobody is citing your lower level estimate. I've yet to see a single newspaper article that cites your lower bound, not one. And, you know, it's, it's all on your, your, your main bound estimate. Every single article, every single policy paper, everything is on main bound estimate. And that's the one that has the vast majority of the problems. and so, you know, there is no, there is no certainty, there will never be a certainty, hey, today this is exactly the amount of electricity, that the network is consuming, hey, this is the breakdown of the machines, like, this is something that's changing every second of the day because there's a lot of volatility both in, you know, different site situations, company situations, mining economics, there's a lot of volatility, no one will know for sure. What we can do is, Like any good scientist, is we can rule out what is obviously false, right? We can't prove that this is necessarily true, but we can prove something to be false, and that's what these metrics do. These metrics prove something to be false. you know, what can we do to, to improve that? I think there's lots of different ways to change that, to improve that, to make it actually reflect what a company like Bitfarms is doing on the ground from an operational perspective. I mean, my model is all based on what we do. us a company because, you know, it's designed to help us inform our decisions. Are we being competitive enough? Do we need to improve our efficiency? What are other people's, you know, cost position? Like, we are trying to compete on a relative basis. We, we're of the firm belief that, you know, if we're in the lower quartile or lower third of low cost producers, we're effectively hedged against, you know, lower mining market economics, at least relative to, to our peers. and so, you know- These sorts of things, we can, we can verify what doesn't work, and the Walker TerraHash is a great way for verifying that, that something doesn't work. When it comes to, you know, other estimates too, like there are other estimates that we have here. I'm not even gonna get into the, the carbon estimates because that is a, that is a really, that is a really wonky math that I don't think anybody has figured out in any industry. there's a lot of different definitions for how to measure that, how to track that, and then when you That specificity on the dataset, you don't know where the electrons are going anywhere. I mean, you look at like our operations in Quebec, it's all powered by, by hydro, but if you look at us as Canada, you might be including a bunch of nat gas into our mix, right? If you look at the United States as a whole, you might be looking at a bunch of, of nat gas and coal and, and other things in the mix. But if you look at where miners are specifically locating behind the meter at nuclear power plants, behind the meter at solar panels, behind the meter at We, like you said, we don't have a whole lot of that data outside of the public companies who are making press releases and are being audited, to make sure that that information is, is accurate. But the public companies, I think, are starting to change the space because there are now a, a significant amount of the hash rate, it's a meaningful sample size, you know, it's, it's well over twenty percent of the network hash rate is, is owned and operated by, by public companies at this point. This is a meaningful sample size, that is, is verifiable information and it's audited information, and I think that probably is gonna be a good basis for, for things rolling forward. certainly there, there's, you know, things that have been proven wrong, ne-need to be adjusted, and, and if, you know, we agree that, you know, the model breaks down in certain economic conditions, then maybe the conversation should be, \"Why is it breaking down under certain economic conditions? And what can be done to the model so that it doesn't break down under those Scenarios." }, { "speaker": "bitcoin_mining_energy_debate", "time": "49:14", "start": 2953.76, "text": "Yeah, sure. I mean, you know, when, when it comes to the question which model sh-should you like more, a l- a little bit of a fun fact is that my model gets cited mostly by crypto media during bull markets because my numbers are always, in bull markets lagging the Cambridge Index and because the-- because of the embedded lag, which is not in the Cambridge model. So if you look at the headlines earlier this year, you will see headlines by Barron's and others saying, \"Oh, Bitcoin is consuming a record amount of energy consumption.\" That's From me, that's coming from Cambridge. And if you put the numbers next to each other and next to the price, huh, you will see that price goes up, the Cambridge model is going to be the first one. So if you're concerned about what's going, huh, what are people going to be looking at in bull markets? Well The Cambridge model is probably going to be the one you like the least, but the Cambridge model has the advantage that they don't really look, you know, they don't, they don't, they don't do the energy consumption per transaction that a lot of people in the Bitcoin world and crypto world hate. But you do need some kind of metric to capture that, you know, Bitcoin isn't as large as the current financial industry. So if we estimate it to consume as much power as the rest of the financial industry with everything that's going on, cash and digital payments and, and all that The offices that are out there, yeah, then, then we need some additional metric to show that, you know, this, this, this, well, the proportionate