The TRUTH About Bitcoin Mining & AI Data Centers

[0:00] So at HUD, we deal with less of the
[0:02] volatility of Bitcoin going up and down
[0:04] and we're focused on building
[0:05] infrastructure, but we're supporting
[0:08] multiple technologies today rather than
[0:10] just Bitcoin or just AI. We believe that
[0:13] the company will continue to support all
[0:15] of the next generation technologies that
[0:16] we think will change the world. What's
[0:18] going on guys? Today we got a great
[0:19] conversation with Asher Gnut. He is the
[0:21] CEO of HUD and he is one of the leading
[0:24] operators when it comes to the
[0:25] intersection of Bitcoin mining and AI
[0:27] HPC. In this conversation, he goes
[0:29] through the excruciating detail of what
[0:31] these companies are, how they operate,
[0:33] how they differentiate, and why
[0:34] investors should consider putting them
[0:36] into their portfolio. This conversation
[0:38] covers a lot and I think you're going to
[0:39] find it very, very valuable. There's
[0:40] lots of information in here that you're
[0:42] not going to find anywhere else. Here's
[0:43] my conversation with Asher.
[0:45] Asher, you guys were one of the biggest
[0:46] Bitcoin miners. You've now converted to
[0:49] AI HPC. You've got a very unique story,
[0:51] I think, in how you've converted this
[0:52] business. It's now 10 plus billion
[0:54] dollar market cap. talk a little bit
[0:55] about the transition the industry is
[0:57] going through and why investors are
[0:58] flocking to this AI HPC trade.
[1:02] >> The demand for AI is really just at the
[1:05] starting point right now. Adoption I
[1:07] think is at the very very early onsense.
[1:09] You see the companies like Enthropic
[1:11] that have increased their revenue month
[1:13] over month over month adding over $10
[1:16] billion of ARR month over month which is
[1:19] the most insane growth that you've seen
[1:21] companies. I mean they've just raised at
[1:23] 100 then 300 then almost a trillion
[1:25] dollars and you're seeing all this news
[1:26] in the market which is super exciting. I
[1:28] think building infrastructure to support
[1:31] those technologies and those use cases
[1:33] is selling the picks and shovels to the
[1:36] ecosystem and for HUD our mission is to
[1:40] create an energy infrastructure platform
[1:42] that supports the technologies of
[1:43] tomorrow and that's building digital
[1:46] infrastructure and so when we started
[1:48] the business we supported Bitcoin
[1:50] compute and that was a technology that
[1:52] we believed could revolutionize the
[1:54] financial system and we created the
[1:56] security behind that network for
[1:58] transactions to happen. As most of our
[2:01] peers in the ecosystem said, you know
[2:03] what, we have these megawws. We're using
[2:05] them for Bitcoin. Let's convert them to
[2:07] AI because they're we can make more
[2:08] money off of that. We took a very
[2:10] different approach. We said, we've built
[2:12] this platform and become one of the
[2:14] largest people in the world securitizing
[2:16] the Bitcoin network. Let's use the skill
[2:18] set of finding power, building
[2:21] infrastructure at scale, and managing
[2:23] power and build for the AI use case. And
[2:26] so I think very uniquely the last two
[2:28] deals that we've announced, some $17
[2:30] billion in contract value, those are net
[2:32] new projects that we built from the
[2:34] ground up and developed rather than we
[2:36] converted over from existing Bitcoin
[2:39] facilities. We still manage about 700
[2:41] megawatts of Bitcoin infrastructure and
[2:43] we spun out that business to a company
[2:46] called American Bitcoin which is
[2:47] publicly traded as well and they're a
[2:49] customer now of HUD infrastructure. So
[2:51] at Hud, we deal with less of the
[2:53] volatility of Bitcoin going up and down
[2:56] and we're focused on building
[2:57] infrastructure, but we're supporting
[2:59] multiple technologies today rather than
[3:02] just Bitcoin or just AI. We believe that
[3:04] the company will continue to support all
[3:06] the next generation technologies that we
[3:08] think will change the world.
[3:10] >> Now when you think about building these
[3:11] data centers, you guys are going and
[3:13] finding land, you're powering these
[3:14] things, you're going and striking these
[3:16] business development deals. Um, one of
[3:18] the things that has happened in society
[3:19] is there's a lot of people who do not
[3:21] want data centers. They're worried about
[3:22] water. They're worried about electricity
[3:24] prices. They're worried about the
[3:25] aesthetics of the data center in their
[3:26] local community. Talk a little bit about
[3:28] this balance between, you know, kind of
[3:30] the everyday American in a local
[3:32] community and their views towards
[3:34] artificial intelligence and data centers
[3:36] and then the belief that maybe from the
[3:38] uh, you know, kind of uh, Wall Street
[3:39] crowd, the Silicon Valley crowd of we
[3:41] don't have enough data centers, we need
[3:43] more compute. And so how are you guys
[3:45] navigating this balance?
