Bit Digital, Inc. (BTBT) Earnings Call Transcript & Summary

Hello, and welcome to the Bit Digital Fiscal Year 2024 Earnings Conference Call. Good morning, good afternoon and good evening, depending on where you're joining us from. Thank you for being here. We're just giving a few more moments for attendees to dial in, so thank you for your patience. [Operator Instructions] Also, as a reminder, today's conference is being recorded. I'll now hand it over to your host, Cameron Schnier, Head of Investor Relations at Bit Digital. Cameron, the floor is yours.

Thank you. Good morning, and welcome to the Bit Digital 2024 Earnings Call. Joining us on the call today are Sam Tabar, Chief Executive Officer; and Erke Huang, Chief Financial Officer. Before we begin, I would like to remind all participants that some of the statements we will be making today are forward-looking. These matters involve risks and uncertainties that could cause our results to differ materially from those projected in these statements. I therefore refer you to today's 10-K filing and our other SEC filings. Our comments today may also include non-GAAP financial measures. Additional details and reconciliation to the most directly comparable GAAP financial measures can be found in our 10-K filing, which is on our website. After our prepared remarks, we will open the call up for questions. [Operator Instructions] With that covered, I will turn the call over to Sam to discuss our performance. Sam?

Thank you, Cam. Ladies and gentlemen, thank you for joining us on the call today. Today, I'll walk through our 2024 results, highlight key milestones from a transformational year and provide insight into the strategic direction of Bit Digital as we scale our HPC operations. 2024 was a fantastic year for Bit Digital. Revenues grew exponentially by 141%. Margins expanded and adjusted EBITDA reached $73 million. This was driven by the rapid growth of our HPC business, which started in 2024 with one customer. And by early January of this year, we surpassed 20 customers. HPC revenue made up over 40% of full year revenue and more than half of Q4 revenue. The Enovum acquisition was a major leap forward. This acquisition vertically integrated our data center operations, added a strong customer base and brought in a highly experienced team. This team gives us a huge advantage in scaling our infrastructure business. We also built out our organization. Headcount has been focused on hiring seasoned value creators, all focused either on data center operations or cloud services. These are businesses you could just throw capital at. You need the right talent. We now have specialized teams leading both divisions. We have a lot to cover today. I'll walk through each part of the business. Erke will then walk you through the financials before we open the line for questions. First, let's start with our Cloud Services business. As we scale our AI infrastructure, we recently launched WhiteFiber, our new HPC platform that integrates GPU cloud services and data center operations. As part of this evolution, we have updated our segment reporting to better reflect our business structure. What was previously referred to as high-performance computing services is now categorized under HPC, the umbrella term for our entire WhiteFiber business. Cloud Services represents our GPU cloud platform, while colocation services include our data center business from Enovum. Cloud services did not exist in 2023, and it became our largest business in terms of revenue generation by the second half of 2024, producing $13 million of revenue in the fourth quarter of 2024. This segment contributed 50% of total revenue in Q4 and 64% of gross profit. Gross profit contracted slightly in Q4 as we added new GPUs and leased additional data center capacity ahead of revenue generation. We view this as temporary as we expect margins to normalize over time. A key factor was GPU leasing expenses, particularly the H100 sale leaseback from early 2024, which accounted for 70% of cost of revenue for that segment. That was a unique structure reflecting our risk tolerance at that time. Going forward, as we own more GPUs outright or deploy more traditional financing structures, we expect margins to expand. We currently have 9 active customers in our cloud segments of WhiteFiber. The majority are running single-digit servers with annualized revenue or ARR below $1 million, but most were won in the past few months. And this aligns with our strategy of onboarding customers and scaling deployments over time. We are happy to start with smaller initial deployments to earn trust, demonstrate performance and expand contract sizes as the relationships grow. Our cloud services run rate is approximately $62 million. This will increase to around $72 million later this month when the H200 contract of DNA fund begins generating revenue. Revenue generation on the contract was pushed out by a month as we ensured full reliability before going live. A decision made deliberately to prioritize customer experience and long-term satisfaction. Additionally, we have a $15 million in ARR expected to start at the end of June when we deploy 512 B200 GPUs for our anchor customer. And separately, we received our first 512 B200 GPUs, which are being deployed in Iceland. We expect this cluster to go live in April and plan to offer these GPUs through an on-demand pool via third-party platforms. This would be an interim step as we internally -- as we work internally on developing our own on-demand platform. While we see strong demand for reserved B200 contracts, we believe on-demand deployment effectively pulls forward revenue and shortens payback periods. Based on current market dynamics, this cluster could generate approximately $25 million in additional ARR. We also have about 113 H200 servers or [ 908 GPUs ] currently being configured, and we are evaluating reserve contract options for those units. Illustratively, at $2 per hour, that is another $16 million in ARR once contracted. Overall, our customer pipeline remains strong and dynamic. Demand for B200 is surging. Also, since DeepSeek, we've seen renewed enthusiasm for H100s and H200s because one can do less with more or rather one can do more with less. We're consistently engaging new prospective customers, and demand continues to outstrip supply. While we see significant growth opportunities, we are taking a disciplined approach to GPU procurement, carefully managing capital deployment to avoid excess inventory risk. Our focus is on growing at a pace that aligns with customer trust, ensuring that as relationships deepened, we scale deployments accordingly. We've also invested in top-tier technical talent to build a robust, software-driven infrastructure that enhances performance, reliability and scalability. Customers don't just need access to GPUs. They need a trusted high-performance platform that ensures seamless deployment and maximum performance. Our investment in this technology layer to deliver just that is a key differentiator, helping us drive customer trust, retention and long-term growth. Also, with Boosteroid, the third largest cloud gaming provider in the world, we continue to expand our partnership. Currently, we have just under 500 GPUs contracted, represented approximately $1.6 million in annual revenue over the 5-year term. We are in process of finalizing an agreement to deploy an additional 700 GPUs, which if completed, would generate an additional $2.4 million in annual revenue for Boosteroid. And we expect the deployment cadence to accelerate throughout 2025. Our GPU procurement strategy is a balancing act between growth and risk. We are focused on scaling customer deployments and expanding GPU capacity to meet growing demand. Several live opportunities are substantial with contracts representing