This conference call transcript was computer generated and almost certianly contains errors. This transcript is provided for information purposes only.EarningsCall, LLC makes no representation about the accuracy of the aforementioned transcript, and you are cautioned not to place undue reliance on the information provided by the transcript.
spk00: Good morning, and welcome to HUD-8 Q3 2024 Financial Results Conference Call. After today's presentation, there will be an opportunity to ask questions. Please note that this event is being recorded and a transcript will be available on HUD-8's website. In addition to the press release issued earlier today, you can find HUD-8's quarterly report on Form 10-Q on the company's website at www.hud8.com. Under the company's Edgar profile at www.sec.com, and under the company's CDAR Plus profile at www.cdarplus.ca. Unless noted otherwise, all accounts referred to during the call are denominated in U.S. dollars. Any comments made during this call may include forward-looking statements within the meaning of applicable security laws regarding HUD-8 Corp and its subsidiaries. The statements may reflect current expectations and, as such, are subject to a variety of risks and uncertainties that could cause actual results to differ materially from current expectations. These risk uncertainties include but are not limited to factors discussed in HUD-8's Form 10-Q for the three and nine months ended September 30 of 2024 and Form 10-K for the year ended December 31, 2023, as well as the company's other continuous disclosures documents, except as if required by applicable law. HUD-8 undertakes no obligation to publicly update or review any forelooking statements. During the call, management may also make reference to certain non-GAAP measures that are not separately defined under GAAP, such as adjusted EBITDA. Management believes that non-GAAP measures taken in conjunction with GAAP financial measures provide useful information for both management and investors. Reconciliations between GAAP and non-GAAP results are presented in the tables accompanying the press release, which can be viewed on HUD-8's website. I would now like to turn the call over to Asher Janute, CEO of HUD-8.
spk07: Good morning, everyone, and welcome to our third quarter earnings call.
spk06: The digital infrastructure industry is on the brink of transformation. As AI drives unprecedented demand for computing scale and density, we are witnessing a fundamental re-evaluation of every link in the data center development value chain, from power sourcing to infrastructure design and construction. As this transformation unfolds, we believe we are well positioned to capitalize on these secular tailwinds. Our vision is to build a platform that integrates energy and digital infrastructure at scale, transforming how energy capacity is secured and deployed to power the technologies of today and tomorrow. In the near term, we are focusing on capitalizing on the significant demand for AI computing capacity with a distinct power-first approach. Looking ahead, we see an opportunity to build a flexible energy infrastructure platform that evolves alongside breakthrough technologies for decades to come. Before discussing how we are positioning the business to achieve these objectives, I'll share some highlights from the quarter. As a reminder, the current period reflects the performance of the combined company, while the comparison period reflects U.S. Bitcoin Corp's performance as a standalone business prior to the merger. Turning to page three, our revenue grew 102% year-over-year to $43.7 million through the three months ended September 30th, 2024. Looking ahead, we anticipate continued top-line growth driven by two initiatives from the quarter, a 15-exahash co-location agreement for ASIC compute at our Vegas site and the launch of our GPU-as-a-service business. Together, these initiatives are expected to generate nine figures of annualized revenue. Net income for the quarter was $0.9 million versus a loss of $4.4 million in the prior year. And adjusted EBITDA was $5.6 million versus $11.5 million in the prior year. Both net income and adjusted EBITDA reflect a loss on digital assets of $1.6 million and a gain on debt extinguishment of $6 million. Our ongoing focus on operating rigor and cost management continues to drive improvements to unit economics. Despite rising difficulty, declining hash price, and older generation machines in our self-lining business, for example, we continue to prioritize the most challenging levers to optimize our operation and drive costs down. These initiatives have led to a nearly 28% reduction in our average cost of energy per kilowatt hour over the past two quarters. Turning to page four, we executed three major strategic initiatives this quarter. First was our partnership and co-location agreement with Bitmain to commercialize a rack-ready liquid-cooled ASIC miner