And currently there are sites that historically could have been for bitcoin mining, but because of the parameters of where they’re located now, they could be interesting for traditional data centers and AI load as well. So you have more competition and you have technology is coming as well. And so we’ve been hyper focused on that energy infrastructure piece, both from a greenfield perspective and creating new relationships and further deepening the existing relations that we have with energy generators and energy partners. And I mentioned in my script earlier that we have over 1000 exclusivity energy capacity, right? Because we believe building, getting energy capacity, building that capacity is a lot longer of a lead time than buying machines and having machines delivered and plugging them in.
And so as we look at the markets, we are very thoughtful on how we deploy. But if you look at the history of us, Bitcoin Corp. In 2021, we were not behind in how we bought machines, but we were very thoughtful. We didn’t buy machines for over dollar 35 a terrahash, because when we looked at press releases in the market, people were buying for $50, $60, $80 a terrahash. And in creating unique relations with the manufacturers and how you structure those deals, we’re able to get access to supply, but also at a low cost to be able to drive our return on investment higher. And so I think for us, the question is not if we’re going to grow and how long we’re going to wait, but what is the best entry point to deploy that capital to have a good return?
And I think the confidence that I hope people see is in the managed services business, colocation business. In our recent production report, we shared that we’re managing 27 exahash today, right? So if we only had our self mining fleet today and we said, hey, we’re roughly around seven x, we’re going to go and add 20 x a hash. The questions are, can you do it? Do you have the operational ability? Can you scale? But I think those questions don’t really need to be asked or have been answered already, which is we are already doing it and we have the capabilities, we have the softwares, we have the sops to do it. And so the question is, when do we deploy that capital and how do we generate the highest yield and the highest return? But we’re very sensitive to your point.
Josh and I being caught flat footed and being in a bull run where we can’t get access to machines or pricing is elevated. And so we’re actively monitoring the markets and having discussions on a daily basis to have a strong pulse on that.
Josh Siegler is: Yeah, understood. Honestly, that outlook makes a lot of sense. So appreciate the answer there. And then tying back to what you were talking about in terms of your diversification and really looking at expanding even more beyond self mining in the future, I’m curious if you can talk a little bit more about the financing options. I think everyone on this call understands that financing is a very, very difficult task in purely self mining right now. But as you expand to these other verticals, do you expect that financing to really open up?
Asher Genoot: Yeah. And that’s the key. Right? I had this amazing conversation about a year and a half ago with one of the largest renewable company kind of CEO’s that just retired and learned about his journey and his path. And I was like, what did you do that allowed you to grow your market cap by like $100 billion? What was it? We didn’t see ourselves as an energy company. We saw ourselves as a financing company because we’re such a CapEx intensive business. And that really stuck with me. As we look at our business today, it’s no surprise that our business and the business of our peers is a heavy CapEx business. And how do you grow that business with the lowest cost of capital to drive the most amount of shareholder value?
If you look at 2021, there was a lot of debt in the markets. I think people learned that probably structuring that debt more creatively would have been more beneficial because you had a lot of distress scenarios in 2022. And we were on the receiving side of that. And we’re able to grow a lot through that bear cycle. And then recently, most of the growth capital is coming from the equity market and a lot of the ATM’s that are available in the market. And so when we look at kind of the opportunities today, we see a couple of different buckets of assets and investor opportunities. So if you look at mining, you have the land in the substation, which is pretty kind of cookie cutter. There’s a lot of financing there at low cost. Then you have the data center, which is higher risk than a traditional data center financing, because the use cases for bitcoin mining.
Not going to use that existing data center for the bitcoin mining data center. For a traditional data center. To build a bitcoin mining data center. I shared, we’re going to be under $275,000 a megawatt to build a traditional data center for AI, about eight to $10 million a megawatt. So very different infrastructure. And then you have the ASIC chips themselves, which is high volatility, high risk. Now, on the other side of traditional data centers, you have the data center, and then you also have the GPU’s and the Nvidia GPU’s. We look at all of those as different asset classes, that you have different investors that want exposure to that asset and the return and volatility of that asset. And by bifurcating them and looking at more project level financing, we believe we’ll be able to expand the pool of people that want exposure and that are willing to invest and also decrease the cost of that capital because there’s different risks on each component of that business.
And so that’s how we’re thinking about financing, and that’s where a lot of our conversations have been understood.
Josh Siegler is: Appreciate the color. Asher, thanks for taking my question, and good luck in the future.
Operator: Thank you. Our next question comes from John Todaro with Needham and Company. Your line is open.
