So at Hut 8, we deal with less of the volatility of Bitcoin going up and down, and we're focused on building infrastructure, but we're supporting multiple technologies today rather than just Bitcoin or just AI. We believe that the company will continue to support all of the next-generation technologies that we think will change the world.
What's going on, guys? Today we got a great conversation with Asher Genoot. He is the CEO of Hut 8, and he is one of the leading operators when it comes to the intersection of Bitcoin mining and AI HPC. In this conversation, he goes through the excruciating detail of what these companies are, how they operate, how they differentiate, and why investors should consider putting them into their portfolio. This conversation covers a lot, and I think you're going to find it very, very valuable. There's lots of information in here that you're not going to find anywhere else. Here's my conversation with Asher.
Asher, you guys were one of the biggest Bitcoin miners. You've now converted to AI HPC. You've got a very unique story, I think, in how you've converted this business. It's now a 10-plus-billion-dollar market cap. Talk a little bit about the transition the industry is going through and why investors are flocking to this AI HPC trade.
The demand for AI is really just at the starting point right now. Adoption, I think, is at the very, very early onset. You see companies like Anthropic that have increased their revenue month over month over month, adding over $10 billion of ARR month over month, which is the most insane growth that you've seen in companies. I mean, they've just raised at 100, then 300, then almost a trillion dollars, and you're seeing all this news in the market, which is super exciting. I think building infrastructure to support those technologies and those use cases is selling the picks and shovels to the ecosystem. And for Hut 8, our mission is to create an energy infrastructure platform that supports the technologies of tomorrow, and that's building digital infrastructure. And so when we started the business, we supported Bitcoin compute, and that was a technology that we believed could revolutionize the financial system, and we created the security behind that network for transactions to happen. As most of our peers in the ecosystem said, you know what, we have these megawatts, we're using them for Bitcoin, let's convert them to AI because we can make more money off of that. We took a very different approach. We said, we've built this platform and become one of the largest people in the world securitizing the Bitcoin network. Let's use the skill set of finding power, building infrastructure at scale, and managing power and build for the AI use case. And so I think very uniquely, the last two deals that we've announced, some $17 billion in contract value, those are net new projects that we built from the ground up and developed rather than we converted over from existing Bitcoin facilities. We still manage about 700 megawatts of Bitcoin infrastructure, and we spun out that business to a company called American Bitcoin, which is publicly traded as well, and they're a customer now of Hut 8 infrastructure. So at Hut 8, we deal with less of the volatility of Bitcoin going up and down, and we're focused on building infrastructure, but we're supporting multiple technologies today rather than just Bitcoin or just AI. We believe that the company will continue to support all the next-generation technologies that we think will change the world.
Now, when you think about building these data centers, you guys are going and finding land, you're powering these things, you're going and striking these business development deals. One of the things that has happened in society is there's a lot of people who do not want data centers. They're worried about water. They're worried about electricity prices. They're worried about the aesthetics of the data center in their local community. Talk a little bit about this balance between, you know, kind of the everyday American in a local community and their views towards artificial intelligence and data centers, and then the belief that maybe from the, you know, kind of Wall Street crowd, the Silicon Valley crowd of we don't have enough data centers, we need more compute. And so how are you guys navigating this balance?
The US needs to build more infrastructure if it wants to be the leader in the technology revolution around the world, and that's critical. I think about kind of where the US stands today, and we have been and can continue to be a leader in the technological revolution and being at the forefront. I think, but right now, even with us saying let's go build as much infrastructure as possible, we're still competing neck and neck with countries like China, who are able to build energy infrastructure, we're able to build data center infrastructure, frankly, faster and cheaper than we are today. And so I think we need a common unity to say we should be behind the infrastructure that supports the growth of these technologies, continue to make sure that our LLM models are at the forefront. We had a head start because we have Nvidia, which is a US company, being able to support us with the best generation chips. But as they start building their own chips and as they start competing, infrastructure is going to be the key bottleneck, and you're seeing that today: power infrastructure, data center infrastructure. And so I think there's this general fear of what does AI mean for me across the US today, and whether that fear be around what's going to happen with my job, whether that fear be around what is this infrastructure and all this investment in my neighborhood going to do. And I think there's a lot of these kind of false fallacies that have percolated and gotten people scared, right? And so some of them you mentioned: one, data centers are going to make my energy prices go up. That's not true. We work with the leading energy utilities across the country, and the reality is every time a data center is built, especially today at large scale, we come in, we pay for what is called a CIAC, which is we pay for all the infrastructure and system upgrades to be able to transport that energy, and that goes to the overall system. So all the people who are consuming energy get to benefit from those system upgrades as well. And then the energy supply itself, we pay and commit to the consumption of that power as well. And so in a lot of these utilities, and you're seeing more of them start to post data about this, but energy prices actually come down with data centers coming online because you increase the utilization and you upgrade the infrastructure. But like there are some areas where you might have a really bad kind of summer storm and energy prices go up, or you have other criteria that are happening and people blame it on data centers, right? But the real kind of facts have seen the opposite, which is energy prices have gone down in a lot of areas where they're structured correctly. The second is water usage. There's two ways to cool these chips. One is the analogy I like to use is if you're walking around at Disneyland with your family and you're kind of spraying a fan and like misting water, and you're basically using the cooling temperature of the cool water to cool down your chips. Alternatively, you have a refrigerator at home, and that refrigerator doesn't use active water to cool it down, right? It's a closed-loop system that is able to cool down the chips. And so we use that closed-loop system where we don't actually use any water on an ongoing basis other than the restrooms, the sinks, and things of that nature. It's like any other office building, and we're much smaller consumption than a traditional office building because everything is in that closed-loop system. Historically, not everyone did that because one, it's more expensive from an infrastructure perspective; two is that you have to invest into kind of higher PUE, so you consume kind of more electricity prices and so forth. And so those are some of the reasons why historically you've had different types of use cases on the water side. But the water issue can be solved where you're not actually using water in the community, and you're using much less water than what people have kind of day-to-day when they drive by any office building, car wash, and anything of that nature.
