Ladies and gentlemen, thank you for standing by. Good afternoon and welcome to the DMG Blockchain Solutions Q2 2025 Update Conference Call. Participants of this call are advised that the audio of this conference call is being broadcast live over the Internet and is also being recorded for playback purposes. A webcast replay of the call will be available on the company's website. Joining us today from DMG Blockchain Solutions is Sheldon Bennett, the company's Chief Executive Officer, and Steven Eliscu, Chief Operating Officer. During this call, management will be making forward-looking statements, including statements that address DMG Blockchain Solutions' expectations for future performance or operational results. Forward-looking statements involve risks and other factors that may cause actual results to differ materially from those statements.
For more information about these risks, please refer to the risk factors described in DMG Blockchain Solutions' most recently filed periodic reports and the company's recent press releases, particularly the cautionary statements within. The content of this call contains time-sensitive information that is accurate only as of today, May 22nd, 2025. Except as required by law, DMG Blockchain Solutions disclaims any obligation to publicly update or revise any information to reflect events or circumstances that occur after this call. It is now my pleasure to turn the call over to Sheldon and Steven. Sheldon.
Thank you, Chantelle. Good afternoon and thanks to everyone who has joined the call today. My name is Sheldon Bennett. I am the CEO and founder of DMG Blockchain Solutions. With a similar format as recent quarters, first, I will provide an overview of the company's achievements in the past quarter. I will then pass the call to Steven, who will review the company's performance. We will end the call with our Q&A session based on questions submitted to us prior to the call, as well as those using Zoom chat. Now to our highlights of recent achievements. First, our Core+ software and services strategy. As an update on Systemic Trust, we recently reported in a press release that Alvin Young has been appointed interim CEO on the departure of his predecessor.
As Alvin's prior role was Chief Revenue Officer, his primary goal remains customer acquisition and onboarding new clients by mid-calendar year, with a revenue ramp-up expected exiting the year. We continue to focus on partnering with crypto trading platforms, including Bosonic, with which we are also aiming to realize value from our investment. We have completed the testing phase of utilizing Fireblocks Wallets that incorporate DMG's Petra technology so that Bitcoin held by Systemic Trust can maintain its carbon-neutral status by being sent through Terra Pool for transaction processing. This is the last remaining component we have needed to enable DMG's carbon-neutral Bitcoin ecosystem. For Terra Pool, we are focused on partnerships and client acquisition. Our focus is to work with a limited number of larger mining entities looking to leverage the portion of their energy mix that is carbon-neutral to have the opportunity to increase their revenue.
The value proposition of Terra Pool becomes even greater when we can offer a suite of complementary products, including Helm, Reactor, and Explorer. Now, a few words on each. First, Helm, DMG's data center infrastructure management software, continues to be enhanced to be best-in-class software for maximizing the profitability of Bitcoin mining. We are utilizing Helm in-house to ensure the software scales to manage tens of thousands of air-cooled and direct liquid-cooled mining servers. We plan to incorporate AI features not only with respect to our code development, but also as how to maximize the productivity of the mining technicians on the floor. We plan to deploy Helm with beta customers by the end of this quarter and offer general availability in the September quarter.
For Terra Pool clients, we see Helm as an upgrade to their existing solutions, which are either general-purpose offerings built on antiquated software stacks or were built for in-house use only. Next, Reactor is our proprietary software for assuring the delivery of hash rate over the term of a hash rate contract and gives Terra Pool clients the option to sell hash rate and be paid upfront for delivering hash rate over the term of a contract, which is a useful treasury management tool that is unique for any pool to offer. We have relationships with several brokers of hash rate that can readily supply us with a market of buyers, including those who may pay a premium for carbon-neutral hash rate.
Optimization of Reactor continues to be a key capability we plan to offer concurrently with Helm, as we plan to integrate Reactor along with Terra Pool and Helm in a single seamless environment. Lastly, Explorer, which is the original software we acquired from Blockseer purchased back in 2018. We have pulled in our development of Explorer to this quarter to be a product initially to offer downloadable reports for the Bitcoin blockchain, a level of basic functionality that seems to have disappeared from any provider. We initially did this for ourselves but decided to offer it to the market. Longer term, we plan to add more advanced analytics capabilities to Explorer.