a 9-figure annual revenue and 3- to 4-year locked-up terms. Executing these would require the right financing structures and we are actively evaluating lease financing and other capital-efficient options. We are firmly in the mix for blue-chip deals. As our AI compute grows, our focus remains on execution, efficiency and customer relationships. WhiteFiber is scaling rapidly, and we believe we are well positioned to be a leader in AI infrastructure. Turning to our colocation services segment of WhiteFiber. This business was established with our acquisition of Enovum in October 2024, marking a major step in our evolution as an HPC platform. Before the acquisition, we had no colocation business. Now, we operate a Tier 3 data center with a full roster of clients, which currently stands at 14 active customers. We added a recurring revenue stream and expanded our expertise. Beyond its immediate contribution, Enovum provides a scalable foundation for future growth, backed by an experienced team and a very robust development pipeline. The Enovum acquisition is a gift that keeps on giving. Since closing the acquisition, we have moved quickly to expand our colocation capacity and secure strategic customer segments -- sorry, customer agreements. In Q4, we acquired Montreal 2, a 160,000 square foot industrial site in Montreal for approximately $23 million as part of our planned expansion to 32 megawatts by 2025. We Montreal 2 is being developed into a 5-megawatt Tier 3 data center expected to go live in mid-2025. The facility will be powered by 100% renewable hydroelectricity and will feature direct-to-chip liquid cooling. The site provides key advantages in accelerating our development pipeline. The property was well suited for a retrofit and includes transferable HVAC infrastructure allowing us to reduce costs and bring capacity online faster. A core tenet of our growth strategy. We still expect to complete the retrofit for approximately $19 million. We are also in the process of securing cost-effective mortgage financing to support the build-out in a non-dilutive manner. We plan on announcing the customer for Montreal 2 at a later date. In February, we announced a multiyear colocation agreement with a leading AI hardware innovator. This client is Cerebras, the manufacturer of the fastest inference LLM processor in the world. Cerebras is launching 6 new data center sites in North America and shows us to be their partner for their first-ever Canadian data center. This agreement is a major validation of our colocation strategy, reinforcing our ability to provide high-performance build-to-suit infrastructure for industry leaders. Under this contract, we will provide 5 megawatts of customized high-density colocation capacity over a 5-year term. The location for the development has been selected, and we are in the process of finalizing legal ownership of the site. Once that process is complete, we will formally announce the location. We expect the contract to commence in mid-2025. Cerebras is pioneering wafer-scale technology, which enables ultrafast AI infrastructure for some of the largest and most complex AI workloads in the world. Their deployment with us will be the first of its kind in Canada expanding AI compute access to enterprises, research institutions and government entities. This deployment required a highly customized, high-density solution validating our ability to design infrastructure for next-generation AI workloads with very unique technical requirements. Beyond this initial deployment, we see significant potential for future expansion as Cerebras continues to scale its infrastructure. Their rapid growth reflects the increasing demand for high-density AI-optimized colocation. Our ability to meet their highly specialized requirements underscores the strength and adaptability of our platform. We are super excited to support them in their next phase of development. As mentioned, we are deep in the mix for blue-chip deals. And that's a prime example of one. Stay tuned for more. Beyond Montreal 2, our development pipeline has expanded significantly, now totaling 510 megawatts including 156 megawatts under exclusive LOI. This includes sites in both Canada and the U.S. with 6 locations under exclusive LOI ranging from 8 megawatts to 100 megawatts. The major driver for the pipeline expansion was the addition of locations in the United States, we recently brought a U.S. site under LOI that could redefine our data center platform. It's developed. It could be our largest project to date, significantly expanding our scale and market position. Even with our planned capacity expansions, we continue to receive more customer demand than we can currently accommodate. This underscores the urgent need for additional high-performance data center space and reinforces our approach of prioritizing execution speed and customer alignment. Everyone knows about the current tariff wars that are playing out. We're currently monitoring and assessing their potential impact on our data centers build out. Many critical components such as generators, HVAC systems and electrical infrastructure are imported from the United States, Canada and Mexico, and new tariffs could increase build costs. We are evaluating strategies to mitigate potential increases including diversifying supply chain and optimizing procurement, while monitoring if, how and when these tariff policies may take place. Looking ahead, we believe inference will be the largest driver for long-term AI compute demand. And we are positioning our data centers to capture this shift. Strategically, we are developing in metropolitan areas where we expect the broadest customer appeal over time. This ensures we can meet the needs of enterprise, government and research institutions seeking low latency, high-performance AI infrastructure. Turning to our Bitcoin mining business. We remain focused on maintaining a cost-efficient and optimized fleet rather than growing hash rate for the sake of expansion. Mining accounted for 54% of revenue in 2024, down from 98% in 2023 as we prioritized investments in HPC. That said, mining remains an important part of our business, and we are taking targeted steps to improve efficiency and reduce costs. Our active mining fleet is currently around 1.6 exahash with an efficiency of approximately 25 to 26 joules per terahash. We have now fully exited all claimant facilities and are refreshing our fleet with more efficient miners at new hosting sites. To replace [ lost deployment ] capacity, we secured 30 megawatts of new hosting, 19 megawatts with our Digital and 11 megawatts with Soluna. The 11-megawatt site was filled with 1,800 redeployed K-Pros and about 1,400 S21s and S21 plus units. The 19-megawatt site will support some redeployed assets as well as 3,800 S21 plus miners adding 820 penta hash to our fleet. To date, we have deployed 941 S21 miners and 500 S21 plus units improving overall fleet efficiency. These upgrades are expected to bring our operational hash rate to approximately 2.5 exahash by May with pro forma efficiency to around 22 joules per terahash. Reaching 3 exahash would require securing an additional 6.6 megawatts of hosting and acquiring approximately 1,800 more S21 plus miners. The power is available from multiple sources, and we are actively evaluating the best path forward. Our strategy remains unchanged. We are not allocating significant growth capital to mining. Instead, we are structuring the business to maintain Bitcoin exposure on a capital-efficient way, focusing on fleet optimization and cost reductions while keeping a disciplined approach to capital deployment. Mining remains a part of our portfolio, but our investment priority remains on scaling our HPC business. I'll now hand over the line to Erke, who will discuss our financial results.