[3:48] >> The US needs to build more
[3:51] infrastructure if it wants to be the
[3:53] leader in the technology revolution
[3:56] around the world. And that's critical. I
[3:59] think about kind of where the US stands
[4:02] today and we have been and can continue
[4:05] to be a leader in the technological
[4:07] revolution and being at the forefront. I
[4:09] think but right now even with us saying
[4:12] let's go build as much infrastructure as
[4:14] possible we're still competing neck
[4:16] andneck with countries like China who
[4:18] are able to build energy infrastructure
[4:20] we're able to build data center
[4:21] infrastructure frankly faster and
[4:23] cheaper than we are today and so I think
[4:25] we need a common unity to say we should
[4:29] be behind the infrastructure that
[4:32] supports the growth of these
[4:33] technologies continue to make sure that
[4:35] our LLM models are at the forefront we
[4:37] had a head start because we have Nvidia
[4:39] which is a US company being able to
[4:41] support us with the best generation
[4:43] chips but as they start building their
[4:45] own chips and as they start competing
[4:47] infrastructure is going to be the key
[4:49] bottleneck and you're seeing that today
[4:51] power infrastructure data center
[4:52] infrastructure and so I think there's
[4:56] this general fear of what does AI mean
[4:58] for me as across the US today and
[5:01] whether that fear be around what what's
[5:03] going to happen with my job whether that
[5:05] fear be around what is this
[5:06] infrastructure and all this investments
[5:08] in my neighbor neighborhood going to do.
[5:09] And I think there's a lot of these kind
[5:10] of false fallacies that have like
[5:14] percolated and gotten people scared,
[5:15] right? And so some of them you
[5:16] mentioned, one, data centers are going
[5:19] to make my energy prices go up. That's
[5:21] not true. We work with the leading
[5:23] energy utilities across the country. And
[5:25] the reality is every time a data center
[5:27] is built, especially today at large
[5:29] scale, we come in, we pay for what is
[5:32] called a kayak, which is we pay for all
[5:33] the infrastructure and system upgrades
[5:35] to be able to trans transport that
[5:37] energy. And that goes to the overall
[5:38] system. So all the people who are
[5:40] consuming energy get to benefit from
[5:43] those system upgrades as well. And then
[5:45] the energy supply itself, we pay and
[5:47] commit to the consumption of that power
[5:49] as well. And so in a lot of these
[5:51] utilities, and you're seeing more of
[5:53] them start post data about this, but
[5:55] energy prices actually come down with
[5:58] data centers coming online because you
[5:59] increase the utilization and you upgrade
[6:01] the infrastructure. But like there are
[6:04] some areas where you might have a really
[6:05] bad uh kind of summer storm and energy
[6:08] prices go up or you have other criteria
[6:10] that are happening and people blame it
[6:12] on data centers, right? But the real
[6:14] kind of facts have seen the opposite
[6:15] which is energy prices have gone down in
[6:17] a lot of areas where they're structured
[6:19] correctly. The second is water usage.
[6:21] There's two ways to cool these chips.
[6:24] One is the analogy I like to use is if
[6:28] you're walking around at Disneyland with
[6:29] your family and you're kind of spraying
[6:32] a fan and like misting water and you're
[6:34] basically using the cooling temp
[6:36] temperature of the cool water to cool
[6:37] down your chips. Alternatively, you have
[6:40] a refrigerator at home and that
[6:42] refrigerator doesn't use active water to
[6:44] cool it down, right? It's a closed loop
[6:46] system that is able to cool down the
[6:48] chips. And so we use that closed loop
[6:50] system where we don't actually use any
[6:54] water on an ongoing basis other than the
[6:56] restrooms, the sinks, and things of that
[6:57] nature. It's like any other office
[6:59] building and we're much smaller
[7:00] consumption than a traditional office
[7:01] building because everything is in that
[7:03] closed loop system. Historically, not
[7:05] everyone did that because one, it's more
[7:08] expensive from an infrastructure
[7:09] perspective. two is that you have to
[7:14] invest um into kind of higher POE so you
[7:17] consume kind of uh more electricity
[7:20] prices and so forth and so those are
[7:22] some of the reasons why historically
[7:25] you've had different types of use cases
[7:26] on the water side but the water issue
[7:28] can be solved where you're not actually
[7:30] using water in the community um and
[7:32] you're using much less water than what
[7:34] people have kind of day-to-day when they
[7:36] drive by any office building car wash um
[7:39] and anything of that nature
[7:41] And so I think those are like two big
[7:42] ones. The third one is is is noise uh
[7:45] and how you design and how you build
[7:47] these data centers. If you look at
[7:48] Ashurn, Virginia, there are literally
[7:50] data centers right next to homes. And so
[7:51] these things can be built in a much more
[7:54] quiet manner. And lastly, we're actually
[7:56] spending a good amount of time with this
[7:57] is historically a lot of these things
[8:00] look like kind of large infrastructure
[8:02] warehouse like buildings, but I think
[8:04] there's an opportunity to invest a bit
[8:05] more money and make these things look
[8:07] beautiful and have them drive by and
[8:09] feel more like museums than they do um
[8:11] kind of warehouses. And so we're
[8:13] thinking about all of those things and
[8:14] as we go into new markets, we're kind of
[8:16] talking through those. But I think as a
[8:18] nation, we've got to figure out how to
[8:20] get people excited that we could be at
[8:22] the leading forefront of this technology
[8:24] rather than nervous and scared because
[8:26] if that is really what ends up really
[8:28] driving our decisions in this country, I
[8:30] think it's going to inhinder us from
[8:33] leading in this AI revolution and allow
[8:35] other countries like China to be able to
[8:38] surpass us. And that's why I think this
[8:40] sense of coming together in this
[8:41] nationalistic pride of let's build the
[8:43] infrastructure that supports this
[8:45] compute is critically important because
[8:47] if it doesn't get built in the US it
[8:48] will get built elsewhere and we don't
[8:50] want that compute lying in other
[8:51] countries.