With our schedule for launching all of these products in the next few months, we will soon have a critical mass of products to be able to realize our vision of monetizing a carbon-neutral Bitcoin ecosystem where Terra Pool supplies carbon-neutral blocks that are in turn filled by Systemic Trust, a digital asset custodian subsidiary, with transactions from financial institutions that want the option to have a regulatory-compliant and carbon-neutral way of sending Bitcoin. This is a monumental achievement for DMG as we finally are now able to match the vision we have discussed with you for years with a comprehensive product offering that can enable us to achieve this goal. Beyond the next few months, we have no shortage of ideas for further building our software and service offerings.
More important for us than adding more new products is first to gain new customers to drive revenue while we incrementally add capabilities to existing products to build and strengthen our moat. Regarding our core data center infrastructure, first for AI, we purchased 2 megawatts of prefabricated data center infrastructure from a party that we had announced previously in an MOU to acquire a full 10 megawatts of infrastructure. While we did not disclose the purchase price, the fact that our unaudited Bitcoin balance declined in April should give an indication that we are serious about investing into AI computing. We are encouraged by our discussions with public sector entities and enterprises regarding potential colocation and direct offtake agreements.
There is tremendous appetite not only for AI compute in Canada, but to keep the data in Canada, the infrastructure, and as much of the IP within Canada in light of the uncomfortable geopolitical dialogue between the U.S. and Canadian heads of state. How quickly we can translate our discussion to agreements remains to be seen. We have dedicated staff who understand the public sector agencies focused on this effort, which is table stakes to participating in delivering sovereign AI. Last October, we signed a memorandum of understanding with the Malahat Nation to build out 30 megawatts of generative AI compute capacity split between both parties. We are now focused on executing agreements with the Malahat as well as more actively pursuing the financing of smaller projects ahead of a larger AI data center project financing, which could be several hundred million dollars just for infrastructure.
Next, for Bitcoin mining, with a portion of our hydro miners energized during the quarter, we realized 1.76 exahash with a fleet efficiency of 22.8 Joules per Terahash. We were negatively impacted by three days of curtailment in February, during which we operated with only about 15 megawatts of firm power at a Christina Lake facility. At the very beginning of May, we reached our short-term hash rate goal of 2.1 exahash. As disclosed in our recent press release, we are likely to operate under 2.1 exahash through the spring and summer months as we optimize our fleet to operate in higher ambient temperature environments. Regarding Bitcoin mining site expansion, we remain focused on lowering our cost of energy, which includes expanding the use of non-firm energy in Christina Lake and finding new sites with low-cost energy.
Regarding the site that DMG announced in May of 2023, we're still working towards concluding an agreement. We continue to evaluate locating Bitcoin miners at other locations. We will provide updates when we sign definitive agreements at these locations. Regarding our 3 exahash goal by the end of this calendar year, we are looking to achieve this goal with non-dilutive financing. We plan to achieve this goal by converting a portion of our Christina Lake facility into hydro mining. We are in the planning stage of making this conversion. We have somewhat shifted our stance on keeping Bitcoin mining at Christina Lake as we encourage that for the near term, non-firm power provides us with more competitive pricing. We are now looking into calendar 2026 for expansion to other sites, which would most likely be in Canada. Now for a summary of our strategy.
First, DMG's Core+ software and services. As we have stated previously in calendar 2025, we are focused on customer acquisition, a platform expansion for both Systemic Trust and Terra Pool. We need very limited headcount expansion to do this. As for both platforms, we are initially focused on onboarding relatively few large customers. We remain focused on onboarding first Systemic Trust clients by mid-year, along with Terra Pool throughout the rest of the year. Regarding platform expansion, we plan to add all the needed pieces that clients are demanding by mid-year, a more exhaustive platform for Systemic Trust, along with a critical mass of software for Terra Pool, including custodial wallets for clients by Systemic Trust, Helm data center infrastructure management, Reactor for Bitcoin treasury management, and Blockseer Explorer.