through a 15-exahash co-location deployment at our Tier 1 Vega data center, which will be capable of cooling server density up to 200 kilowatts per rack. Second was a launch of our GPU as a service business, High Rise AI. And third was a conversion of the outstanding balance of our Anchorage loan to equity. After the quarter, we also signed a purchase agreement with Bitmain to upgrade 111 megawatts of self-mining capacity across our existing fleet in the first quarter of 2025. Let's discuss how these initiatives advance our ambition to build a next generation energy infrastructure platform. We start at the foundation of our platform in the power layer. Historically, data center operators often treated power as an afterthought in developing digital infrastructure. We are taking a different approach, putting the acquisition of high quality power assets at the forefront of digital infrastructure development. Our playbook is straightforward. we seek to secure as many megawatts as possible, assign each megawatt to the highest return use case, and maximize yield over time. Building tier one data centers for Bitcoin mining ASIC compute is central to this approach because we believe it enables us to monetize power assets rapidly and cost effectively to generate strong unlevered returns, even when traditional tier three data center workloads are not immediately viable due to factors like time to energization, land size, load capacity dynamics, or fiber access. In other words, developing tier one data centers for Bitcoin mining is a tool that allows us to scale our power layer and maximize load capacity secured for the long term. In practice, the advantages of this model are exemplified by our 205 megawatt Vega project. When Vega came across our desk, its potential was clear. A large scale behind the meter site with an existing substation and immediate access to some of the lowest power prices in North America. However, a contractual obligation to begin consuming power at the substation by the second quarter of 2025 impose a tight timeline to commercialize and develop a tier three data center for AI compute at the outset. Demand for tier three data centers is robust, but commercialization timelines are typically longer compared to the rapid deployment possible for Tier 1 data centers for Bitcoin mining. Our underwriting, however, supported acquisition and rapid development of the site for a Tier 1 data center to power Bitcoin mining ASIC compute as the initial use case. Today, we are on track to energize Vega and start generating approximately $135 million in annualized revenue at the site in the second quarter of 2025. less than a year after acquiring it, we retain the flexibility to repurpose the site for other use cases like AI to optimize long-term value. We continue to target high potential assets like Vega, prioritizing both speed and quality in developing our portfolio. Increasingly, we are focusing on front of the meter opportunities to meet rising demand from AI hyperscalers and other strategic partners. Front-of-the-meter assets can provide crucial benefits for AI data center development projects, including full ownership of interconnections and robust substation infrastructure designs. Turning to page five, we've applied our power-first approach to expand our development pipeline significantly since our last update by more than two gigawatts of assets under diligence and 500 megawatts in assets under exclusivity. As of October 31st, we were actively evaluating more than six gigawatts of expansion capacity with more than 1.5 gigawatts under exclusivity. Three projects from this pipeline are particularly promising for large-scale AI data center projects. Collectively, they represent over 430 megawatts of capacity with power delivery expected to be available before the end of 2025. We are actively exploring various commercial structures for these projects across a range of customer profiles. Next, I'll turn to the digital infrastructure layer, where we design, build, and monetize purpose-built facilities to maximize returns on our power assets. Our goal is to develop a platform that meets current demands while evolving alongside new technologies and markets over the coming decades. This requires us to design for where the industry is headed, not just where it stands today. Our framework for digital infrastructure development uses a tiered classification system based on redundancy and resiliency levels, rather than specific workloads like ASIC or GPU compute. This application agnostic approach allows us to innovate and deliver scalable value engineered solutions that support a range of current and future workloads. Today, we focus on designing, building, and operating data centers at both ends of the spectrum. At one end are Tier 3 data centers for GPU compute, powering AI workloads, which require significant capital investment and 5.9 reliability with N plus 1 redundancy for