[8:52] >> Now I'm going to throw out a couple of
[8:54] things that I've heard various players
[8:55] doing and I want you to react to agree
[8:57] disagree you know and why. Uh the first
[9:00] is I've heard some people pitch well we
[9:01] should do collocation. we should do some
[9:03] like high performance manufacturing and
[9:05] colllocate it with data centers that
[9:06] changes the economics of uh the actual
[9:09] site and it allows us not only to get
[9:11] the compute online but then it also
[9:13] allows us to do high performance
[9:14] manufacturing. How do you guys think
[9:15] about that?
[9:18] >> I don't really fully follow uh kind of
[9:22] connecting those two. I think energy
[9:24] campuses are going to be a bigger and
[9:25] bigger thing which is you bring
[9:28] generation. So you not only bring the
[9:30] load and the consumption, you actually
[9:32] bring netu generation as well. And there
[9:34] might be other use cases, right? Because
[9:36] like we have a 500 megawatt data center,
[9:38] but it the IT load is 352 and the rest
[9:41] is kind of the cooling infrastructure
[9:43] that doesn't always get used. And so you
[9:45] can have kind of colloccated
[9:46] infrastructure that's able to consume
[9:49] some some of that power that's not being
[9:51] used. So I think there's a lot of
[9:52] creativity of how to build these energy
[9:54] campuses but it depends on what do these
[9:57] loads need what are the different
[9:58] requirements and how do you create kind
[10:00] of these utilized systems. I think the
[10:02] next phase historically for data centers
[10:04] was we go we ask a utility hey can we
[10:06] get power obviously we don't have much
[10:08] power and so we need to go build new
[10:10] power and so now a lot of uh
[10:11] infrastructure providers are thinking
[10:13] about how do we bring power with us so
[10:14] we bring power and we bring load and
[10:16] then we can help add capacity into the
[10:18] market while consuming as well and I
[10:20] think you'll see a lot more of that
[10:21] >> now that collocation or like the other
[10:24] offtakes of the energy what would they
[10:25] be doing is it other types of data
[10:27] consumption or or is there something
[10:29] else
[10:30] >> so as we Think about it from a Huda
[10:32] perspective. Today we have two large use
[10:34] cases, right? We have use cases for ASIC
[10:36] that kind of power the Bitcoin network
[10:38] and security. And we're also big
[10:39] believers. If Bitcoin is going to be a
[10:41] currency around the world that people
[10:43] trust, we want that compute or a large
[10:45] portion of that compute to be in the US
[10:47] rather in countries that um whether it
[10:50] be kind of Iran, Russia, Ukraine, China,
[10:53] like we we want it to be in countries um
[10:56] that are more on in in the west. And so
[10:58] maintaining compute in the US we think
[11:00] is kind of critical to national
[11:01] security. Separately is from an AI
[11:04] compute perspective. So these two loads
[11:05] are very different. AI compute is more
[11:08] bursty more based on workloads and kind
[11:10] of a traditional data center they
[11:12] consume and it's less price sensitive
[11:14] where Bitcoin you trade around price
[11:15] volatility. And so these are interesting
[11:17] loads that can actually work together
[11:19] because one doesn't have to be online
[11:21] all the time and one does. Um and so
[11:23] other use cases we're seeing in the US
[11:25] today, you see advanced manufacturing,
[11:27] you see these kind of onshoring, you
[11:28] know, these chip manufacturing
[11:30] companies, fab manufacturing companies
[11:32] that are all large scale energy campuses
[11:34] in the hundreds of megawatts, if not
[11:35] gigawatts as well. And you'll see more
[11:37] of that as we step into the world of
[11:39] robotics and advanced manufacturing.
[11:41] Now, another thing I've heard is there's
[11:43] obviously the mega sites, you know, uh,
[11:45] hundreds of um, megawws, but then I've
[11:48] also seen people talk about, well,
[11:49] actually closer to urban areas, we're
[11:51] going to need 50 to 100 megawatt sites.
[11:53] How do you think through the mega sites
[11:55] versus maybe the smaller sites that are
[11:56] closer to the density?
[12:00] >> I'll talk about it from a power then
[12:02] kind of a latency perspective.