For DMG's core data center infrastructure, over the past quarter, we've had our heads down focused on client acquisition for our AI infrastructure. As we now physically own 2 megawatts of infrastructure, we see this as a catalyst for winning deals with a focus on public sector clients where the SCIF military-grade rating is a key selling feature. Given we are past the election of Canada's new Prime Minister, we're now working with the respective ministries that actually procure AI infrastructure and their new leadership. Even with a big portion of our attention on AI, Bitcoin mining remains foundational to our core strategy, and hence we are still planning to grow to 3 exahash by the end of the calendar year and to do this in a non-dilutive way. Now I'll hand it over to Steven to review the company's performance.
Thank you, Sheldon. I'm Steven Eliscu, DMG's COO. First, a few words about the company's overall position. In the March quarter, our cash short-term investments plus Bitcoin balance was $61.9 million, a decrease of 3% sequentially and up 42% year- over- year. We are utilizing more than half our Bitcoin balance as collateral for our Sygnum Bank loan facility, which we have utilized for capital purchases. Note that while our Sygnum loan balance was unchanged at $20 million in the March quarter versus the prior quarter, we previously guided that we intended to reduce our loan balance, a process we began in April and have continued in May. We believe we should lower our level of debt in light of what could be a lot more market volatility ahead.
While some may have already forgotten the Liberation Day exacerbated Bitcoin price pullback and subsequent recovery with Bitcoin now at all-time highs, we still remain cautious about the macro environment and will likely continue to gradually sell Bitcoin to pay down the loan at least for this quarter. For our mining operations, we reached our 2.1 exahash goal at the beginning of May. As we believe hydro mining and direct liquid-cooled AI servers are the future of high-performance computing, this deployment positions us to execute on that future. We have learned much about hydro infrastructure for which we have had issues, in part that delayed our deployment. By and large, we are pleased with the operation of the hydro miners themselves, which have run as expected, although we've yet to see how they run in the heat of summer.
We were also negatively impacted by increased energy rates in the March quarter related to seasonality. This was DMG's first full quarter of operating about half its fleet on non-firm power. As we're now subject to seasonality, we expect the winter months to be a high-cost period. Supporting this expectation, we have seen significant declines in our blended energy rates in March and April. Finally, as Sheldon detailed in his update on AI, we're making material progress towards securing offtake agreements with our focus on Canadian public sector and private enterprise clients. Additionally, as we purchase 2 megawatts of prefabricated data center infrastructure, this will add to our asset base that we will report in the June quarter. Now to review our financial results. In our March quarter, our revenue increased 9%, at $12.6 million from $11.6 million the prior quarter.
Mainly, self-mining revenue increased a similar percentage on 8% higher average realized hash rate. We mined 91 Bitcoin, down from 97 the prior quarter, as our hash rate increase was more than offset by a 14% in the network's Bitcoin per exahash generation. Our hosting revenue decreased 4% sequentially to $0.2 million in our March quarter. We expect hosting revenue to decline to near zero this calendar year as our existing customers retire their fleets and we utilize our capacity for self-mining and AI. Operating and maintenance costs increased 14% to $7.6 million from the prior quarter as we operated 8% more hash rate on slightly improved efficiency, 22.8 Joules per Terahash. The balance of the increase was caused by higher seasonal energy rates.
Note that our non-firm power is subject to curtailment, and as we disclose in our financial statements and a press release, we curtailed once in the March quarter for three consecutive days. We have had no subsequent curtailment events and believe our Christina Lake energy rates may end up slightly lower this calendar year than if we operated solely on firm power, and they should be sequentially down in the current quarter. Also note that the 15 megawatts of firm power we have will ultimately support our AI venture with the Malahat, and we are working with our utility to secure additional firm power for future AI growth. Our margin percentage on our revenue less operating and maintenance costs was 40% in the March quarter, down from 43% the prior quarter, mainly on higher energy rates. Consequently, our energy cost to mine a Bitcoin was about $55,000 U.S.