And so I think those are like two big ones. The third one is noise and how you design and how you build these data centers. If you look at Ashburn, Virginia, there are literally data centers right next to homes, and so these things can be built in a much more quiet manner. And lastly, we're actually spending a good amount of time with this, as historically a lot of these things look like kind of large infrastructure warehouse-like buildings, but I think there's an opportunity to invest a bit more money and make these things look beautiful and have them drive by and feel more like museums than they do kind of warehouses. And so we're thinking about all of those things, and as we go into new markets, we're kind of talking through those. But I think as a nation, we've got to figure out how to get people excited that we could be at the leading forefront of this technology rather than nervous and scared, because if that is really what ends up really driving our decisions in this country, I think it's going to hinder us from leading in this AI revolution and allow other countries like China to be able to surpass us. And that's why I think this sense of coming together in this nationalistic pride of let's build the infrastructure that supports this compute is critically important, because if it doesn't get built in the US, it will get built elsewhere, and we don't want that compute lying in other countries.
Now I'm going to throw out a couple of things that I've heard various players doing, and I want you to react to agree, disagree, you know, and why. The first is I've heard some people pitch, well, we should do colocation. We should do some like high-performance manufacturing and collocate it with data centers. That changes the economics of the actual site, and it allows us not only to get the compute online, but then it also allows us to do high-performance manufacturing. How do you guys think about that?
I don't really fully follow kind of connecting those two. I think energy campuses are going to be a bigger and bigger thing, which is you bring generation. So you not only bring the load and the consumption, you actually bring net new generation as well. And there might be other use cases, right? Because like we have a 500-megawatt data center, but the IT load is 352, and the rest is kind of the cooling infrastructure that doesn't always get used. And so you can have kind of collocated infrastructure that's able to consume some of that power that's not being used. So I think there's a lot of creativity of how to build these energy campuses, but it depends on what do these loads need, what are the different requirements, and how do you create kind of these utilized systems. I think the next phase historically for data centers was we go, we ask a utility, hey, can we get power? Obviously, we don't have much power, and so we need to go build new power. And so now a lot of infrastructure providers are thinking about how do we bring power with us, so we bring power and we bring load, and then we can help add capacity into the market while consuming as well, and I think you'll see a lot more of that.
Now that collocation or like the other offtakes of the energy, what would they be doing? Is it other types of data consumption, or is there something else?
So as we think about it from a Hut 8 perspective, today we have two large use cases, right? We have use cases for ASIC that kind of power the Bitcoin network and security. And we're also big believers. If Bitcoin is going to be a currency around the world that people trust, we want that compute or a large portion of that compute to be in the US rather than in countries that, whether it be kind of Iran, Russia, Ukraine, China, like we want it to be in countries that are more on in the West. And so maintaining compute in the US we think is kind of critical to national security. Separately is from an AI compute perspective. So these two loads are very different. AI compute is more bursty, more based on workloads, and kind of a traditional data center they consume, and it's less price sensitive, where Bitcoin you trade around price volatility. And so these are interesting loads that can actually work together because one doesn't have to be online all the time and one does. And so other use cases we're seeing in the US today, you see advanced manufacturing, you see these kind of onshoring, you know, these chip manufacturing companies, fab manufacturing companies that are all large-scale energy campuses in the hundreds of megawatts, if not gigawatts as well. And you'll see more of that as we step into the world of robotics and advanced manufacturing.
Now, another thing I've heard is there's obviously the mega sites, you know, hundreds of megawatts, but then I've also seen people talk about, well, actually closer to urban areas, we're going to need 50 to 100-megawatt sites. How do you think through the mega sites versus maybe the smaller sites that are closer to the density?