or not, and that you, you need some kind of relative measure, because Bitcoin is, still, growing and, the fin- well, the, the, the regular financial industry is growing as well, but" }, { "speaker": "stephan", "time": "50:48", "start": 3048.44, "text": "yeah, but Alex, I think to be fair here, this is something where I've seen you commonly say this, even in your tweets, where you commonly cite this three to seven TPS No, that's not really, that's not comparable, as Ben was saying, as many Bitcoin people are saying, that's, that's, we know as an example, as I'm sure you know, every ten minutes on average there's a Bitcoin block. Those blocks have on average about four thousand transactions each, three or four thousand, maybe five thousand at the max. So it's not that-- it's not like even if Bitcoin had greater adoption, that it would go much above that, because that's basically the theoretical maximum of, to be clear, on-chain transactions per block. Now we know that He's doing one on-chain transaction but paying out to one hundred customers, is that one transaction or is it really fifty or a hundred? And then don't forget, that's again not counting Lightning, that's not counting all these transactions that rarely ever hit the chain. So how can you defend this three to seven TPS number when you are, in my view, unfairly conflating or comparing that with Swift or with, you know, or, or with, sorry, Visa, Mastercard?" }, { "speaker": "bitcoin_mining_energy_debate", "time": "51:53", "start": 3112.92, "text": "you men-- you mentioned the Lightning network as a, as a payment layer. But the thing is, had the most recent estimate that I saw for the Lightning Network was that it was processing one transaction a second. Now that estimate is a little bit outdated, but it was coming from, some crypto research institution, and, if you're interested, I can, send you the link after this conversation. But that doesn't add a whole lot on top of the handful of transactions that the Bitcoin network is doing. It adds a lot" }, { "speaker": "stephan", "time": "52:22", "start": 3142.42, "text": "more in theory, because what happens is a lot of people will extrapolate, right? They'll say, \"Oh, look, three TPS per second. Oh, let me just scale that to the entire existing modern day payment system and just naively, quote unquote, naively scale it when Lightning could do a lot more than one TPS per second. And I think it's quite clear that it is doing a lot more than one TPS per second if you think about a lot of these payments that are happening on, for example, podcasting, value for value, Nostra Zapping, and all, all of these little things that are just not being me-- not able to be so easily measured, right?" }, { "speaker": "bitcoin_mining_energy_debate", "time": "52:56", "start": 3175.8, "text": "Well, sure, Any new, number out there that says, okay, there's this many transactions happening on Lightning, so we can add them. And even if we did have a number, it's no-- it's, it's not going to be, and we're not going to be talking thousands. you know, in order to bridge the gap with the regular financial system, you need, in, in order to have a comparable energy consumption per transaction, you need thousands of more transactions per second, not just a few more. Hey, if you have one or two more per second, yeah, Or is, is, I think it's a different discussion, and we can probably have a separate podcast about that, because in the end, in order to use the Lightning Network, well, you still need to go through the main chain, and, and well, if you don't do that, then you're going to be dependent on these intermediaries again, and then that's kind of what we try to avoid with Bitcoin. I mean, the whole thing about Bitcoin is that it's decentralized peer-to-peer money, huh? That's, that's, that's the whole core of the, of the, of the system. And, and actually, you know, I always say, and, and people are surprised to hear that, that, that very concept is, is the very thing that I personally always did find interesting, I wouldn't be active around cryptocurrencies if I didn't ha-- at some point, found that a very interesting idea. It's just that in practice, we're seeing that a lot of people aren't using Bitcoin in that way, they're leaving their money at exchanges," }, { "speaker": "bitcoin_mining_energy_debate", "time": "54:24", "start": 3263.95, "text": "I had, if you try to facilitate better peer-to-peer interactions, you're limited by a main chain that is just doing a handful of transactions per second, and you still need to go through that main chain in order to get to the second layer. And, it's still very, debatable what is the real added capacity, of the Lightning Network, i-if you account for, all those limitations, right? That's something that has to be seen over time. I mean, it does The efficiency is, isn't a fixed value, huh? It's something that can go up and down over time. You know, we've seen people ever since twenty fifteen state that the energy consumption per Bitcoin trans-- per-per Bitcoin transaction on average is X amount. In twenty fifteen, they specifically said like, okay, it's as much as a US household for a period of one and a half days. Now, if you were to look at the same metric today, you're going to conclude, oh, no, that's two months. Or, m-m, it went back a little bit," }, { "speaker": "bitcoin_mining_energy_debate", "time": "55:21", "start": 3320.79, "text": "value that evolves. one of the key problems that Bitcoin has is that energy consumption is closely related to the price. the higher the price goes, the more mining activities