As we will likely utilize non-firm power for much of our energy mix going forward, we will experience more volatility in our mining costs than we've had historically. As a proxy for cash flow from our business, which assumes we are selling 100% of our generated Bitcoin, our earnings before other items, excluding depreciation, amortization, and stock-based comp, was $2.5 million, or 20% on a percentage basis in the March quarter, a decrease from $2.6 million and 22% the prior quarter, again due in large part due to higher energy rates. Our cash flow from operations was - $1 million in the March quarter as we sold $7.1 million less Bitcoin than we earned. Our cash balance declined by $3.5 million as we funded much of our capital needs through our cash balance, which allowed our Bitcoin balance to rise 13% from the prior quarter to 458 Bitcoin.
As we already reported, our Bitcoin balance subsequently declined in April to an unaudited amount of 351 Bitcoin. Investors should not expect our Bitcoin balance for the remainder of the current quarter to rise from April levels as we liquidate Bitcoin for operational expenses and paying down debt. Non-mining expenses, excluding depreciation, amortization, and stock-based comp, were $2.5 million in the March quarter, down 4% from the prior quarter of $2.6 million. In our 2025 financial year, given lower expected interest expense on our Signum loan for the balance of the year and cost control on headcount, we now expect non-mining expenses to rise only modestly this year. We're adding headcount to support program management, especially given what we need to accomplish on AI, business development for our software initiatives, and operations.
For at least the near term, these are targeted hires, helping us to ensure operational execution and make the needed transition from being a company that historically did not need to have marketing and sales talent. Depreciation expense of $4.3 million in the March quarter was flat from the prior quarter. As a percentage of revenue, our depreciation expense was 34%, which we believe is among the lowest of our peers and demonstrates our rapid improvement in capital efficiency as we have grown hash rate. We believe we are making good choices to balance the need to maximize cash generation while minimizing our total cost of mining by having purchased near leading-edge miners, which includes the T21 and S21 series of hydro miners at approximately $1 million a megawatt or less, while underclocking our legacy miners to extend their useful life.
Our earnings before other items was - $2.6 million in the March quarter, similar to the - $2.5 million the prior quarter. Our net income was - $3.3 million, or - $0.02 a share, versus - $3.1 million and - $0.02 a share of the prior quarter. Regarding our balance sheet, our cash short-term investments plus Bitcoin holdings decreased 3% to $61.9 million from the prior quarter and increased 42% from the prior year. The value of our property and equipment and long-term deposits decreased 8% to $55.9 million from the prior quarter as depreciation exceeded our capital additions. Accordingly, our total asset base decreased 6% to $129.5 million from the prior quarter and increased by 9% from the prior year. In the March quarter, we sold 39 Bitcoin, generating $5.1 million of cash.
Thus, we sold 43% of the Bitcoin amount mined versus the prior quarter of selling 81% of the Bitcoin mined. It is possible that with our Bitcoin sales this quarter, DMG will sell more than 100% of its mined Bitcoin in fiscal 2025. As a Treasury policy, investors should continue to expect us to sell most or all of the Bitcoin we mine and that our Bitcoin holdings should decline as a percentage of our total asset base over time. We do not expect to purchase Bitcoin on the open market as investors are free to do so independently of their investment in DMG. Regarding raising new capital, as for a future where AI could be a major component of our business, our capital raising would most likely be debt instruments tied to offtake and colocation contracts.
For Bitcoin mining expansion, we are most likely to continue to utilize Bitcoin-backed loans as well as other forms of debt financing. Note that as we look to our next phase of expansion towards 3 exahash, we now believe we can source near leading-edge 13-16 Joules per Terahash equipment later this year that should cost in the million dollar range, million dollar per megawatt range, significantly lower than the $2 million-$2.5 million per megawatt that we expected prior. I will now hand the call back to Sheldon to summarize our prepared comments, and we will answer questions. Sheldon.
Thank you, Steven. First, to reiterate our key results and outlook, we are positioning DMG to expand into AI in a meaningful way with a differentiated strategy based on unique, strong relationships with indigenous communities, offtake agreements, along with execution partners.