Thank you, Sam. I will now discuss our financial results for 2024. As a note, we completed our transition to domestic issuer status and filed our first Form 10-K with the SEC this morning. Total revenue for the year was $108 million, a 141% increase from 2023. Revenue increased across all business lines. Bitcoin mining revenue was $58.6 million, up 32% year-over-year. Bitcoin production declined 37% to 950 Bitcoin due to higher network difficulty and the April [ halving ] event. However, higher Bitcoin prices and increased hash rate led to the overall revenue growth. Ethereum staking revenue more than doubled to $1.8 million for 2024 as we earned approximately 566 Ethereums in staking rewards for the year. Cloud Services generated $45.7 million in its first year of operations, starting revenue in Jan 2024, colocation services acquired through M&M contributed $1.4 million from October 12 through the year-end. Our total cost of revenue excluding depreciation and amortization was $62.4 million compared to $29.6 million in the prior year. The increase was driven by a larger mining fleet, higher network difficulty and the launch of our cloud and colocation businesses. Cloud services costs were $19.5 million with $13.6 million from H100 server lease expenses. Gross profit was $45.7 million and nearly threefold increase from 2023. Gross margins expanded approximately 500 basis points to 42.3%, driven by cloud and colocation revenue, which offset lower mining margins due to the block rewards reductions and higher network difficulty. General and administrative expenses were $41.5 million, up from $27.7 million. The increase was mainly due to higher payroll and professional fees. A large portion of these fees were tied to the M&M acquisition and are not expected to recur. Depreciation and amortization was $32.3 million compared to $14.4 million in 2023, reflecting a larger miner and GPU fleet. 2024 adjusted EBITDA was $73 million compared to $12.4 million in 2023. Adjusted EBITDA for 2024 includes $55.7 million of gain on digital assets, which are predominantly unrealized gains. GAAP earnings per share was $0.19 for 2024 on a fully diluted basis compared to a loss of $0.16 in 2023. Turning to our balance sheet. We held approximately $98.9 million of cash and restricted cash as of December 31, 2024, and our digital asset positions was worth approximately $161.4 million. Total assets were $538 million, and shareholders' equity was $463 million. We remain debt-free by actively exploring financing options for our HPC business. Capital expenditures for 2024 totaled $94 million, the majority of the CapEx was deployed in the fourth quarter and was used to found GPU purchases and the acquisition of Montreal 2. I will now turn the call back to Sam for closing remarks.