mission-critical applications. At the other end are Tier 1 data centers for ASIC compute, powering Bitcoin mining workloads, where lower capital requirements and minimal redundancy requirements make low operating costs and rapid deployment a priority. The cost difference in data center development between these extremes is substantial. Bitcoin mining ASIC workloads can be deployed in tier one data centers that cost less than $400,000 per megawatt to develop. While mission critical AI GPU workloads require tier three data centers that cost approximately $10 million per megawatt. Looking forward, however, we see a new paradigm emerging between these two extremes. Let's turn once more to our Vega project, where we are pioneering a new Tier 1 data center form factor that narrows the gap between Tier 1 and Tier 3 infrastructure. Historically, Tier 1 data centers designed for Bitcoin mining relied on shelving and forced air cooling, limiting them to ASIC compute. With our Vega project, we're disrupting this paradigm. The custom infrastructure we developed and engineered in-house for the project features a high-density rack, direct liquid-to-chip cooling, and HVAC-supported air cooling. Inspired by Tier 3 data center architecture, this design will enable Vega to support rack-based deployments of ASIC compute at densities of 180 kilowatts per rack. surpassing even the 120 kilowatt density required by NVIDIA's latest Blackwell GPUs. And despite these innovations, we expect to build Vega for less than $400,000 per megawatt, a fraction of traditional data center costs. Our partnership with Bitmain to launch the U3S21EXPH was integral to the development of this new tier one data center form factor. This will be the first ASIC miner mass commercialized by Bitmain to feature direct liquid-to-chip cooling in a U form factor, allowing for higher density deployment of ASIC compute in the rack-based architecture we developed for Vega. Turning now to page six. With construction progressing rapidly, we expect to energize Vega in the first half of 2025. The project will be an early proving ground for our next generation rack-based Tier 1 data center design, where we can demonstrate our ability to commercialize innovative data center architecture with the speed and capital efficiency that have historically set us apart in Bitcoin mining. Thoughtful commercialization also plays an important role in maximizing returns in our digital infrastructure layer. Our co-location agreement with Bitmain at Vega exemplifies a balanced, risk-adjusted approach to growth. structured to offer the benefits of a traditional data center co-location deal, but with a shorter 18-month term, this agreement is expected to generate $135 million in annualized hosting revenue upon full rent, supported by a $30 million payment credited towards a fixed purchase price option for the 15 exahash ASIC deployments. With recent catalysts for Bitcoin price appreciation, We believe this structure positions us well to capitalize on the future upside while mitigating risk on the downside. More broadly, we apply a use case agnostic approach to monetizing our digital infrastructure, focusing on return profiles and risk-adjusted economics. When signing a tenant lease to co-locate a customer at a data center like Vega, for example, the primary factors we evaluate are normalized revenue relative to investments, contract duration, counterparty credit profile, and cost of capital, not whether the tenant operates ASICs or GPUs. A publicly traded miner or a major ASIC manufacturer with a strong balance sheet and stable cash flows, for example, might offer a more secure profile than a Series A funded AI startup. Finally, let's turn to the compute layer, where we invest strategically in application-specific hardware like ASICs and GPUs to capture the lucrative economics offered by emerging technologies like Bitcoin mining and AI compute. Our approach to scaling at this layer is fundamentally similar across applications given shared investment risk dynamics. Revenue in both ASIC and GPU compute markets is closely tied to supply demand volatility due to rapidly evolving technologies. For ASICs, revenue is driven by hash price. reflecting the market rate for computing power in Bitcoin mining. For GPUs, on-demand revenue is derived from compute markets where real-time demand determines pricing. On page seven, we outline the impact of our recently announced A6 fleet upgrade. Last week, we announced the purchase of more than 31,000 Bitmain S21 Antminer units to upgrade 111 megawatts of self-mining capacity across our existing fleet in the first quarter of 2025. We expect this initial upgrade to increase our self-mining hash rate by 66% from 5.6 to approximately 9.3 exahash per second, while improving our fleet efficiency by 37% from 31.7 to 19.9 joules per terahash. At a hash price of 4.5 cents, this upgrade could drive up to 18 percentage points of gross margin