Systemic Trust is focused this year on acquiring clients and ramping revenue. Terra Pool is evolving into a complete platform with Helm, Reactor, and Explorer. We are focused this year on building partnerships and acquiring clients. DMG mined 91 Bitcoin in the March quarter on a hash rate of 1.76 exahash and a fleet efficiency of 22.8 Joules . Our hash rate goal is to grow to 3 exahash this calendar year, but to do so only in a non-dilutive way. Cash, short-term investments, and digital currencies at quarter end were $62 million, with total assets of $130 million. On a net income basis, we had a $0.02 per share net loss in the March quarter.
In addition to maximizing the cash generation of our Bitcoin mining operations, we are focused on realizing revenue from our AI and Core+ software and services initiatives that can help return us to profitability and drive shareholder value. In summary, we are proud of the accomplishments we have made this past quarter, demonstrating additional progress towards realizing a substantive AI revenue stream, positioning ourselves with a complete platform to generate revenue from our Core+ strategy based on Systemic Trust and Terra Pool, and having deployed hydro mining, which will allow us to remain competitive and grow as a Bitcoin miner. We continue to make real progress, and we are highly focused on our initiatives that can significantly grow our revenue and cash generation. As always, we appreciate your continued support. Now on to our Q&A.
As with past Q&As, Steven and I normally split the questions up a bit, and we take them both sent to us earlier over the last few days. If there are some that come in over the chat, we'll try and read through some of them and answer some of those questions as well. The first question that we have is, how long is it going to take to secure AI agreements? In particular, doesn't it take a really long time with government, given all the bureaucracy? This is a great question. As we all know, it can take a long time with any government, especially when we're new to government procurement, and especially when we're focused on SCIF-rated or military-grade data center infrastructure, which requires additional security clearances to participate in some of the tenders and procurement opportunities that the Canadian Department of National Defense has.
We are actively working through this process, but we do remain optimistic that we can secure agreements in the coming months. Months, I say not years, months, mainly because the Canadian government now has a sense of urgency to achieve AI self-sufficiency and sovereignty. Through our discussions with various agencies, we understand their sense of urgency. We understand their process, and they've been very helpful in helping us get through the types of qualifications, certifications, information they need to be able to have us as a potential vendor. We do know that working through all these issues with these various agencies can pay huge dividends down the road with what we see as large multi-year agreements that will command a premium. We are very much focused on this.
There is an alternative that we are trying to avoid, which is to go after sort of shorter-term agreements that are revenue right now, but they could sort of fall short in paying off the cost of the very high cost, the very high capital cost of the physical GPU hardware that we would need to purchase. We are really, in the first instance, really looking at more long-term, focus on government, prefer government, if possible, contracts to start our AI expansion. Hopefully, that answers that question. There is another one, and it is not much different. If the military just wanted to buy the prefab data center from you, would you just sell it, assuming a good markup? Yes, I mean, business is business. This is possible.
We could just sell it, mark it up, let them take it, and take the money and move on and continue with our strategy. But it's not really what we want to do. We do want to try and lock up an agency or multiple agencies so that we don't just sell them a piece of infrastructure and be done, but we sell them a service based on that infrastructure, whether it's us procuring GPUs or whether they colocate their equipment they want into our infrastructure. We see trying to get long-term contracts based on infrastructure that we can provide is a more practical business plan for us than just finding data center infrastructure and reselling it. It doesn't really meet our strategy of getting into implementing sovereign AI quickly in Canada if we just sell the equipment. It's kind of a roundabout answer.
I mean, yes, if the price is right, we'll sell it, but this is not what we're trying to do. This is not our strategy. Another question, why didn't you use debt financing to fund AI equipment purchase? It appears you liquidated Bitcoin to pay for 2 megawatts. Yes. So why we didn't use debt? The nature of a debt instrument is that it requires a revenue source to backstop it in general. If we want to get some money, normally that debt source will say, "How are you going to pay it back? What's your cash flow from the money you're borrowing to pay it back?" With this purchase of this 2 megawatts, there is no cash flow coming out of it. We bought the 2 megawatts to sort of jumpstart our movement into AI, and there was no cash flow behind it.