Thank you, Erke. Before we open the line for questions, I want to touch on the broader trends that we're seeing across our business. We are experiencing significant and sustained demand for compute infrastructure, far exceeding the capacity we are currently bringing online. The need for high-performance computing continues to expand, and we believe we are well positioned to capitalize on this long-term trend. This demand stands in stark contrast to broader market sentiment, where we've seen risk assets, including our own stock come under pressure. Specifically for Bit Digital, the bigger are HPC business, the more our stock seems to trade like a pure-play Bitcoin miner, which is ridiculous. We believe that we are deeply misunderstood, but time is our friend. And eventually, we believe we will be valued properly. While we can't control the macro environment, we are focused on what we can control, executing our strategy, expanding our infrastructure and positioning our business for long-term value creation. We believe we are well on our way. A key advantage for our model is the synergistic nature of our colocation and cloud business. These segments are complementary serving different stages of the AI infrastructure value chain with distinct earnings profile and payback periods. Colocation provides long-term contracted revenue streams. Cloud services offer high-margin shorter duration contracts with greater flexibility. Together, they create a durable and diversified cash flow that supports sustainable growth. From a capital allocation perspective, we've taken a pragmatic approach to financing growth. In the past, we have used equity issuance as a bridge financing tool to fund our expansion while we evaluated longer-term cost-effective capital solutions. Given our current valuations, we recognize that issuing equity is clearly less attractive than ever. Our focus is on securing alternative financing options that allow us to scale in a non-dilutive manner that's much more sustainable. On the data center side, we are actively pursuing commercial mortgage financing to support our build-out. We are making progress on the front -- on this front and have finally received an attractive term sheet that we can move forward with and continue to explore the best terms before finalizing an agreement. For our Cloud Services business, we are also exploring vendor financing and leasing structures to optimize our GPU investments. Given the rapid evolution of AI hardware, we are structuring our GPU procurement strategy to balance growth and risk management. We are seeing compelling opportunities to scale and our approach ensures that we remain nimble while preserving capital efficiency. As our business continues to evolve, we are assessing the best structure to highlight the value of each of our business lines. We remain committed to making decisions that drive long-term value creation and position Bit Digital for growth. With that, I would like to open the line for some questions. As a note, we have Billy Krassakopoulos, who leads our data center business; and Ben Lamson, Head of Revenue for WhiteFiber on the line for some Q&A.