[12:03] Historically, there's was this belief
[12:05] that like you could only do training
[12:08] farther out and you have to be much
[12:10] closer to population density to to to
[12:12] have better latency to run workloads. If
[12:14] folks use Chad GPT or use claw or use
[12:17] Grock, you realize that sometimes you
[12:18] ask it a question and it goes and it
[12:20] thinks for some time and then comes
[12:21] back. So that really quick latency is
[12:24] less relevant compared to a world where
[12:26] you're browsing
[12:28] your emails or social media or whatever
[12:30] it may be. And so we've seen that these
[12:33] large scale campuses actually the the
[12:35] ability to build kind of these inference
[12:37] loads not just in population dense areas
[12:41] and first uh kind of tier markets is a
[12:44] reality today. At the same time you have
[12:46] a power constraint which is how much
[12:50] power at scale can you get and that's
[12:52] becoming harder and harder. And so I
[12:54] think you're going to see a lot more
[12:55] smaller campuses being deployed because
[12:58] it's easier to be able to say all right
[12:59] let's consume 5 15 50 megawws rather
[13:02] than go build a thousand megawatt campus
[13:04] and so you'll see both and you'll see
[13:06] people building infrastructure across
[13:07] both sets.
[13:08] >> Now another thing that I think people
[13:10] are trying to figure out is uh at the
[13:12] same time we're trying to get power
[13:13] we're trying to get the data centers
[13:14] built and there's a bunch of um kind of
[13:16] constraints or obstacles in just those
[13:18] two things alone. Also there's a lot of
[13:21] movement in like what is going inside
[13:22] the data center. The hardware itself is
[13:24] being rapidly innovated on and these
[13:27] large language models are now putting a
[13:28] lot of money and they're trying to
[13:30] figure out like who can get the
[13:31] advantage on the actual metal itself and
[13:34] so are you guys participating in that or
[13:35] you are just trying to deliver you know
[13:37] powered shells or just the power itself
[13:39] and and leave that for somebody else.
[13:41] >> We are building full turnkey solutions.
[13:43] So we're not just building the powered
[13:45] shell we're building everything that
[13:46] goes inside the data center as well. And
[13:48] as we think about the long-term kind of
[13:50] competitive edge in the data center
[13:52] market, what matters most is your cost
[13:54] per token, which is like your unit of
[13:56] compute and your cost per token on a
[13:59] fully depreciated and advertised basis.
[14:01] And so that means you don't really care
[14:03] if the costs are in capex or in opex.
[14:06] You care about the all-in cost of that
[14:08] infrastructure. And the three pieces
[14:09] that really drive those costs are one
[14:11] the cost of energy, two the cost of the
[14:14] data center, and three the cost of the
[14:16] chips and the technology that
[14:17] orchestration runs it all to run it the
[14:19] most efficiently to get the most compute
[14:21] out of that infrastructure stack. And so
[14:22] at HUD, we're proud because we're not
[14:25] just a data center business. We've own
[14:28] power plants in our history. We own four
[14:30] power plants. We have behind the meter
[14:32] assets, front the meter assets. And so I
[14:34] think understanding that full value
[14:36] chain on the compute side we have
[14:38] another company called high-rise.ai that
[14:40] we incubated and that's like a neocloud
[14:42] business cororeweave. And so we
[14:44] understand this full stack from
[14:45] beginning to end and can think about how
[14:48] to drive efficiencies across a full
[14:51] chain to drive the thing that matters
[14:52] most which is your cost per token.
[14:54] >> Today's episode is brought to you by
[14:56] simple mining. Bitcoin mining has a
[14:57] reputation for being complicated, risky,
[14:59] and hard to evaluate as a real
[15:01] investment. If you're considering mining
[15:03] in 2026, what actually matters isn't
[15:05] headline profitability, it's uptime,
[15:08] repairs, and whether the operation is
[15:09] run like a real business. That's why
[15:11] I've been using Simple Mining. They're
[15:13] based in Cedar Falls, Iowa, and they run
[15:15] a white glove hosting operation where
[15:17] you own your miners. You choose your own
[15:19] pool, and you have Bitcoin sent directly
[15:21] to your wallet. They were featured on
[15:23] the Inc. 5000 list as the fastest
[15:25] growing company in Iowa with over 40,000
[15:27] machines under management. What stands
[15:29] out to me is execution. They have the
[15:31] number one rated ASIC repair center. And
[15:33] for the first 12 months, repairs are
[15:35] included. If mining margins get tight,
[15:38] you can pause with no penalties. And if
[15:39] you want to resize or upgrade your
[15:41] fleet, there's a marketplace to resell
[15:42] equipment instead of being stuck. To
[15:44] help people think it through whether
[15:46] mining actually makes sense right now,
[15:47] they put together a short resource
[15:49] called the 2026 Bitcoin mining
[15:51] blueprint. It walks through the five
[15:53] mistakes investors make when allocating
[15:54] to mining, and they also explain how to
[15:56] avoid them before deploying capital. If
[15:58] it sounds interesting to you, you can
[16:00] get it for free at simplemining.io/p.
[16:04] That's simplemining.io/pom.
[16:07] Go check it out today and see if you
[16:09] should get into the mining game. Today's
[16:11] episode is brought to you by Fountain
[16:13] Life.
[16:14] Your body is incredibly good at hiding
[16:16] disease
[16:19] unless you look.