expansion in our self-mining segments. Our selection of the S21 Plus was guided by rigorous return-driven analysis. We evaluated ASIC models ranging in efficiency from 13.5 to 17 joules per terahash and determined that the S21 Plus would deliver the fastest payback at any hash price above $0.02. In choosing the S21 Plus, we are optimizing for financial metrics like capital efficiency and payback rather than chasing less meaningful metrics such as hash rate and fleet efficiency. More broadly, our ability to execute across a wide range of hash price scenarios and machine efficiencies is rooted in operating rigor and proprietary technology. In recent quarters, we've demonstrated our ability to manage the most operationally demanding drivers of profitability, optimizing energy costs and enhancing unit economics. Building on this foundation, our initial fleet upgrade now represents a straightforward lever to drive future growth and bottom line impact. Our purchase option for 15 exahash of hosted miners at Vega provides further potential for disciplined growth. Taking into account the fleet upgrade and assuming we fully exercise the option, we have a path to approximately 24 exahash per second with an average fleet efficiency of 15.7 joules per terahash as early as the second quarter of 2025. At a hash price of 4.5 cents, this could drive up to seven additional percentage points of gross margin expansion. As for AI compute, our GPU as a service vertical is now fully online and generating revenue under the high-rise AI brand. Our customer agreement provides for fixed infrastructure payments and a revenue share that will enable us to capture market demand upside. Initially incubated with balance sheet capital, the business unit is now progressing initiatives like capital raising and development of a cloud product. We recognize that businesses in both the ASIC and GPU compute markets are highly subject to supply-demand volatility, which can be partially offset by implementing hedging strategies such as forward cash price derivative markets for ASIC compute and long-term customer agreements that lock in hourly pricing for GPU compute. While we continue to explore strategies to mitigate risk, we believe our continued focus on operating rigor, proprietary technology, and structured dealmaking is central to our ability to capture the upside of that volatility while protecting against the downside risk. I'll conclude my remarks today by taking a step back and reaffirming our commitment to disciplined, fundamentals-driven growth. As we scale aggressively across each layer of our platform, our focus remains on building a robust balance sheet and maintaining high standards for capital allocation. We believe this discipline, combined with a strategy rooted in innovation, thoughtful commercialization, and operating rigor, positions us not only to deliver outsized value in the near term, but also to build an enduring generational business at the intersection of energy and technology. With that, I'll turn it over to Sean to discuss the impact of our strategy on our financial results in detail.
spk02: Thanks, Asher, and good morning, everyone. I'll start by reviewing our third quarter results by segment. As a reminder, the current period reflects the performance of the combined company, while the comparison period reflects the U.S. Bitcoin Corp's performance as a standalone business prior to the merger. Our digital asset mining business experienced a year-over-year revenue decline from $15.6 million to $11.6 million, driven primarily by a reduction in Bitcoin mines following the halving and an increase in network difficulty. While the halving and attendant increase in network difficulty are outside of our control, we remain laser focused on managing the variables that are within our control. Notably, our cost optimization efforts delivered significant bottom line impact with segment gross margins expanding by seven percentage points year over year from 30% to 37%. We achieved a 9% quarter over quarter reduction in our average energy cost per kilowatt hour from 3.2 to 2.9 cents, building on the 21% reduction we achieved last quarter. Looking ahead to Q1 2025, we expect further improvements to profitability as we roll out our initial fleet upgrade with up to 18 points of gross margin expansion, assuming a fixed market hash price equivalent to that in Q3 2024. Moving on to managed services, revenue increased more than four-fold year-over-year to $20.8 million, driven by the full ramp-up of our MSA with Ionic Digital, along with proceeds from a $13.5 million termination fee for Marathon related to exiting the Kearney and Granbury sites. Scale efficiencies contributed to a 16 percentage point expansion in gross margin from 70% to 86%. As we shared last quarter, we have temporarily paused business development in this segment to focus on scaling our own portfolio. However, as these results demonstrate, revenue