[Operator Instructions] And our first question will come from Mike Grondahl with Northland Securities.

First question is just on cloud services. Sam, I think you said the current run rate is $62 million, and you have a contracted customer that's coming on late March, have you named that customer, that gets you to $72 million run rate.

Yes, that's DNA funds.

Okay. DNA funds. Okay. And then I think there is another -- the 512 GPUs come on, I'll say, roughly July 1 for a $15 million ARR, that gets you to $87 million run rate. Is there anything else contracted to come on right now? I guess that's the first question. Anything else we should be aware of that's contracted that lifts that $87 million?

Well, we're already well past $100 million from a contracted basis, our WhiteFiber business includes both the GPU and data center business. So we -- so we will -- we go ahead.

Great. So let me ask this question then. You also talked about setting up this on-demand pool. And I think you said that as the potential for $25 million of ARR once it's contracted. And then there was another, I'll call it, GPU pool of 908 that has a $16 million ARR once it's contracted. Did I hear those right?

Yes. I would love to -- yes, I would love to pass you over to Ben, who heads our revenue for that division. Ben?

Yes. Thanks, Sam. So I want to be clear on this. On Demand is not contracted, but it commands a much higher price per hour than the contracted rates. So that $25 million number for the B200s that are coming online at the beginning of April, that is an annualized run rate based on those GPUs reaching full capacity on demand. Now we may choose to sell some or all of those on to reserved contracts 1, 2 or 3 years as that derisks things long term. But in the short term, we expect demand to be so high that we may choose to keep those on demand for a much higher price per hour.

Got it. And that $25 million and then the $16 million, do you think those are going to be, I don't know, generating revenue in the next quarter or 2 or 3? Like how should we think of them contributing or adding to the $87 million, a rough timeline?

Yes. We expect that the B200s to start generating revenue in April. As for the H200s, I want to be careful to give a date on those as those are part of some R&D around some product developments and technological developments, we'll be talking about later this year. So we're going to hold on to those for a little bit to run some R&D around some pretty cool stuff that we're building. And because of that, I don't want to give a hard date on when they're going to start generating revenue. But we'll be talking about some of that in our product road map later this year.

Great. Okay. And then just one more on the colocation or data center business, I don't think I heard what fourth quarter revenue was -- and then once we have that number, could we kind of walk through what's contracted and how that's ramping in '25, just like what we did for the cloud services business?

Yes. I think we have Billy on the line who leads our colocation data center division.

This is [ Krassakopoulos ]. Let me just jump in to that first one, Mike, it was 1.4 was the colocation revenue we recognized, but that was from the date of the acquisition. So like you can annualize that. And for Montreal 1, that would be the prevailing run rate until the new capacity comes online.

Got it. And then I think you guys talked about Cerebras coming online, and that's a $10 million run rate. Is there anything else we should be factoring in today?

I don't think we specified the run rate for that beyond just it being 5 megawatts, and you could sort of extrapolate the market rate there...

Got it. But just -- Cerebras the only one we should layer in for data center right now. Okay.

Well, Montreal 2, which would be a separate customer that we have not announced, which would also start in the middle of 2025. And then just beyond that, the planned deployments for the second half, which would come on later in the year.

And the next question will come from Nick Giles with B. Riley.

Guys, congrats on the progress thus far. First question. You mentioned a 100-megawatt site under LOI that would obviously be transformative. So I was hoping to get some additional color on when this capacity could come online? Is there any existing power infrastructure in place at the site? And then -- would there be a desire for a larger anchor tenant? Or how should we think about customer mix?

I'm happy to answer that question in a preliminary way, but Billy, would you like to give some more detailed color on that?

I think -- so we've identified sites in the United States that we are under LOI presently. There is currently 24 megawatts of power available at this location with a very easy path to 48, basically double within 60 to 90 days. And additionally, we have discussions with the utility provider to get us another 100 megawatts towards the end of 2025.

Got it. Billy, that's very helpful. And so maybe just a follow-up on that. How much of this would be included in your 156 megawatts of exclusive LOI capacity?

This is about 90% of that right now...