will typically be going on because miners will simply have more resources to deploy, while the, at least the on-chain capacity is fixed. So, huh, the, the payment capacity isn't keeping up with the growth, of the energy consumption. and that is a trend that needs to be Be completely reversed, You, what you want to have going forward is that there is no more growth in energy consumption, and a lot more capacity for transactions, And then the energy efficiency can improve. the fastest way to get the, energy efficiency down would simply be to, drop mining altogether, but that's also a very sensitive discussion." }, { "speaker": "stephan", "time": "56:19", "start": 3379.09, "text": "Yeah. Look, I mean, obviously, I think- I think that just, it just really ignores a lot of the security model of Bitcoin. It ignores the reason, you know, it ignores a lot of things about how Bitcoin works. It would be like saying, \"Oh, don't fly somewhere on a plane, just ride your bicycle there.\" Well, no, like it's just, that's, it's misunderstanding. Anyway, Ben, do you have a, do you wanna have a response? Yeah," }, { "speaker": "bitcoin_mining_energy_debate", "time": "56:39", "start": 3399.38, "text": "I mean, I, I think we're already running a little bit out of time, so and, and again, this is something that we You know, Ethereum on proof of stake, since last year, consuming a lot less energy, but their landscape changed in more ways than one. And they don't just consume less energy, they also change their, security model, as you call it. the only thing we know is that they have been, they managed to run, uninterrupted since they made the switch, but there's no guarantee for the future. I, I just wanna make one last note in the context of the conversation we've been having, and that is that, you know, when, when we, we Estimates in a broader context. You know, what we can do for cryptocurrencies is unrivaled. There is no other industry where every single day you can just take a, a little piece of paper and you start, very simply making a, a very easy calculation what is probably the likely energy, energy consumption of the network, especially the lower bound, that's a certainty that you can just calculate every single day, and then with a little bit of assumptions, you can calculate a more likely number. but you can do it every single day, and you don't need a lot of inputs to do it. There is, there is no other industry out there where we have the same capability. So if you're talking, for example, about data centers in general, the estimates that are there, they have a much bigger range, you know, they, they will range from one hundred terawatt hour all the way to a thousand terawatt hours, per year, and that's because it's much harder, the, the, the quality of data available is much less, and there is Still room for improvement what we're doing in Bitcoin. I'm not saying that what we're doing in Bitcoin is perfect, and I think that maybe we can have some agreement there, but compared to whatever we are doing for every other industry out there, it's, it's unrivaled transparency. We can do these calculations on a daily basis. that we can't do it for anything else." }, { "speaker": "guest_2", "time": "58:43", "start": 3523.3, "text": "I, I absolutely agree. Any last comments? Yeah, I absolutely agree that it is unrivaled transparency in the, in the Bitcoin space, which is why everybody can go out there and verify these numbers. It's incredibly frustrating to me that nobody does. You know, news, news journalists aren't going to go out there and, and verify any of this math. They're really just looking for numbers to cite that confirm their, their headline and the story that they're trying to write. And this is something that I've seen dozens of times speaking to mainstream journalists, literally, do not care. End of the interview, hey, that's really interesting, but that's not the story we're trying to write, and, and they move on and they'll go on and they'll cite something like, like DigiConomist, or, or possibly Cambridge, as you pointed out, can get some citations too. I don't like the transaction fee methodology because it, again, it, it's not really what I think we're doing here as miners. I think we're providing security first and foremost, I think transactions come secondary, and when we look at, you know, electricity consumption," }, { "speaker": "guest_2", "time": "59:43", "start": 3583.04, "text": "mentally disagree on that point. I think energy consumption should continue to go up. You know, when we look at the electricity consumption for the, for the industry as a whole, like I said, we're a fraction of one percent. You know, the exact number, nobody's entirely sure, but we're talking about somewhere in the range of probably twenty to fifty basis points, right, in terms of world electricity supply that's going to, to Bitcoin mining. When we look at what are the, sources of electricity, how it's generated It, how it's distributed and transmitted, you know, who are the biggest consumers, where does that electricity go? The number one source of electricity consumption worldwide is waste. And the reality is, is that I as a miner, I'm not looking to go to a place like Manhattan and set up a Bitcoin mine, you know, downtown, like it's, the energy is too competitive. I'm not gonna get a price that's gonna make it profitable for me. So we go to these remote areas where nobody's invested, or people invested decades ago and, and it's been sitting vacant for- For years, you know, in, in Quebec, we have six of our seven sites are in former industrial