It was not really something we could go to a financier and say, "Hey, give us some money because we want to buy this." By the way, it has no cash generation. This is sort of why we looked at using our own funds to finance this 2 megawatts. I hope that answers it. Another question, how do you see the new Canadian government acting on its talk about sovereign AI? That is a great question, and especially the word talking. Lots of governments like to talk. We will see what the actions are. It is kind of important that Canada has appointed a federal minister of AI. That is Evan Solomon. They have done this ahead of any other G7 nation.
This sort of gives us an idea that Canada is really taking AI under the Carney leadership as a serious effort and something that they want to deploy quickly. As we have on our staff a former Canadian government employee, we are quite well connected with senior government officials and have reached out all the way up to Prime Minister Carney's office. We are finding out what they're saying, how much they're supporting this, what they're planning on doing. We're learning more about Canada's sovereign AI and its self-sufficiency initiatives. We're just kind of learning that Canada, at least under this Prime Minister, does not want to be dependent on the US or other countries for that, for AI sovereignty.
While we still sort of need to follow the protocols to be part of an actual AI procurement process, which I alluded to in an earlier question, we do believe that our work with the new government in power now is going quite well and that we have a very strong position to potentially win some sovereign AI projects. Another question, why do you seem to be more optimistic about long-term Bitcoin mining at your Christina Lake facility? Have you given up on finding other sites? Yes, we are more optimistic about longer-term Bitcoin mining in Christina Lake. Just to put that in a bit of context for people that do not watch the Christina Lake site that closely, as we have said, we have always been on firm power.
This is power that's delivered to us, and it's the same price for everybody in our territory that is a commercial, large commercial customer. We're no different than pulp mills and lumber mills and things like that, which is great. The power rate's always the same. We've always looked at and always thought about having what people call market power or wholesale power or what we call non-firm power, or some people call PPAs. Texas miners have PPAs. It's normally some type of contract where they can buy in the market. In our market, which is called Mid-C or Mid-Columbian, for the last few years, that price has been, in general, higher than what we would buy at our fixed price.
We have been very focused on not deviating and taking the bulk of our power or all of our power up until recently on our fixed-rate power. However, power prices have dropped. They started dropping, and the future price of power for our Mid-C area is looking very advantageous. We are quite happy to move over to taking more of our power. I think we are around 50% of it right now, and we may increase that on what we call non-firm power or wholesale power, just because, as Steve said in his comments, our overall blended cost is becoming lower. We think that even riding up the seasonality, we will end up getting a lower cost per kilowatt than if we were to stay solely on fixed power, as we have done in the past. That is one.
I'll give you another reason why we are is when we talk about our sort of move away from air-cooled to hydro, we have about 36 megawatts of power infrastructure that's inside our building. All of that is air-cooled, and our plan is to retrofit all of that air-cooled into hydro, which will be really a low cost because we won't have to change our power distribution and racking and all that. We'll just be adding in some of the piping and pumping and dry coolers. This allows us to increase our hash rate in the building. It allows us to no longer deal with the seasonality of the hot temperatures that we experience. Sometimes we've experienced some of these heat domes in the past that make air-cooled mining a bit more difficult in our location.
There are a couple of good things about Christina Lake that have kind of popped up. One is this lower overall cost, even with seasonality of market power. Two is just the very low cost for us to be able to convert our air-cooled into hydro-cooled. That is great. There was a second part of this question, "Have you given up on finding other sites?" The answer is no. We are always actively looking at pockets of low-cost energy where we can make incremental additions to our mining capacity. We will continue to give updates on that. In some ways, it is really easy to find power on natural gas, Alberta, Texas, sort of higher carbon areas. It is a little bit harder to find it if we want to stay as a carbon-neutral company.
Not impossible, but it takes a little bit more work, and we got to do a little bit more due diligence, and it takes a bit longer to get those things done where it makes a material difference and is worth doing. Probably a lot on that question. Next question, what happened to Lawrence Truong? No, sorry. What happened as to why Lawrence Truong left as CEO of Systemic Trust? So another good question. I guess the simplest thing to say is that Lawrence decided to pursue other opportunities. What we can say about this is that we're still confident Systemic Trust will be a major digital asset custodian in Canada and that Alvin Young is highly capable as our interim CEO.