from managed services continues to offset G&A and other fixed costs across the business. We turn now to our high-performance computing, co-location, and cloud business. Today, we operate five traditional data centers in Canada that serve more than 260 customers across government, financial services, media, and other industries. As the segment originates from the legacy HUD-8 business acquired through the merger, there is no revenue in last year's comparison period. Third quarter 2024 segment revenue was $3.4 million. Segment gross margins decreased slightly quarter over quarter from 26% to 24%. Finally, other segment revenue grew more than fivefold year over year to $7.9 million, driven largely by $5.4 million in power revenues from our four natural gas power plants in Ontario. In the coming quarters, this segment is expected to reflect the nine-figure annualized revenue we secured through our hosting agreement at Vega and the launch of our GPU as a service offering. As we expand and diversify our business, we expect that more predictable and financeable lower cost of capital segments will form a larger share of our revenue mix. With that in mind, we believe we are well positioned to capitalize on favorable fundamentals in power and digital infrastructure to scale business segments that offset the volatility of our compute layer revenue streams, such as self-mining and GPU as a service. In the near term, this means our focus will be on opportunities in AI data center development and co-location. Looking ahead to the remainder of the year, I'll outline our approach to capital planning, financing, and balance sheet management. Our capital allocation framework centers on funding balanced, risk-adjusted growth. Within this framework, our approach varies based on the dynamics of each segment and opportunity. For projects with contracted cash flows, we are comfortable with longer payback periods provided they remain within contract terms. For more volatile merchant-driven opportunities such as self-mining, we target shorter payback periods of up to two years based on current market dynamics. With substantial near-term growth on the horizon, we expect our capital needs to increase across both Bitcoin mining and AI. With respect to Bitcoin mining, near-term CapEx will primarily fund the ongoing development of our Vega project. We are targeting a build-out cost of approximately $400,000 per megawatt, which reflects the incorporation of direct liquid-to-chip cooling technology and implies total CapEx of approximately $80 million. We have already deployed more than one-third of this CapEx to date. and we expect to deploy additional capital through Q2 2025, with revenue generation anticipated in the same quarter. Additional CapEx for machine purchases, including our initial fleet upgrade and the potential acquisition of approximately 15 ExaHashel-posted machines at Vega, is planned for fiscal 2025. With respect to AI, the large-scale projects Asher outlined may require substantial CapEx, depending on the commercial structures we finalize with our partners. While we don't have definitive guidance to share at this stage, we are committed to disciplined capital allocation. With that, let's turn to our financing strategy. Our strategy centers on two objectives, securing the lowest possible cost of capital and minimizing enterprise risk. At the project level, we are actively exploring a range of financing options suited to the distinct needs of the opportunities in our pipeline. Project financing, historically unavailable for Bitcoin mining, is a viable option for AI data center development. Other structures like equipment financing and development credit facilities can help us optimize balance sheet exposure. At the parent level, aligned with our view of the company's intrinsic value, we will continue to take a pragmatic approach to equity issuance, balancing it with strategic leverage. And while we remain sensitive to dilution, we also recognize the importance of retaining the flexibility to raise capital when compelling growth opportunities arise or when market conditions are favorable. Finally, I'll discuss our approach to balance sheet management. We closed the quarter with a robust balance sheet supported by a well-laddered debt structure with no significant maturities until June 2025. With a combined $649 million in cash and Bitcoin as of the end of Q3 2024, which does not account for the recent increase in Bitcoin price, we believe we are well positioned to execute on our near-term growth initiatives. Our proactive approach to balance sheet management is exemplified by the recent conversion of our $37.9 million Anchorage loan to equity at a 51% premium to our 20-day VWAP the day prior to the signing of the debt repayment agreement. This transaction not only underscores Anchorage's competence in our growth trajectory, but it's also expected to yield $17.6 million in interest savings