I'll talk about it from a power then kind of a latency perspective. Historically, there was this belief that like you could only do training farther out and you have to be much closer to population density to have better latency to run workloads. If folks use ChatGPT or use Claude or use Grok, you realize that sometimes you ask it a question and it goes and it thinks for some time and then comes back. So that really quick latency is less relevant compared to a world where you're browsing your emails or social media or whatever it may be. And so we've seen that these large-scale campuses, actually the ability to build kind of these inference loads not just in population-dense areas and first kind of tier markets is a reality today. At the same time, you have a power constraint, which is how much power at scale can you get, and that's becoming harder and harder. And so I think you're going to see a lot more smaller campuses being deployed because it's easier to be able to say, all right, let's consume 5, 15, 50 megawatts rather than go build a 1,000-megawatt campus, and so you'll see both, and you'll see people building infrastructure across both sets.
Now another thing that I think people are trying to figure out is at the same time we're trying to get power, we're trying to get the data centers built, and there's a bunch of kind of constraints or obstacles in just those two things alone. Also, there's a lot of movement in like what is going inside the data center. The hardware itself is being rapidly innovated on, and these large language models are now putting a lot of money, and they're trying to figure out like who can get the advantage on the actual metal itself. And so are you guys participating in that, or are you just trying to deliver, you know, powered shells or just the power itself and leave that for somebody else?
We are building full turnkey solutions. So we're not just building the powered shell, we're building everything that goes inside the data center as well. And as we think about the long-term kind of competitive edge in the data center market, what matters most is your cost per token, which is like your unit of compute, and your cost per token on a fully depreciated and amortized basis. And so that means you don't really care if the costs are in capex or in opex. You care about the all-in cost of that infrastructure. And the three pieces that really drive those costs are one, the cost of energy; two, the cost of the data center; and three, the cost of the chips and the technology and the orchestration that runs it all to run it the most efficiently to get the most compute out of that infrastructure stack. And so at Hut 8, we're proud because we're not just a data center business. We've owned power plants in our history. We own four power plants. We have behind-the-meter assets, front-of-the-meter assets. And so I think understanding that full value chain on the compute side, we have another company called Highrise AI that we incubated, and that's like a neocloud business, CoreWeave. And so we understand this full stack from beginning to end and can think about how to drive efficiencies across a full chain to drive the thing that matters most, which is your cost per token.
Today's episode is brought to you by Simple Mining. Bitcoin mining has a reputation for being complicated, risky, and hard to evaluate as a real investment. If you're considering mining in 2026, what actually matters isn't headline profitability, it's uptime, repairs, and whether the operation is run like a real business. That's why I've been using Simple Mining. They're based in Cedar Falls, Iowa, and they run a white-glove hosting operation where you own your miners, you choose your own pool, and you have Bitcoin sent directly to your wallet. They were featured on the Inc. 5000 list as the fastest-growing company in Iowa with over 40,000 machines under management. What stands out to me is execution. They have the number one rated ASIC repair center, and for the first 12 months, repairs are included. If mining margins get tight, you can pause with no penalties. And if you want to resize or upgrade your fleet, there's a marketplace to resell equipment instead of being stuck. To help people think it through whether mining actually makes sense right now, they put together a short resource called the 2026 Bitcoin Mining Blueprint. It walks through the five mistakes investors make when allocating to mining, and they also explain how to avoid them before deploying capital. If it sounds interesting to you, you can get it for free at simplemining.io/pom. That's simplemining.io/pom. Go check it out today and see if you should get into the mining game.
Today's episode is brought to you by Fountain Life. Your body is incredibly good at hiding disease unless you look.
True wealth isn't measured in possessions. It's measured in the time you gain, the energy you feel, and the life you create. It may be the smartest investment of your time you'll ever make.
Life is so very, very precious, and I want each of you to be truly awakened to your aliveness and live longer without limits.
Your number one job is to educate yourself, and Fountain wants to help educate you.
Now you guys are striking very large deals. Last one was over $9 billion of revenue. I think that's part of that contract. What's that process like? So you guys know, okay, we got the land. We're going to go power with this thing, but we need to have a partner. We need to have somebody who's actually going to pay us for this. Do you just call up a bunch of companies and say, hey, I got land, I got power, let's have a conversation? You know, just walk through maybe how do you get in touch with these people and what are those conversations like right now?
We've built really deep relationships with some of the largest consumers of compute in the world. That's some of the largest investment-grade hyperscalers. That's the AI labs like the Anthropic, OpenAI, xAI's. That's also the fast-growing companies that are AI companies that are growing as well and creating use cases across the board that are consuming hundreds of millions, if not billions, of dollars of compute. And so you have a massively growing market. And then enterprise is slowly starting to percolate right now as well. And so for us, we're excited because we've approached these relationships as partnerships. And the way that we've approached it is not we have this piece of inventory, do you want it or not? But what is your long-term growth map, and can we be infrastructure partners to you all? And that's how we've really approached and been building our deals that come together. And that's really kind of the mindset that we're taking as we continue to grow and scale.