Systemic Trust has already moved beyond the startup stages of building a team, building out a platform, and successfully working with the regulator, all things that Lawrence managed prior to his departure. During Lawrence's tenure in the background, Alvin was focused on customer development, defining what is needed to further build out the platform, which he continues to be executing on with the DMG software team helping along the way. We anticipate that Systemic Trust going forward will not skip a beat. I think it's going to add customers. I've been in a lot of the customer meetings with Alvin. It's kind of a small world in the world of crypto. I know a lot of the same people he knows. I don't think he's going to have too many difficulties onboarding and getting people going and working on the trust. I think things will be fine.
Maybe since I've been rambling along here for a little bit, I'll let Steven take over a few questions.
Yeah, thanks, Sheldon. One of the ones we just got in is just asking us about us utilizing our prefab data center with liquid cooling. I think it's important for investors to understand just how we're thinking about deploying AI. Initially, the prefabricated data center that we have purchased and plan to purchase more of is an air-cooled asset. Its SCIF rating makes it very—which is what the military wants. It is very hard for third parties to come and try to listen in to what's going on inside the physical structure. If we were to modify it in any way to, let's say, support direct liquid cooling, we would probably impair the ability to maintain that rating.
That initial asset should be looked at as an air-cooled-only asset. Especially as we're gaining experience with direct liquid cooling, anything that we build greenfield going forward would be based on direct liquid cooling technology. We think that's the future. We think data centers, the way they're built, especially as a lot of them, especially the ones we think in Canada, are going to be for inference-type workloads, going to be relatively modest in size, actually quite physically small as we look to the next generation coming out next year with 600 kilowatt racks. The future data centers are going to look very different, and we're going to intercept that future. For now, the prefabricated data center is air-cooled. It's most likely going to stay air-cooled as we go forward. That was a technology question. Let's kind of switch gears here.
One of the questions that came in ahead of time was, how does DMG compare to some of the other mining stocks like HIVE and MARA? What strategies going forward could lead us to all-time highs? We appreciate that question. We know that on an EV to hash rate basis, we've been valued less than our larger peers, even as we've kept up more recently with hash rate growth and fleet efficiency. We've maintained a reasonable huddle. We've limited dilution. We know we've been making these investments in software and services and AI more recently that have yet to show results. Our focus is to translate these investments into meaningful results as we believe that can be the catalyst for the stock.
For Bitcoin mining, we are going to continue to talk about it, but in a more subdued way because it does continue to generate cash for the business. We need it at least in the short term. Next question is, DMG needs to learn how to market itself as a publicly traded company. Investor relations needs to be much higher on DMG's priority list. Is there a plan to aggressively promote and market DMG's transformation into an AI technology company? We really appreciate the criticism and really appreciate the question. We've been hesitant over the past year to aggressively market ourselves regarding Core+ software and services, and more recently, AI, as we're also cognizant of pushing a story too early and then having investors lose interest when it takes longer than expected, which, given what we're trying to do, these things do take time.
We have held back, but we are now in a much different position than we were even six months ago. We've added two top-tier sell-side analysts who have increased our visibility with investors. The person who asked the question is right that we do broadly need to step up our IR, and it hasn't been lost on Sheldon and myself. Just going forward, we do plan to invest more. We are going to be evaluating over the summer, utilizing outside firms versus bringing in an in-house dedicated resource. We'll also look to other avenues to get our social media presence also to step that up as well. Just related on asking about hydro, just can we use the same systems as we use in AI as Bitcoin mining? The answer is no. This is important for investors to understand in general.