over the next three years. And when you capitalize this number, you can see we have created value through financing. We will continue to explore transactions that allow us to optimize our capital structure while minimizing impact on shareholder value. With respect to treasury management, we aim to take a flexible approach to asset management. This may include putting our stack to work to enhance liquidity, converting assets to working capital, using Bitcoin as collateral, or employing options strategies. This may also include simply holding Bitcoin as a strategic reserve asset. To conclude my remarks, I'd like to share my priorities as CFO for the remainder of the year. As Asher noted, the digital infrastructure sector is on the brink of transformation, and we believe our distinct power-first approach to data center development positions us well to capitalize on these secular trends. As I support the execution of our strategic priorities, my focus is fourfold. One, increasing the velocity of our power origination pipeline, which has grown by more than two gigawatts in assets under diligence and 500 megawatts in assets under exclusivity quarter over quarter. Two, credentializing our brand. Three, ensuring the soundness and cost effectiveness of our financing strategy. And four, building out the internal infrastructure required to support these initiatives. Reflecting my first quarter as CFO, I'm both energized by the progress we have made and excited by the immense potential we have yet to capture. And I look forward to advancing our vision to build an enduring generational business that will shape and lead the digital infrastructure industry for decades to come. With that, I'll turn the call over to the operator for Q&A.
spk00: Thank you. To ask a question, please press star 1-1 on your telephone and wait for your name to be announced. To withdraw your question, please press star 11 again. Please stand by while we compile the Q&A roster. Our first question comes from the line of Mark Palmer with the Benchmark Company. Your line is now open.
spk01: Yes, good morning. Very nice progress on all fronts. I wanted to ask about the current status of the company's discussions with... for its data center capacity, particularly AI hyperscalers, and what is the nature of those discussions at this point?
spk06: Thanks, Mark. Good to hear from you again. We have been really focusing hard on our AI strategy and specifically around commercialization of large-scale data centers, liquid to cold, for the purposes of AI. Similar to the beginning of the year when I first took over, the market asked, when is our fleet upgrade coming? And we didn't talk much about it until it commercialized and materialized. And then we started giving all the details of what we did, how we did it, and why we think it is accretive to the business and to our shareholders. So today, what we shared today earlier in the earnings is that we have sites. that we have isolated and marketed regarding AI projects and AI data centers. And we're in different stages of commercialization on those sites. And as things materialize and become definitive, we'll open to the market and explain how we did the deal and what the nuances in terms of the terms and the structure of those deals were.
spk01: Very good. And just one quick follow up along the lines of some of the comments that Sean just made. As you're looking at options for financing the build out of the Bitcoin mining side of the business, how do you think about the potential for equity issuance on the one hand versus the potential sale of some of the company's Bitcoin on the other as a means of financing that growth?
spk06: If we look at the Vega project, which we've already started building, we've made payments at that project, we've had cash on the balance sheet in order to fund those obligations and those billed CapEx expenses. I think we're constantly looking at what is the cost of capital across the board. Obviously the markets have been strong in holding our Bitcoin stack, has been accretive to the business over the course of the last couple of months. And so as we look at continued growth opportunities, will weigh both the investment returns of the actual projects that we're investing in, whether it be on the mining or AI side, and then looking at our cost of capital and what we believe is the most accretive at that moment in time.
spk01: Thanks very much.
spk00: Thank you. Our next question comes from the line of George Sutton with Craig Hallam Capital Group. Your line is now open.
spk03: Thank you, and I really appreciate the update. When you say you have 430 megawatts of capacity that could be powered by the end of 2025, I certainly recognize how valuable an asset that is. I'm curious how you actually could get that done. Are we talking brownfield sites, or can you just give us any sense on how you actually get something powered to that extent by the end of 2025?