[16:31] True wealth isn't measured in
[16:32] possession.
[16:34] It's measured in the time you gain,
[16:36] the energy you feel, and the life you
[16:39] create. It may be the smartest
[16:41] investment of your time you'll ever
[16:42] make.
[16:49] Life is so very very precious and I want
[16:53] each of you to be truly awakened to your
[16:57] aliveness and live longer without
[17:00] limits.
[17:06] >> Your number one job is to educate
[17:08] yourself and Fountain wants to help
[17:10] educate you.
[17:16] Now you guys are striking very large
[17:17] deals. Last one was over $9 billion of
[17:20] uh revenue. I think that uh that that's
[17:22] part of that contract. Um what's that
[17:24] process like? So you guys know, okay, we
[17:25] got the land. We're going to go power
[17:26] with this thing. But we need to have a
[17:28] partner. We need to have somebody who's
[17:29] actually going to pay us for this. Do
[17:31] you just call up a bunch of companies
[17:33] and say, "Hey, I got land. I got power.
[17:35] Let's have a conversation." you know
[17:37] just walk through maybe how do you get
[17:38] in touch with these people and and what
[17:40] are those conversations like right now
[17:42] >> we've built really deep relationships
[17:44] with some of the largest consumers of
[17:46] comput in the world that's some of the
[17:47] largest investment grade hyperscalers
[17:49] that's the AI labs like the anthropics
[17:52] open AI X's that's also the fast growing
[17:56] companies that are AI companies that are
[17:58] growing as well and creating use cases
[18:01] across the board that are consuming
[18:03] hundreds of millions if not billions of
[18:04] dollars of compute and so you have
[18:06] massively growing market. Um, and then
[18:08] enterprise is slowly starting to
[18:10] percolate right now as well. And so for
[18:11] us, we're excited because we've
[18:13] approached these relationships as
[18:14] partnerships. And the way that we've
[18:16] approached it is not we have this piece
[18:18] of inventory. Do you want it or not? But
[18:20] what is your long-term growth map? And
[18:22] can we be infrastructure partners to you
[18:24] all? And that's how we've really
[18:26] approached and been building our deals
[18:27] that come together. And that's really
[18:29] kind of the mindset that we're taking as
[18:31] we're continue to grow and scale.
[18:32] >> Now $9 billion. I think there's a lot of
[18:35] companies who would love to sign a deal
[18:36] like that. How do you know the numbers
[18:38] are right? You know, so much is changing
[18:39] in terms of some of these are 10 plus
[18:41] year contracts. And I think one of the
[18:43] questions that investors have of the
[18:45] hyperscalers in particular is like are
[18:46] they going to get a return on their
[18:47] investment? You guys are a little bit
[18:49] different because you're basically
[18:50] signing up it as revenue, but how do you
[18:52] protect yourself or what are some of the
[18:53] things that if you're an investor
[18:54] looking at a company like HUD or other
[18:56] that people should be looking for to
[18:58] know, hey, this contract is good, it's
[19:00] high quality, they're likely to get paid
[19:01] on it.
[19:03] So today, if you look at the two deals
[19:05] that we've announced, they're 15-year
[19:07] contracts. Both of them, both of them
[19:09] are investment grade counterparties that
[19:11] back the payments of those obligations
[19:13] during the 15 years. And so you have
[19:16] basically the most valuable
[19:18] multi-trillion dollar companies in the
[19:19] world supporting uh the cash flows of
[19:22] these and those are diversified
[19:23] businesses that have a variety of
[19:25] different revenue streams. Um and so you
[19:27] have an investment grade kind of
[19:28] counterparty standing behind it. It's
[19:30] not just a startup that is using the
[19:31] compute. It's a take or pay kind of
[19:34] triple net lease where they're obligated
[19:36] to pay those payments every single month
[19:38] regardless of if they want to use it or
[19:40] not. Like they are committing to that
[19:41] infrastructure stack and we have a
[19:43] commitment and belief in them that
[19:45] they'll continue to grow and continue to
[19:46] scale. And so in total between those two
[19:48] over the duration of those contracts,
[19:50] it's almost $17 billion in total
[19:52] contract value between the two projects.
[19:54] Um but stabilized contract value with
[19:57] some of the most creditw worthy
[19:58] companies in the world um backing them.
[20:02] Now you have this Riverbend project.
[20:04] It's one of the two big ones that you
[20:05] guys have. Um there's some unique or or
[20:07] like innovative financing that you've
[20:09] put around this project. Can you talk
[20:10] about you know what is the financing and
[20:12] how it's structured? But why is this
[20:14] unique and no one's done it before?
[20:16] >> So most of the financing that we see in
[20:18] the market in the data center space is
[20:21] kind of short-term construction like
[20:23] bridge financing. So whether that be 2
[20:24] to 5 year 10 years meaning you get the
[20:26] money it's a little bit more expensive
[20:28] but you go and you construct and build a
[20:30] data center and then once the data
[20:32] center is stabilized then you go and you
[20:34] refinance that project and you get a
[20:36] little bit of a cheaper cost of capital
[20:37] because you don't take the risk of
[20:38] construction and you refinance it for
[20:40] the duration of the project. What was
[20:42] interesting is from a credit
[20:44] perspective, so we had S&P and Fitch uh
[20:47] kind of radar project. Usually you get
[20:49] kind of like a subinvestment grade
[20:51] rating during construction because the
[20:53] construction risk and then once you
[20:55] finish the construction depending on who
[20:57] your tenant is and who your offtake is,
[20:59] then if it's an investment grade part
[21:01] counterparty, then you'll get an
[21:03] investment grade uh credit after
[21:04] commercialization.