Okay. Great. That's very helpful. My next question was just on some of the GPU contracts. What kind of extension options are embedded in these -- some of these smaller 6- to 12-month contracts. I mean, should we think about all of these customers as very likely to further expand or could some rotate out and you would plan to backfill that capacity? I'm just trying to get a better sense for GPU utilization?

Absolutely. That's a great question for Ben.

Yes. I mean, we always -- with these customers, most of the time we're expecting them to both expand the amount of GPUs that they're using and the term. We've seen just broadly across the industry that customers prefer to start with, especially earlier-stage companies prefer to start with shorter deployments as they become comfortable with a new provider. And as we show improve in those -- that shorter deployment term, they end up renewing for a longer and/or larger term.

The next question comes from George Sutton with Craig Hallum.

And an impressive range of opportunities in front of you. So first, on the more traditional financing structures that you're working on, that seems to be the primary governor to further growth. Can you just give us a sense of where those discussions are currently?

With respect to the mortgage financing, I think you're referring to. We don't want to get too deep into it so that we don't jinx it, but we have a very attractive term sheet that we are currently planning on making the terms even better. But this is definitely a really strong start for us. And -- so we proceed this would be the most cost-effective way to tap into sources of financing without diluting our equity and which is correct because we couldn't do that with Bitcoin mining. You could easily do that with data center and real estate. So we look forward to proving to the markets that we can do this. And now we have finally going -- after going through a process, a term sheet that we really like. And once we finalize that, we will be able to discuss it publicly.

Got you. So relative to Boosteroid, I'm curious if you could just talk about what determines the deployment cadence there?

It's a mix of their GPU needs and also our own, I guess how we want to deploy the capital. I mean, there's only so much runway we have from the balance sheet. So we couldn't realistically do a $200 million deployment for them tomorrow. So it's measured, but we see a lot of opportunity to expand that with Boosteroid throughout the year, and we expect that to gradually increase.

Got you. And then just a question for Ben relative to WhiteFiber. Can you just give us a sense of the response that you've gotten from the rebranding? And it sounds like you could now do an on-demand platform. I wasn't aware that you could do that yet with your platform. Is that a little bit of an update on this call?

Yes. So let me clarify there. So we're going to be offering on-demand instances through a partner. So we do not yet have our on-demand platform live. That's -- I don't want to take -- I don't want to pontificate on when that will be, but we're looking definitely end of this year, early next year for our on-demand platform. So we'll be leveraging a third party to put those instances into an on-demand pool. And in terms of the response on the rebrand, it's been really positive. You think the way that we're positioning our offering is really well received. We've had more demand than we've had -- than we've been able to capture due to lack of supply being live, which is a champagne problem. So we're really excited about how the rest of this year is going to go.

And the next question will come from Kevin Dede with H.C. Wainwright.

Great to talk to you. Thanks for having me on. Let me echo Georgia's sentiments impressive array of opportunity. I think as I take a step back and look at it, though, Sam, it's hard to get arms around all the moving parts that you have, obviously, and I think that's what most people have been asking about. But beyond that, is the equipment, I know you referenced tariffs, but I guess what I'm wondering is how comfortable are you in the sourcing -- the infrastructure equipment you're going to need to support all these -- all this effort, number one. And number two, is there a contract recourse for Cerebras or for the eventual tenant of Montreal 2 to push back on the contract if you're not able to deliver on time?

Yes. So there are 3 things you mentioned there. There are a lot of moving parts of our story in our business. I think that sometimes that's a weakness and not a strength, because we have a lot of great things happening, and we're not exactly a one-trick pure play, Bitcoin miner. So we're not a one-trick pony. We're doing a lot of different good things on different fronts. So sometimes that story and that narrative gets a little bit lost. So we appreciate when analysts tease out every -- all the moving parts and provide a good report in terms of like the information of what's happening with respect to our businesses. So just wanted to comment on that initial statement you made. Your second question, is it more on the financing side or on the operational side in terms of the logistics potential disruption of tariffs. Just wanted to understand if you're looking at it from an equipment sourcing perspective or a financing sourcing perspective?

More on the equipment side, Honestly, Sam, there's -- I mean, obviously, all the demand that you're seeing is echoed throughout the industry throughout North America. So that's the question, right? Will you be able to source? Are you comfortable there? And then there's a recourse for your potential tenants.