When Bitcoin miners talk about getting into AI and they talk about direct liquid cooling, it's not the same direct liquid cooling. It's much more expensive. An example is the water, if it's a closed-loop system like we're currently using, the water that recirculates probably at least some times of the year needs to be actively refrigerated before it reenters and is used to cool the GPUs. That's really expensive. There's also a lot of redundancy. Also, if there's a leak of the water, it's catastrophic in the case of the AI servers, just given the amount of cost and future systems are probably upwards of $100 million per megawatt, that these systems are going to be built very differently. The experience we're gaining is helpful, but the actual implementations are going to be very different, and we're very cognizant of that.
Just also asking about would we start with HPC with a non-government customer before scaling to potential government customers because clearly there is a time frame associated, as Sheldon talked about, to get government customers. Yes. I mean, we are in parallel. We are talking to enterprise, potential private enterprise clients, and we will be looking at that as an option as well. Just also, we talked about getting to 3 exahash. If we are not going to raise equity and we want to reduce our debt, then how are we going to be able to raise the money to be able to buy new miners? That is a good question. I mean, the view is that we take down the level of debt substantially, at least through this quarter. We evaluate where we are at.
Really, we may look towards later in the year, depending on the market conditions, as we are really looking to some of the next generation equipment that is not available now but will be available later in the year at very cost-effective price points, really looking more to that time frame. At that time frame, market conditions may very well support us growing our debt levels again as long as we feel they are manageable. We may utilize a Bitcoin back loan. We may utilize other types of debt financing. We are going to be very careful, and it is all dependent on the market conditions and what kind of ROI we think we can get. Let's see. Sheldon, are there any other questions that you wanted to take as we are coming up on the hour?
No, I am just looking at the chat a little bit.
What is the current outstanding balance of the loan?
We can't disclose that until the end of the quarter.
I can't. I mean, as we said, it's coming down. It's in our financials where it was at the end of the quarter, but we're kind of doing what we're saying we're doing, and we're paying that down. Has Systemic Trust added any clients other than DMG? Again, companies have the right to have privacy. Some of them will allow us to press release and say who they are. Some will not. As we're just starting client acquisition, we don't really want to say anything until we get the okay, and it will come out in a press release. I'm not too sure what we can say on that. I think you're just going to have to wait a little bit for some public announcements.
Hopefully, someone will come out soon. Let's see what else is in there. Is there anything else? Buying more shares. I bought a lot of shares in DMG. I guess you can talk about yourself rather than Steven, but I'm probably the happiest to buy more shares of DMG. It's actually getting harder because there's a whole bunch of blackout periods that we have now in our governance that make it a bit harder for us to buy. I bought some not too long ago, and probably will be happy to buy more. I hopefully kind of answered a lot of this HPC.
There's one more I'll answer here, which is just the strategic rationale for buying the prefabricated data center. I think we've indicated this.
We think we need to have some amount of physical infrastructure that we own that's in flight that gives us towards we've said we may locate it on our Christina Lake property. We have other options as well. For us, in terms of the credibility that we establish when we have these conversations, actually owning the infrastructure and talking about time frames that we can deploy it, we appear to be much more credible. Hence, because we're in a race with others to deploy sovereign AI in Canada, this we believe just gives us another piece that gives us that edge in terms of being able to deliver it first and win this race. We see there's a limited window here for winning sovereign AI.
The prefabricated data center is a bridge to be able to versus these much more ambitious 15 megawatt direct liquid cooled next generation data centers. That is out in time, as we have said. It is going to take a while for us to build that. That in turn, also if we can win initial some smaller agreements with the prefabricated data center, that will give us the momentum to get us the offtake agreements that in turn help us to fund these larger data centers. Hopefully, I answered a couple of questions in that. I think with that, we are probably nearing time. Sheldon, why do I not just pass it back to you to finish our call?
No, thanks, Steven. I think that is sort of a good explanation.
We have been on a few calls where with different people responsible for AI asking about blockchain, why are we talking to a blockchain company? We have an answer for that. Why they're talking about a blockchain company is because there is a component on the software layers why blockchain is important to AI. That is a different topic. Let us now officially end our Q&A. As a reminder, both Steve and I will be participating in the upcoming Bitcoin 2025 in Las Vegas next week. Please reach out to us if you would like to have an in-person meeting with both of us or either of us. Therefore, we thank everyone for attending, and our call is now over.