plants that have sat vacant for years or in some cases, decades. Right? This is, this is hydroelectric infrastructure that was built to power heavy industry, which has gone, left, moved on, and the, you know, the snow still falls, the rain still falls, the water still flows through the dam, the power's still being generated, but the demand is gone, the distribution doesn't exist, and, you know, you have to transport that thousands of kilometers or miles away to get to Assume it, you'd have to agree with the different states, the different countries, the different regulators, you'd have to line up the capital to build the transmission lines, you have to-- like these are processes, processes that take a long time, and the situation in Quebec, some of these deals have been going on for over ten years and they still haven't transmitted a single watt of power. And so we have a situation in Quebec where there's a-- I can't remember the, the latest number, but the last time I checked, it was forty terawatt hours a year was being" }, { "speaker": "guest_2", "time": "01:01:43", "start": 3703.08, "text": "Power continuously, every single second of the day, and this is water that-- This is power that could be generated, but there's no local demand, there's no local transportation to bring it to a market, and so they're spilling it over the side of the dam. You know, that's enough to power about a third to forty percent of, of the Bitcoin network, and this is, this is just a source of waste. You know, when, what happens if we go in there with Hydro Quebec and start monetizing all this waste electricity? Well, Hydro Quebec is gonna have greater profitability. They" }, { "speaker": "guest_2", "time": "01:02:12", "start": 3732.98, "text": "Social programs, roads, highways, schools, hospitals, whatever kind of social infrastructure they find, or further renewable energy infrastructure. You know, right now they've got a plan to continue growing their renewable energy generation capacity by twenty forty, and they're sitting on massive excess capacity right now that they're not monetizing, and they're wondering how are we gonna pay for it in twenty forty? Well, geez, you know, we as an industry could continue to grow, we could grow a hundredfold and just be consuming marginal cost electricity. marginal electricity that nobody else wants, wasted electricity that would otherwise be on the line. And so, you know, in that point, and this is, this is where I think the industry is going, the idea that, you know, everybody's gonna be operating at six cents or, or five cents, you know, in twenty thirty, I think is, is absurd. You know, everybody is moving towards marginal cost sources of energy, otherwise overlooked, underutilized sources of energy, energy that would otherwise be wasted, because if that energy had any other demand, its cost is gonna be too You have to go with energy sources that literally have no other application, and therefore the cost is gonna be incredibly low. And in this case, Bitcoin mining is, is solving a problem that the electrical industry never thought they could solve, which was, how do you monetize and make use of the gap between electricity supply and electricity demand? Because you always have to have electricity supply exceeding electricity demand. If it ever drops below electricity demand, that's when you have a brownout or a blackout. Right? And so you always have to have this excess capacity on the line, 'cause you don't know when somebody's gonna turn on a light switch, turn on a washing machine. You don't know when, you know, people are gonna power up this, this industrial cooking oven or whatever they're using their, their power for. So you have to have that excess capacity on the line. And what Bitcoin miners are proving again and again is that that's the energy that we're, we're looking for. And if we're looking for those sources of energy, we are very much incentivizing the creation of Especially on the renewable side, it's something that you see throughout most of the publicly traded miners. a lot of this is going to new renewable projects because the, the economic incentives on renewable projects are incredibly screwed up. if you look at where the miners are going in, in Texas, in the United States, they're all going to one load zone in ERCOT, it's called West Texas. And why are they going to West Texas? Well, they're going to West Texas because they've built up all these hydro-- sorry, not hydroelectric, they've built up all in West Texas, with no demand and no transmission capabilities, and so why are they building them there? Because the subsidies and the natural resources and the cheap land and the, the regulatory, footprint of building in West Texas is relatively competitive, and so they're putting massive investment in renewable infrastructure, billions of dollars of investment in renewable infrastructure, into an area where there's no demand for that energy infrastructure. At all, and it's only been financed because of the subsidies. And now Bitcoin miners are coming in here and we're saying, \"Okay, well, let's, let's actually try and improve the economics of this project because there's a big gap here between electricity supply that's being generated by this plant and electricity demand that's on the line. There's no reason why we should be building, you know, continuously gigawatts more projects in West Texas without distribution and transmission and local demand, but that's where, that's where the projects are going. And I think the free market, works better. Like, Bitcoin isn't powered by subsidies, Bitcoin