spk06: So you're right. So power availability by end of 2025 is something that's extremely sought after and valuable. And that's why we're having such good and deep and engaging discussions on those projects. The actual commercialization of the data center has to do with the customer and the types of bills they're willing or not willing to do. We've been creative in the Bitcoin mining side of things, and we've been creative in finding the right partners and right vendors and right suppliers on the AI tier three data center side as well. And so we have some solutions. that are middle of the fairway, that have longer build times in the call it 12 to 24 month range, just like other large scale data center platforms have shared. And we have other solutions that are faster that can actually realize power coming available next year and being able to commercialize those. So based on the customer discussion and based on the actual customer, some of those sites will be able to realize that power when it becomes available and others we may realize that power quarter or two after it becomes available, but we'll still be far ahead in terms of available power and capacity. I think a lot of data center platforms today are looking at capacity, not in 2026, but really at end of 27 and 28. And so we think we're still in a very good position today and we're excited by the conversations we're having.
spk03: Great. And just on the purchase option of your product, your U-shaped Bitmain miners. I'm just curious, since you've announced that deal, I don't know the exact percentage, but Bitcoin's up a good 50% or so. Is it as simple as looking at the Bitcoin price as the major determining factor as to whether or not you exercise those options?
spk06: hash price so uh as we all know now that bitcoin price doesn't necessarily correlate with hash price so we run a range of sensitivity scenarios the great thing is we have a locked in price in order to purchase those machines right and that price was locked in when bitcoin price and hash price were much lower and so we feel very comfortable as we get closer to those machines being energized we'll in that period of that six-month option, every day we'll be analyzing what is the hash price, what is the return profile, and is this accretive or not? And I think that's similar if you look at our recent kind of modest fleet announcement in terms of the fleet upgrade of 111 megawatts. I mean, we saw kind of the trends in terms of Bitcoin elections and so forth, and we worked very hard in a short period of time to execute. And I think how we think about the business and this year has, I hope, shown it to the market is we will be contrarian and not necessarily go with the herd, but we'll also move quickly when we need to and be patient when we need to as well to drive the best returns on the CapEx that we invest into the business.
spk03: Okay, gotcha. Great job. Thanks, guys.
spk00: Thank you. Our next question comes from the line of Mike Colonese with HC Wainwright and Company. Your line is now open.
spk05: Good morning, Asher and team. Congrats on all the progress and I appreciate you taking my questions. First one for me, great job on seeing that nice reduction in power costs quarter over quarter last few quarters here. How should investors think about power prices going forward? Could we settle in around the sub three cent level like we saw in 3Q or could we see a little bit of a range kind of going forward?
spk06: So our power costs this last quarter obviously were extremely competitive at around 2.8 cents per kilowatt hour. I think how investors should look at our power costs is our ability to control one of the biggest operating costs that we have. And what I mean by that is our revenues on hash price, unless we hedge them out, are volatile, right? Hash price is volatile based on supply and demand. And unless that's hedged out, our energy costs, if we lock them in, we take basically a fixed price in terms of cost on one side, and then we take volatile top line revenue on the other. What we've been able to show the market is our machines are one of the oldest fleet in the market. Yet our gross margins and the way we operate them, I think, have been extremely competitive. And so in markets where we have volatile energy prices, like a lot of our exposure in ERCOT and also the ASO markets in Canada, we're able to use our proprietary software to manage our energy costs to maximize gross profit on those machines. And so that's very different. If our goal was to maximize revenue, we would increase our power costs, curtail less often and drive more Bitcoin produced. But our mindset has been, let's try profitability and use those profits to invest more thoughtfully into the business. And so as we upgrade new generation fleet of machines, we're going to optimize for profitability of those machines to drive the fastest payback. We're not optimizing for most amount of Bitcoin mined, even if we're losing money. And we're not optimizing for lowest energy costs, if that means we're giving up Bitcoin mined. We're optimizing for profitability and return on those investments that we make.
spk05: Great. That's helpful color. And I just wanted to ask a question around the natural gas power plants in Ontario. It's often overlooked, but you guys generated some revenues in the quarter from that asset. How do you think about using that asset going forward? Could we expect a potential sale to potentially free up some capital to fund some of your AI or Bitcoin mining initiatives? Or are there other use cases that we could expect from a monetization standpoint?