Yes. Billy is going to give an answer to that. But just before he does, I do want to mention one of the main reasons we acquired Enovum and that team is because they've been doing this their entire careers and have established logistics supply systems and contingencies. So none of this is -- it's just not their first rodeo. And that's why it was really important for us, not just to acquire that Tier 3 data center, but really the team and the pipeline because they know what they're doing. And frankly, before we acquired them, we didn't know what we were doing in colocation services. And I salute the others who are trying it for the first time, but we just didn't have that courage and that's why we acquired that team because they know what they're doing. So anyway, with having for that, I would love to turn it to Billy to answer the question more deeply.

Thanks, Sam. So the equipment for the 2, 5- megawatt deployments that we have coming online has all been secured. Some of it has actually delivered. A large portion of it is in the production line right now to be delivered within the next 30 to 60 days. Equipment for remainder of 2025, we have purchase orders out deposits on equipment for the 20-or-so megawatts that we forecasted for the remainder of 2025. So we're very confident on that. And we're looking to place orders for deliveries on equipment in 2026 right now on equipment that's site agnostic. So stuff like generators, HVAC equipment, battery equipment, even though we have sites that are under exclusivity LOI that we plan on closing on very shortly. The equipment has already been preordered. And to answer your other question, the 2 clients that are coming online in the next quarter or early third are -- do not have recourse because of our confidence in delivering. I mean we have the equipment. We've secured the sites, the construction has begun. So there's no recourse for these clients if we do not meet the expected delivery date.

Thanks very much, Billy, appreciate it, Sam. On the Bitcoin mining side, you teased us a little bit with a 3 exahash target. I guess what I'm wondering is why you would even bother. So that's like one question. The other one is on Boosteroid. I understand that your -- if I have it correct, that is -- that you're leasing -- your leasing space in order to host them. And I'm wondering how you've seen that pricing dynamic change, say, I don't know, over the past 6 months that you've been working with them.

Sure. With respect to another 2 questions. The first question is why bother with our Bitcoin mining business. And the other question is the pricing dynamics as the passage of time has gone on with respect to our relationship with Boosteroid. Is that correct?

Yes, yes. No, I understand the involvement in Bitcoin mining. I do understand that, but -- why would you boost from 2.5 or 2.6 to 3. It just doesn't seem to correlate with all the other opportunities you have.

I mean, look, that's a good question. Our priority is the HPC business. There's no doubt. That's where our business is. We think -- we think 3 is somewhat of an arbitrary number. We do the -- we do think there's magic in the number 3. But we also don't want to give up on our Bitcoin mining business. We believe in the thesis of Bitcoin, especially these days with the institutionalization of Bitcoin at the stud level and increasingly with financial institutions. And so we are actually really excited in the future to talk a little bit more about digital assets and what we will be doing on that front. But for the time being, we believe that -- when you're mining Bitcoin, you're getting that at a discount to open markets, and that is an important thing to do. But we're not going to expand for our expansion sake. We're optimizing our fleet. We're making it lean. We're reducing costs and the methods that I talked about earlier today. With respect to Boosteroid, your question is kind of not clear to me. You're asking how has pricing changed during the course of time? Is that -- it wasn't clear. Sorry, Kevin, what he asked for?

I can just interject, Kevin. I mean the way that contract is structured, it's not like our traditional cloud services. It's structured effectively as an equipment lease, which basically means all the data center expenses are effectively pass-throughs or more specifically, we're not even build. So it's pretty agnostic, like say, directly pay the data center costs.

And the next question will come from Greg Pendy with Clear Street.

In for Brian Dobson. But just, I guess, a question. You're one of the only companies out there that I'm aware of that have deployed Ethereum staking yield strategy, but we're hearing from a lot of the other miners and you said earlier, you're going to remain committed to having some Bitcoin mining, at least for now. But how are you evaluating the yield strategies that some of the pure play Bitcoin miners are looking to do on their HODL balance? And would that shift are you still going to very comfortable with staking Ethereum to get yield off the coin?