[21:06] What was unique is because of how we
[21:08] de-risk execution of the project and
[21:10] delivery, our project was actually
[21:13] investment grade day one and that
[21:15] allowed us to be able to go and raise
[21:18] capital from the investment grade
[21:19] markets day one instead of from the high
[21:21] yield markets or from the project
[21:22] finance uh uh kind of bank markets and
[21:25] so that the investment grade markets
[21:27] gave us duration. So instead of doing a
[21:29] two to fiveyear financing because I
[21:31] think the risk you face there is in two
[21:33] to five years you you have all these
[21:35] data centers being built today that are
[21:37] all going to need to be refinanced and
[21:39] so you'll have maybe 100 plus billion
[21:41] dollars of infrastructure that needs to
[21:43] be refinanced at that time and so what
[21:45] is going to be the appetite to refinance
[21:46] in that moment what is going to be the
[21:48] overall sentiment in the market and I
[21:50] think that's a risk that we really
[21:51] wanted to take away in the first project
[21:53] when we announced Riverbend and because
[21:55] of the investment grade status we felt
[21:57] like we could take away that risk. And
[21:59] so what we ended up doing was we're one
[22:01] of the first investment grade kind of
[22:03] construction phase data center projects
[22:05] in the market ever. And we were taking a
[22:06] bit of a risk when we did it because it
[22:07] was kind of a first of its kind. And we
[22:10] size the uh capital we raise to over 16
[22:14] 1/2 years. So instead of 2 to 5, it
[22:17] basically covers a construction period
[22:18] plus the full 15-year lease. and the
[22:21] full debt gets paid down via the cash
[22:23] flows of the project rather than via a
[22:27] refinancing that you expect to have. So
[22:29] what we felt like is for Had it fully
[22:32] derisks any refinancing risk and we've
[22:34] already derised so much of the project.
[22:35] It took the financing piece out of it as
[22:37] well. And for investors, they didn't
[22:39] have to take the bet that we were going
[22:41] to be able to refinance us to pay them
[22:42] back. We had the cash flows that are
[22:44] backed by investment counterparts to be
[22:46] able to do so. And so it was led by JP
[22:48] Morgan, Goldman Sachs, and Morgan
[22:50] Stanley. And we had a phenomenal
[22:52] outcome. Really proud. And had some of
[22:54] the most institutional, some of the most
[22:56] blue chip names in the market. So we
[22:58] raised over $3.2 billion. And we had 10
[23:02] billion dollars plus of uh demand across
[23:04] the book from the most well-known
[23:06] institutions around the world.
[23:08] >> How many of these can you do a year?
[23:09] Right. So you guys have two major ones.
[23:11] You obviously have the Bitcoin mining,
[23:12] you've got some other projects that are
[23:13] underway, but is this like one mega
[23:15] project per year? is the pace at which
[23:17] you think you can go. Can you do four or
[23:18] five of them?
[23:20] >> So, we're doing two already. Um, and
[23:22] we're looking at expansion on those
[23:23] campuses as well. But as a company, the
[23:26] way that we're really building is if you
[23:29] think about like some of the most
[23:30] successful kind of hedge funds, they
[23:32] have this pod model, right? Where you
[23:34] have pods that go and run different
[23:35] initiatives. And that's really kind of
[23:37] how we've built our business. We have
[23:38] this PR these principles that are part
[23:40] of a task force and they go and they
[23:42] build and run a project. And so for us
[23:44] and then we have partners that can scale
[23:45] with us. The reason we chose Jacobs and
[23:48] Vera from an EPCM which is engineering
[23:51] uh construction management to supply
[23:54] chain is because these companies are
[23:56] some of the biggest companies in the
[23:58] ecosystem. Jacobs is 40,000 employees
[24:00] worldwide. Ver is a 100 plus billion
[24:02] dollar company and so they can grow with
[24:04] our scale and our demand as well. Um,
[24:07] and then from a financing perspective,
[24:09] who has a stronger balance sheet than JP
[24:10] Morgan and Golden Sask and Morgan
[24:12] Stanley? And so we're really proud of
[24:13] the partners we brought together so that
[24:15] as we scale, they can scale with us and
[24:18] support our growth as well.
[24:19] >> As a CEO of HUT, how do you think about
[24:21] building stuff yourself versus acquiring
[24:24] other companies or other uh properties
[24:26] and sites? Is there a world where you'll
[24:28] do both or are you guys pretty focused
[24:30] on just uh development?