is powered by private capital trying to allocate capital efficiently, and we're going to places where that energy is, is, is wasted. It's the renewable energy sector which is funded by subsidies and is growing out in a way that doesn't make sense. I mean, there are, there are numerous areas in the United States that, are run entirely on carbon, like Puerto Rico is, is my favorite example. For example, it's an island that's run entirely on diesel. You know, it's a very meaningful population around three and a half million. And, you know, why are people not installing solar in Puerto Rico? Why are they just installing in Texas, where they're getting-- Texas, you're getting negative energy prices. Puerto Rico, they charge thirty-two cents a kilowatt hour. I mean, it, it doesn't make any sense. Bitcoin miners aren't in Puerto Rico consuming diesel. We're, we're sitting there on these tail ends of the distribution curve, absorbing the inefficiencies By absorbing those inefficiencies in the market, we're actually making a stronger market, we're making a more resilient market, and we're improving the economic incentive to drive a transition to renewables. And I think we, we see this everywhere, every-- like we see this in Paraguay with the, the Itaipu Dam, we're seeing this in West Texas, we see it in Quebec, and there are certainly examples of, you know, funding, you know, non-renewable generation as well, but mostly this is going towards renewable, because you don't want that of the fossil fuel price. You know, the marginal cost to generate a kilowatt hour from hydroelectric power plant is zero. You know, the things aren't even built, the capital costs are, are already there, the marginal cost to generate one extra kilowatt hour, zero. The marginal cost to generate one extra kilowatt hour on a natural gas turbine or a coal power plant is not zero. And, and, and this is where the entire economics are going. They're, they're going long run marginal cost, entirely overlooked sources of energy, in which case consuming larger amounts of electricity is And it also means, too, Alex, that, that the economic lifespan of those machines is going to get pushed out longer and longer and longer, because the lower we can reduce that cost of energy, the longer that like, longer, you know, you can generate profit with that miner. And so, you know, I don't wanna open up a whole can of worms here, but, you know, the miner lifespan is like five to seven years, predictably, in terms of a useful, a useful lifespan. and that's only gonna get longer and longer and longer as people Sources of electricity. So this is, this is a very, very powerful market, it's a very powerful economic incentive, and what I don't like is, is I don't like a bunch of, Misleading figures and KPIs that are prohibiting that progress, because this is the best thing that's happened to the energy industry, probably since we invented nuclear power." }, { "speaker": "bitcoin_mining_energy_debate", "time": "01:08:31", "start": 4111.19, "text": "Okay, so that would be what I, what is known as, debating as a gish gallo where you just spew a lot of arguments all together and you leave your opponent with, well, way overwhelmed to be able to respond to everything, making it seem as if some of those arguments, actually have merit. but I will try to respond to some of them. And one, one of them is that, you know, in, in Quebec, you talked about, mine, miners using, excesses The amount of, renewable fixing effi- efficiencies in the market, Quebec actually imposed a hard quota on the amount of energy that miners are allowed to use simply because, you know, we're, you know, this is where you enter the, the complicated world of grid management, and what they are dealing with is that, okay, they have an access for a big part of the year, but then at the same time during the summer months, winter months, they also have peak, demand, huh? So they need to- Serve, demand peaks during certain times of the year, and they need to keep their access available during for, serving those, periods. And the thing with Bitcoin miners is that, well, they typically like to consume power, on a twenty-four seven basis. I mean, you know, even in, in China before where a lot of these miners were- So now it's just, just" }, { "speaker": "guest_2", "time": "01:09:53", "start": 4193.36, "text": "one, one quick point on that. One, I mean, Hydro Quebec has excess capacity basically all year round, and two, every single miner in Quebec operates on a curtailment program with Hydro Quebec, and so during those coldest hours of the year, where that power is in, you know, shorter supply, we all turn off, we all turn off. Every single one of the Bitcoin miners in Quebec turns off almost simultaneously and restores that power to the grid. And so, you know, really that's not even a shortage of supply, it's more a accounting- Deal, where a lot of different municipal utilities are working with the provincial utility and they're trying to manage their peak load with the provincial utility. This is more about managing the accounting and, and the economics of it as opposed to a shortage of supply. And then, is one more quick point on that, the, the moratorium that's in place, the three hundred megawatts, that came into place because there was eighteen gigawatts worth of power applications that went into Quebec in twenty seventeen and twenty eighteen, eighteen gigawatts, at the time the industry was consuming four and a half gigawatts. Right? And so, you know, this was a-- this was absurd, the amount of applications that went in there, and of course, it created a, a huge amount of fear. But, you know, if they took any chan-- any time to look at the numbers and say, \"Hey, the network industry is consuming four and a half gigawatts, we've received applications from a bunch of unknown Chinese people for eighteen gigawatts, these aren't real.