John Todaro: Great. Thanks for taking that question, and, yeah, congrats. Thank you for picking up coverage on us. Yeah, yeah, of course. So, two questions for me. First, on the GPU cluster. So I guess we got a little bit more color here, but as we think about this kind of, what are the next steps? So does a site still need to be built? Is there going to be more of a colocation strategy where you use a third party site? So, yeah, I guess just as we think about that, is this something that, you know, there’s revenue in calendar ’24, or is this more like a ’25 story?
Asher Genoot: Yeah. So we’ll be announcing the full plan regarding that first order within the next quarter, and it would be something in this calendar year that we’re looking to execute and bring on. We see the data center business and the GPU business as two opportunities that will converge down the line. And what I mean by that is today, as I mentioned, the current HPC business, it needs to go through some restructuring and investments in order to be equipped to handle AI machines. But however, we don’t want to wait for that in order to get our entry point into the GPU market and get exposure to that market. And we think of that similarly to kind of, if I use analogy to mining, owning and building the data center is different than just buying the chips and kind of hosting and co locating, especially if you have a good customer and kind of partnership in place in order to scale and have good cash flows.
And the markets today are long shifts in short supply. And I think the demand for AI chips is only increasing. We’re just touching kind of the tip of the iceberg. And so we’re excited to share updates on that in the near term and look to be launching that business unit in the near term as well.
John Todaro: Okay, great. And then one on the mining side, I believe you’ve got in south mining, I think it was mentioned around 4.5 cents a kilowatt hour, but either hosting or managed services was a bit higher at $0.06. As you think about the business, where there is the higher power cost, how are those contracts structured? Do you expect some of that hash to come offline after the having? How profitable would that be, say bitcoin? We just say it stays at 70k?
Asher Genoot: Yeah, no, appreciate that question. So let me give a quick background on why we started hosting machines and go to what the strategy is today. So the only hosted machine that we had in kind of the recent announcements that we discussed was machines hosted at the Charlie and Delta facility, which is Bacarney and Granbury facilities. And the reason we had hosted machines there is during the compute north bankruptcy, when generate took over and hired us as the managed services operative to run the facilities. At that moment in time, we believe that our capital was better spent investing in distress opportunities than just going and building infrastructure. And we looked at the return profiles. We created a good deal where it was based on a kind of a hash price, based on hash price that would determine our hosting rate.
And it also created an alignment would generate that we had our own skin in the game and that we cared deeply about uptime because we had customers and ourselves there. And it also helped during the transition where other customers saw, hey, you guys have your own machines there. You’re putting kind of your money where your mouth is. And so we can have comfort as well. And so that’s, that was kind of the thesis of why we had hosted at that moment in time, but where we are now in the market. So obviously those sites have been sold off and there’s a new owner and there’s a new new operator. Those facilities that are looking to operate on their own. And so for us, even if we had hosted miners, controlling the operations and manage them is critical to kind of how we think about kind of that growth.
And so from that perspective, this month we already started moving machines from the hosted facilities to vertically integrate them into our self owned facilities. And so you’ll see some downtime across the end of this month and next month in those machines being offline and being relocated to our facilities. And I mentioned on the Salt Creek facility, we announced kind of the groundbreaking of that last month in February. We expect that site to be fully online in April, which I said is under three months. Or from those of you following kind of build out timelines, I think from Greenfield to fully online and hashing in less than three months is a pretty short timeline. And by doing so, we’ll be driving all of our machines to self managed facilities and driving those costs down, especially with the having coming up in the next couple weeks.
John Todaro: Got it. Thank you for that. I appreciate the answers and good luck.
Operator: Thank you. Our next question comes from Bill Papanastasiouj with Stifel. Your line is open.
Bill Papanastasiou j: Hi guys, thanks for taking my questions. Just a quick one here. Hoping you could provide an update on market opportunities for managed services in a post having environment. Curious to hear what growth could look like on iconic digital. Obviously a number of miners have been acquiring a significant amount of equipment and many of which don’t have the infrastructure to house these rigs. And according to your earlier comments, we will also see newer generation equipment down the road. Any color you can provide there.
Asher Genoot: Thanks. Yeah, so when we look at kind of our two non self mining business units, we have the colocation business and we have the managed services business. So the managed services business is a third party, hiring us to either build and or manage all their facilities. And that’s not just the operations of the construction, but that’s turnkey solutions, that’s managing the energy, the curtailment, that’s doing the financials, putting audited financials and books together, really running a turnkey suite. And the thesis behind that business is when you look at a mature market, usually you can hire an operator to come in and operate. And so if you want exposure to a wind farm or an oil and gas well, you can hire an operator to come in and operate.