[24:32] >> We're really proud of our ability to
[24:34] develop. We think that we take a first
[24:36] principles approach towards how we
[24:38] develop. How do we develop more
[24:40] efficiently? How do we develop at a more
[24:42] efficient cost? How do we develop at a
[24:43] higher quality? And really proud of what
[24:46] we build. At the same time, kind of the
[24:48] market es and flows. Sometimes people
[24:51] are really excited, things are really
[24:52] expensive, sometimes things are cheap.
[24:53] And I think as you move forward in the
[24:55] next couple years, execution is going to
[24:56] be key. And so there might be
[24:58] opportunities where we can buy
[24:59] opportunities, assets where they say,
[25:01] you know what, I need an execution
[25:03] partner to come into this project with
[25:04] us. and that might be an earlier stage
[25:05] project or I want some type of liquidity
[25:07] and that might be a later stage project.
[25:09] And so we have a corporate development
[25:10] M&A team that looks at opportunities all
[25:12] day. Um but we also our core kind of
[25:14] growth engine is our development growth
[25:16] engine because that's where we're going
[25:17] to get the best yield out of the time we
[25:19] spend.
[25:19] >> All right. So people are very convinced
[25:20] we need more data centers, we need more
[25:22] compute. What are the bottlenecks? Like
[25:24] energy is a bottleneck. Are there
[25:26] critical minerals or other things that
[25:28] you look at and you're like these are
[25:29] the things that are slowing us down
[25:31] right now?
[25:32] So if you think about the kind of the
[25:34] full chain, I would say right now it's
[25:36] energy capacity at scale. Um and so
[25:39] that's generation of energy capacity,
[25:41] that's transmission capacity if you're
[25:42] pulling it from the grid. And so that's
[25:44] from like an power infrastructure
[25:45] perspective. Then when you go up the
[25:47] chain to the data center perspective,
[25:48] it's long lead time supply items. So
[25:50] that's breakers, transformers, switch
[25:52] gears, and you're seeing some things
[25:53] that the government is doing in calling
[25:55] these kind of national security assets
[25:57] and bring manufacturing, invest into
[25:59] manufacturing and reshoring that. And
[26:01] the third is the chips, right? You're
[26:02] seeing all these memory companies
[26:04] skyrocket in terms of their share price
[26:06] like SanDisk and um and SK and some of
[26:09] these like large memory companies
[26:11] because it's not just the GPU itself.
[26:14] It's all the components that go into
[26:15] that infrastructure stack on the ship
[26:17] side that support that no matter how big
[26:19] or small. So I think across the whole
[26:21] chain the long lead time item or the
[26:23] bottleneck es and flows depending on
[26:25] where we are in the market. I think
[26:27] today people need more infrastructure
[26:29] quick um and they need more power
[26:30] capacity in the long term as well.
[26:33] >> Now, are there things that you guys are
[26:35] doing to try to fix some of those supply
[26:37] chain stuff or because the way you're
[26:39] developing and you've got these
[26:40] partners, you're kind of just
[26:41] susceptible to whatever the lead times
[26:43] are.
[26:44] >> No, we are we were definitely very very
[26:46] cognizant and deep into the supply
[26:48] chain. We think about how much does it
[26:51] cost to produce every single piece of
[26:54] equipment that we buy from someone else
[26:55] and what are the raw inputs and what are
[26:57] the lead times, what is the time to
[26:59] build look like and that's how we drive
[27:01] kind of our ability to be able to
[27:03] deliver projects in a super efficient
[27:05] manner. Um when we announced the
[27:07] partnerships with Jacob Inverters, we
[27:08] had almost every other kind of
[27:10] manufacturer and partner reach out to us
[27:12] and say can we join the program? Can we
[27:13] be a part of it as well? And so overall
[27:16] we're very very deep into the supply
[27:18] chain. That's how we think we can drive
[27:20] efficiencies, innovation. And right now,
[27:22] we don't just buy equipment or hire jigs
[27:25] and converters to do things. We are
[27:27] alongside them every single day
[27:29] designing the infrastructure, designing
[27:31] the equipment, the modulars that we put
[27:33] together and driving a lot of that
[27:35] innovation alongside.
[27:36] >> What are the things that keep you up at
[27:37] night? Like what what are you worried
[27:39] about?
[27:41] I think what I'm I would say as a CEO of
[27:45] HUD and I'll go kind of more broadly as
[27:47] a CEO of HUD is as we continue to scale
[27:49] and build how do we keep what makes us
[27:52] unique in the culture that we built and
[27:54] how do we scale effectively and I've
[27:57] always shared this is we care about
[27:59] scaling but scaling with credibility and
[28:02] being able to execute on everything that
[28:03] we say we're going to do rather than
[28:05] trying to move too fast or move too
[28:06] slow. So finding that right balance. Um
[28:08] and for HA like the vision we have long
[28:10] term is there's this huge opportunity of
[28:14] physical intelligence. I think if you
[28:16] look back in the '9s like you had these
[28:18] companies that said we're going to build
[28:20] internet companies and today you fast
[28:22] forward 20 30 years you have Microsoft,
[28:25] Amazon, Meta, Alphabet and all these are
[28:28] the most valuable companies in the world
[28:30] but each of them have actually different
[28:32] businesses from social media to search
[28:36] to e-commerce and so on. But at their
[28:39] core, they learn how to build businesses
[28:42] in an internet era. And the opportunity
[28:45] that we believe exists today that allows
[28:48] us to potentially become one of those
[28:49] multi-trillion dollar companies in 10,
[28:52] 20 years from today is physical
[28:55] intelligence. How do you use technology
[28:58] and redefine how you build in the
[28:59] physical world? And that's what we're
[29:01] most excited by in the long term. We're
[29:03] building this data center
[29:04] infrastructure, these critical assets,
[29:06] but how do we use the technologies that
[29:08] we're empowering to reinvent our
[29:10] businesses ourselves? And so as we look
[29:12] forward kind of 5 years and 10 years,
[29:14] data centers shouldn't take 6 months to
[29:17] build with all of our engineers, etc.