Yes. Well, the yield on Ethereum is higher than Bitcoin. Bitcoin, there is no real yield. There is the yield on Ethereum. But we are -- we want to talk about the future of digital assets and our strategy on that in the medium-term future. I don't want to say too much about it right now. But we do believe that there is a very bright future with digital assets. We've seen, as mentioned earlier with Kevin that Bitcoin is now an accepted part of the DNA of the financial structure of the United States, they're also now creating a digital asset stockpile for the long tail of other coin, and I would imagine that would include Ethereum once that's announced. There are many technologies out there outside of Bitcoin with respect to Ethereum and other coins that are really worth looking at and also provide staking value and have their own merits. And that's something we could get into in the medium-term future. But for the time being, we're focused on WhiteFiber, which includes our co-location services with Billy and our Cloud business with Ben.

Okay. And then just one more real quick one. Just we all -- we're hearing from a lot of people that the demand for HPC is out there. But how important is the location specifically near the metro areas? Because there's a lot of other miners out there that are teasing the idea of pivoting maybe some other exahash to HPC. But just from your discussions and what you're seeing in the environment, can you get more specific on what's drawing people? Is it location? Is it near metro areas or even near specific climate areas that are drawing people?

Yes, Billy will certainly offer a more technical answer than mine, but clearly, there is business that we have won and there are certain requirements, particularly on the inference side that require low latency, which means you have to be in cities or near cities in order to provide for such clients. And so we've always taken that view for a long time. Our thesis is proving correct. We understand that other bitcoin miners are building in different places, and we salute them and we hope -- we wish them the best. But with respect to our strategy, we prefer to have our Tier 3 data centers in metropolitan areas or near them, particularly with respect to inference. And it's one of the reasons why, frankly, one, businesses like Cerebras. But Billy, do you have anything more to add to that? I'm sure you do, right? I'm not -- my technical tower is not as good as ours when it comes to this.

We look for sure, we prefer urban markets because latency is very important in these types of installations. But I mean, the markets that we're looking at price per square foot is not in the high to extreme ranges. We always look for locations, we try to balance out their locations.

And the next question will come from Joe Gomes with Noble Capital.

So I just wanted to circle back here to something you said, Sam, it sounds like given where the stock is, you're going to turn the ATM off. And last year, you guys raised over $240 million from that. And just kind of want to get your thoughts that you're comfortable that you can finance through other means that type of capital going forward here.

Yes. At these current levels, there is no desire to tap into the ATM. And frankly, we don't want to. We don't. We -- it hurts us as shareholder owners, which is where we are to dilute the shareholding, and we -- one of the reasons why we decided to pivot and expand and aggressively pursue the HPC business through cloud, through colocation services. Is that -- it is a much easier path to take on debt. But digital has had historically zero debt in the past. And the reason is because when you take on debt, you can't -- if you take on debt for Bitcoin mining, you can't model out your future cash flows unless you're using a different type of Excel spreadsheet that I'm used to. You just can't predict the future cash flows because you don't know where the price of Bitcoin is going to be. So when Bitcoin miners are taking on debt to finance their expansion. It's a wild gamble, and that didn't exactly turn out very good for many Bitcoin miners. So we never took on debt historically. And most Bitcoin miners, they realize the risk of taking on debt there, too, and they are aggressive with their ATM. We decided if we're going to go towards a business that's noncyclical and we can tap into non-dilutive sources of capital, in other words, not touch the ATM. And which is exactly what we're doing now, particularly on the data center side, for example, we already have a very attractive term sheet from a Canadian bank lender where the terms are incredible, and we don't have to use the ATM if that's the kind of financing terms that we can do. And it's easier to use those sources of financing because you can model future cash flows when it comes to colocation services. You can model future cash flows when it comes to the Cloud business. You can't do that with Bitcoin mining, which is why, as I said, we have zero debt because we were Bitcoin miners and now we have become something very different, and that's why we're ready to take on debt at very attractive terms in order to expand our businesses that are growing exponentially.

Equity financing is painful, Joe. But I mean, if you look at the platform we built in 2024, I mean we feel very good about that. And so I think it is very good use of funds and near term as a source of funds, we would likely opt to sell some digital assets before equity.

Thank you. And that does conclude the question-and-answer session. I'll now turn the conference back over to you.

Well, thank you very much, ladies and gentlemen. This concludes the call for Bit Digital. We look forward to the next quarterly call. Stay tuned for some more very interesting news. And that's it. Thank you so much. Have a great day.

And that does conclude today's conference. We do thank you for your participation. Have an excellent day.