\"" }, { "speaker": "bitcoin_mining_energy_debate", "time": "01:11:15", "start": 4275.32, "text": "Yeah, okay, and that was actually when, I was in contact with one of your colleagues to, help show" }, { "speaker": "bitcoin_mining_energy_debate", "time": "01:11:25", "start": 4285.1, "text": "realistic. but, you know, I, I think that now this amount is really capped and, yeah, we, we really need to look at, okay, how, how does this look for the whole network? And You know, if you look at, for example, the US, New York Times recently did it in a separate analysis, they looked at, okay, what's going on in the United States with the miners using energy over there? They did a marginal analysis or a consequential analysis, and what that does is it looks at, okay, what is the real carbon impact of adding more power demand on a grid, are more renewables being used or is it more fossil fuels? At least in that analysis that they did, they found that on most US grids, renewables are getting prioritized and the ex-- and the additional power demand coming from crypto miners is mostly being served from fossil fuels, and they actually concluded that the amount of fossil fuels going into the network over there was going to be reaching, around ninety percent, yeah, which is very much higher than what we conclu-- or what I, in Cambridge previously concluded because we tend to look at grid averages. And we look at the grid and we, we take a grid average to do our carbon emission calculation, huh? And taking a grid average isn't accounting for all these complicated, dynamics. So in some cases, it wor-- may work out better, but at least the general trend seems to be that it works out, worse. Now, if, if you happen to be one of the ones, that are exclusively using, hydropower throughout the year, then, well, that's, that's great, huh? Of the network that doesn't apply. But then here you have other things to consider, you mentioned lifespan of devices, there's electronic waste, lifespan, increasing lifespan of device as well. You could say that's a good thing, 'cause then you have less electronic waste. At the same time, it means that you keep inefficient machines operating for longer, so then you also end up with higher energy consumption." }, { "speaker": "bitcoin_mining_energy_debate", "time": "01:13:30", "start": 4410.1, "text": "and about working together, to improve the amount of renewable generation on the grid. Well, renewable energy projects are typically long-term projects, and ironically, you know, I was on a stage with BitFarms in twenty nineteen at the Consensus conference, and we were talking about it, and I think I literally said at the time that Bitcoin miners make terrible partners to build out your, energy infrastructure for the very simple reason that Bitcoin miners are very dependent on an extremely volatile- asset. So if the pr-- the asset happens to go south tomorrow, then your business partner is going to be gone. And, well, it's not just me making that note, it's actually been repeated, I think, last year by Fitch Ratings, who also warned, utility companies to be very careful about interacting with the mining industry as a whole, because there is a financial risk involved for them if their business partner suddenly disappears overnight. That's a non-zero, risk if you're dealing with Bitcoin. I mean, I, what, what I do in my models, what I can do is I can predict somewhat how the energy consumption is going to evolve given a certain Bitcoin price, but I don't attempt to predict the price, and I don't think anyone can predict the Bitcoin price. So this is a, if you are going to, if you're a utility company, you're going to be involving yourself with Bitcoin miners, you're taking on that risk as well, you're exposing yourself to a highly volatile asset which might not be a good idea, but that's, also a completely separate discussion. Now, I reserved an hour for this, this talk, and we already passed that by twenty minutes, so I don't know if we need to follow up on this on a separate podcast, but I kind of need to leave." }, { "speaker": "stephan", "time": "01:15:10", "start": 4510.83, "text": "Okay. Alright. Alright. Well, I guess we'll, we'll finish it there. Thank you both for joining. thank you, Alex, for being, gracious with your time and, accepting the criticisms from, those of us in the Bitcoin industry, and thank you, Ben, for your time as well. we'll leave it Agreements, on Twitter or on another podcast some other time. So thank you both. If" }, { "speaker": "bitcoin_mining_energy_debate", "time": "01:15:31", "start": 4531.5, "text": "there's high demand, I'm always willing to join again. Happy to do a second one. Right." }, { "speaker": "stephan", "time": "01:15:36", "start": 4536.25, "text": "Thanks guys, and I will put the links in the show notes for listeners. Thank you. Okay. Thank you. Thank" }, { "speaker": "guest_2", "time": "01:15:39", "start": 4539.68, "text": "you." }, { "speaker": "stephan", "time": "01:15:40", "start": 4540.58, "text": "So I hope you enjoyed that, and let me know what you think, whether there should be another follow-up debate episode, and of course, make sure to share this one out there so people can hear different views on Bitcoin mining energy. Thanks for listening, and I'll see you in the citadels." } ] }