[29:19] They should be the ones driving these AI
[29:22] native solutions to be able to design
[29:23] these in days, not months. In addition
[29:26] to that, as we think about building, how
[29:28] do we make sure that we're able to
[29:30] augment kind of the talent that we have
[29:32] to build alongside um robotics and
[29:35] infrastructure as well to build more
[29:36] efficiently? I think like the beauty
[29:38] about AI is it's going to free people's
[29:42] times up to be more creative to be able
[29:45] to innovate more and you're going to see
[29:47] a whole new kind of industry being born
[29:50] out of creativity because I think what
[29:52] the human mind is a beautiful thing and
[29:54] rather than focusing it on kind of
[29:56] monotonous tasks if those can be uh
[29:59] taken over by AI how do we have that
[30:02] human mind focus on innovation focus on
[30:04] pushing kind of the world forward and I
[30:06] think that's we're excited by at HUD
[30:08] long term. And I think what keeps me up
[30:10] as night is kind of messing that
[30:12] opportunity up. I think we have a moment
[30:13] in time where HUD is in this perfect
[30:15] moment. We're big enough as a company
[30:17] where we can raise tens of billions of
[30:19] dollars to support these large
[30:20] infrastructure projects, but we're small
[30:22] enough where we have this mindset of a
[30:25] high growth startup and a fast growing
[30:27] business that wants to win and wants to
[30:29] take over the world and we can rethink
[30:31] and reframe every single thing that we
[30:33] do. And so we're really grateful and
[30:35] fortunate uh to be where we are today
[30:37] and we want to be able to really build
[30:39] out what the potential of this business
[30:41] can be.
[30:42] >> Talk for a second about uh American
[30:43] Bitcoin. You guys spun it out. Um I
[30:45] think people are very interested in the
[30:46] Bitcoin mining still uh as well. What's
[30:49] the latest on that business? How do you
[30:50] think about it?
[30:51] >> It's great. Uh we went public less than
[30:53] a year ago. We have over 7,000 Bitcoin
[30:56] from zero when we spun it out.
[30:57] Obviously, HUD has its own Bitcoin stack
[30:59] as well, but the business has been doing
[31:01] well through the last three quarters.
[31:04] Regardless of Bitcoin price, our margin
[31:06] has been over 50%. Which is amazing. And
[31:09] that's the beauty because even if
[31:11] Bitcoin price comes down, other people
[31:12] that are less efficient stop mining and
[31:15] stop using the compute and that
[31:17] therefore the difficulty comes down. And
[31:19] so our net margins actually remain the
[31:22] same even with the volatility uh of
[31:24] Bitcoin, which is awesome. And so the
[31:25] business, we're super proud of the team,
[31:27] what they continue to do, what they
[31:28] continue to build. They just brought a
[31:30] site on uh last quarter. And I think
[31:32] what's unique around American Bitcoin is
[31:34] because everyone is moving to AI, it's
[31:36] actually becoming easier to use that
[31:38] same compute to buy more Bitcoin. Um and
[31:40] so we think long-term kind of the
[31:42] security and the integrity of the
[31:44] Bitcoin ecosystem is critically
[31:45] important. And we're one of the last
[31:47] remaining large scale operators that are
[31:49] providing that in North America. Um and
[31:51] we're really proud of to do that. And
[31:53] that business is is is going well. the
[31:54] foundation is building. Obviously,
[31:56] there's stock price volatility, but
[31:58] that's kind of more macro that we can't
[32:00] really control, but what we can control
[32:01] is the fundamentals of the business. And
[32:03] what I've always told the team, I was
[32:04] like, look, when I took over the CEO's
[32:06] HUD 8, our stock price was around $6. I
[32:09] mean, you remember this, you were one of
[32:10] the early folks that said, hey, people
[32:12] should take a look at HUD 8, right? And
[32:14] we fast forward two years, we've
[32:16] appreciated by over a,000%. Um, and I
[32:18] and I told the team, I was like, what
[32:19] mattered most is not focusing on what
[32:21] happens on the stock price on a
[32:23] day-to-day basis. If you build an
[32:24] amazing underlying business, the rest
[32:26] will follow.
[32:28] >> Makes uh makes sense to me. Thank you
[32:29] very much for taking the time to do this
[32:31] and uh we'll do it again soon. Thanks
[32:33] for having me