spk06: When I took over and when we discussed those assets in my first earnings calls, I mentioned that our core strength is not operating power generation facilities. It's consuming that power, empowering next generation technologies and infrastructure. And so we spent a lot of time figuring out what are the best ways to commercialize and drive value from these assets. And since Sean has joined us, He has a career being a managing director from sitting in the power utilities and renewable sector in understanding how to value assets and commercialize these assets. I'll pass this question off to him and he can share a little bit more as he's been looking at the best way to maximize value here.
spk02: Yeah, thanks, Asher. And as Asher mentioned, I, in my career, sold many, many, many gigawatts of natural gas fired power plants. As I came into my seat, this was something, these assets were something that were really interesting to me because of that background. And as we look out into the future, there are a few near to medium term catalysts that we think might increase the value of these plants, particularly a five-year capacity auction coming up in ESO. So I think as we balance, you know, looking at strategic alternatives for those plans. We want to make sure that we're maximizing the return that we get from them. So it's something we still have front of mind and we're evaluating our alternatives as we go forward. I don't know that, you know, we expect this to be a major source of our business, our revenues going forward, but we do expect to be able to realize a return on the investment.
spk07: Great. Thanks, Chris.
spk00: Thank you. Our next question comes from the line of John Todaro with Needham & Company. Your line is now open.
spk04: Hey, congrats on all the progress, especially on the growth on the mining side and then the continued focus on profitability. It's really starting to show through. Two questions here. Can we just drill, I guess, a little bit more to the 430 megawatts? So I understand that the customer might have differences and that might take longer to build, but I guess kind of almost a time frame. Is it fair to say that kind of the earliest you could get something would be maybe like mid-26th, and then depending on the customer, maybe end of 26 or early 27? And also, those differences, would you expect economic differences as well between those two customers on top line, maybe a different rental rack, so maybe a difference on revenue per megawatt?
spk06: When we think about kind of tier three data centers and how we build them, there's a range in terms of cost and I've shared kind of the ranges of 100 to 140 kilowatts per month and 10 years of 10 to 15 years. And so we continue to see that range across different customers we speak to. And in terms of structure from triple net to gross modified, we'll adjust that top line relative to kind of what costs are baked in. I think the way people should kind of think about the AI side of our business is that we're aggressively working on it. And as things become definitive, we'll share that more with the market. And I think as a company, we've taken the approach of, Honestly, playing things a little bit closer to the chest and sharing things as their material and their real and being able to explain how we did it. And I think similarly, like we've done so this year with other parts of our business, we'll continue to do that here. And I appreciate you guys asking these questions. And we'll show more as we kind of have today with the number of megawatts we have in that pipeline. But we are aggressively pursuing it. because of the availability of the power, it is valuable, and we're trying to figure out the best way to maximize this value.
spk04: Thanks for that. Yeah, I think that's likely the appropriate strategy. One more related to it, though, and that is just, would you put any kind of CapEx dollars towards it before you got leases signed, or should we expect you to kind of hold off until the lease is signed, then you'd start putting CapEx towards it?
spk06: There's some modest payments that can be made to move things along and some lead time equipment and so forth. Obviously, when you look at these projects, you're talking a billion dollars per hundred megawatts in that ballpark. And so relative to our balance sheet and relative to the risk profiles, we'll make modest investments in order to make sure that we're continuing to secure the value of those projects. And also, some of these projects are in Tier 2 markets where others are in Tier 1 markets. And so based on the scope of customers, that also gives us more confidence in putting some of these initial modest payments down. But they're not going to be drastic and large capex up front on a risk-based investment.
spk07: Great. Thanks, Asher.
spk06: Appreciate it.
spk00: Thank you. And I'm showing no further questions at this time. This does conclude today's conference call. Thank you all for your participation, and you may now disconnect.
Disclaimer