Hello, I'm John Furrier, host of theCUBE here in our NYSE CUBE Studio as part of our Data Center of the Future, AI Factory series in the NYSE Wired Community and program. Big popular series around the future of AI and what's going to power it. And we have one of the major players here to talk about the power and the energy. Asher is the CEO of Hut 8 and the executive chairman of American Bitcoin. Asher, thank you for spending the time with us today. Bill Tai certainly laid down a lot of commentary in previous CUBE interviews, so thanks for coming on.
Yeah, definitely. Thanks for having me on. Excited for the discussion today.
So right out of the gate, I did see some news today. American Bitcoin announced acquiring more Bitcoin and increasing the strategic reserves. Quick comment, because I want to get into the energy conversation with you guys around this, but this is a signal of more action. The market is growing, there's more business model innovation. Just quick comment on the news.
Definitely. We continue to grow all of the verticals we have. So Hut 8 is the energy infrastructure company. Hut 8 is a long energy company. We have energy assets, we have data center assets, power plants, data centers for AI, data centers built for Bitcoin. Then we have two businesses. One is American Bitcoin, which owns the servers for Bitcoin, and then Highrise that owns the GPUs for AI and cloud. And so American Bitcoin is continuing to grow. They had a big announcement today. They're continuing to scale and grow. And so this is our overall ecosystem, and we're lucky that we see growth on all elements of those different pillars of the ecosystem today.
We've been following their journey for a long, long time since Bill started and the team put Hut 8 together. What's interesting is the creation of American Bitcoin and then you said Highrise AI, you have the mining side and now the data center side. You're essentially scaling energy infrastructure under management, and talk about that power envelope. What does the headroom look like? Can you share the dynamics? Because you got the mining. That's key. You got the scalable infrastructure on the GPU side. What does that do from a scale perspective? What does that mean in terms of numbers, gigawatt to what?
So today, Hut 8 has about a gigawatt under management that we've commercialized, we've built and we're receiving revenues and profit from. We have over a gigawatt in the development pipeline of sites where we've either had the option on the land, own the land, the power agreements, and we're developing those projects today. And then we have a couple of gigawatts under exclusivity or diligence in which we are continuing to drive into the development phase. And so we have gigawatts in our overall kind of megawatts that we have. And for us at Hut 8, it's a pretty simple perspective. As technology continues to evolve, its consumption of power will increase. That doesn't matter if it's data centers for AI, data centers for Bitcoin, advanced manufacturing. We have the conviction and belief that power is going to become more scarce, that power is going to become more valuable. And so how do we sit at that intersection at Hut 8 in energy and technology and build infrastructure that supports these use cases? That's why you see us heavily entrenched in the AI theme and the demand that we see there, but also continuing to scale and build on Bitcoin and securitizing the future of finance as well. And so at Hut 8, you see us not being technology-focused, more technology-agnostic and more infrastructure and energy-focused.
You guys have a managed power solution, I call it that. It's my word, probably a word that means it. The power is, everything's bounded by the power. Is there a difference between the energy side for say true AI-ready GPUs, power density interconnects versus standard mining capacity? Is there a difference between the two, and how would you discuss that?
So power is fungible, whether you pull a megawatt or gigawatt from the grid and you use it for AI use cases or for Bitcoin use cases. There's a difference in the build itself. So with AI, you build more redundancy into the data center itself and you have a higher PUE. And so that means that you're a little less efficient because you have so much more cooling and infrastructure and redundancy supporting the actual chips so they have a more stable and predictable workflow. And so when you think about the overall data center, something that we're excited that we're continuing to build and innovate on is how do you build infrastructure that may be able to support different types of chips? At the end of the day, people use AI and people use Bitcoin as the terms, but really it's what actual servers are using in the compute. AI today is using GPUs primarily from Nvidia. And then on the Bitcoin side, it's primarily ASICs, but you will start having ASICs power AI compute as well. And so I think when we think about, it's what type of compute are we using, what type of servers are we using, and what is the actual load dynamics of those servers? And so we're very excited because we see continued demand and growth and energy and infrastructure assets, and we're trying to build and support all of that demand today.
I think you guys did a really good job. I love that, I won't say pivot, because it's more of an extension. You didn't really change directions, you just kind of created a scalable model for energy, and energy infrastructure, I guess is the category. I was talking to a friend about this this weekend and I told him I was going to be interviewing you. He says, 'Oh, ask him a question.' So I'm going to ask you a question from a friend. This is classic social media. Hey, asking for a friend. Here's the question. What's your crisp North star or KPI that you're working through this next year? Is it watts contracted, GPUs deployed, EBITDA mix, something else? What's the focus? How would you answer that question? Because you're bringing up the economics between AI and Bitcoin mining, which you just laid out. As you look forward with this market, what's that KPI, that North star that you look at?
So I'll focus on Hut 8, because American Bitcoin and Highrise have their different goals and KPIs, which is why we separated the companies into its own leadership and management teams as well. For Hut 8, the long-term goal and vision is how do we increase the amount of megawatts in our overall footprint and access to the power that we have over term? Because we believe that the value of that power will continue to increase. In the short term, looking at the next one plus years, it's how do we lock in those megawatts, but also contract those megawatts for different use cases. And today, for example, the return on AI use cases is very high. So continuing to lock in power for AI use cases as well. The beauty about American Bitcoin being a customer for Hut 8 is that we're able to lock up sites really, really quickly, be our own customer through American Bitcoin and its demand, and then be able to then focus on looking at what is the value of those megawatts over the long term. And so for Hut 8, it's continuing to lock in megawatts for American Bitcoin, continuing to lock in megawatts for AI use cases, and being able to commercialize those in the short term. And in the long term, just continue to expand the overall footprint and access the power that we have as a business.
Okay, thank you so much. On the footprint and the deployment, the site deployments, what's the issue there from a balancing of resource speed? Do you optimize for a certain equation? You get speed, cost, efficiency, obviously key factors. Is there a formula or kind of algorithm that you lean into optimize around?
Look, today, speed is paramount in this industry. People need access to these data centers to distribute as quickly as possible, and so we're looking at providing different solutions and innovative designs that offer that with great partners. And so speed is paramount, speed to power, speed to energization, speed to compute. Over term, as we look at overall moat as a company, I believe speed is important, I believe cost is important and I believe quality is important. And usually people say you have to choose two out of the three, but I push our team to try to optimize it all.
