Core Scientific Inc

NASDAQ:CORZ

Earnings Call Transcript 2024-Q2

[Operator]

Greetings, and welcome to the Core Scientific Second Quarter Fiscal Year 2024 Earnings Conference Call. At this time, all participants are in a listen-only mode. A brief question-and-answer session will follow a formal presentation. [Operator Instructions] As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Steve Gitlin, Senior Vice President, Investor Relations. Thank you. You may begin.

[Steven Gitlin, executive]

Good afternoon, ladies and gentlemen, and welcome to Core Scientific Second Quarter Fiscal Year 2024 Earnings Call. This is Steven Gitlin, Senior Vice President of Investor Relations for Core Scientific. At this time, all participants are in a listen-only mode. We will conduct a question-and-answer session after management's remarks. As a reminder, this conference is being recorded for replay purposes.



Before we begin, please note that on this call, certain information presented contains forward-looking statements within the meaning of the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements include, without limitation, any statement other than historical or current facts that predict or indicate future events or trends, forecast, performance or achievements and may contain words such as believe, anticipate, expect, estimate, intend, project, plan or words or phrases with similar meaning. Forward-looking statements are based on current expectations, forecasts and assumptions that involve risks and uncertainties that may cause actual results to differ materially.



For further information on these risks and uncertainties, we encourage you to review the risk factors discussed in the company's annual report on Form 10-K filed with the Securities and Exchange Commission and the special note regarding forward-looking statements contained in the company's current report on Form 8-K filed today and the earnings release and slide presentation contained therein.



Today's presentation is available on our website at corescientific.com in the Events and Presentations section. The content of this conference call contains information that is accurate only as of today, August 7, 2024. The company undertakes no obligation to update statements made today to reflect events or circumstances occurring after today.



Joining me today from Core Scientific are Chief Executive Officer, Mr. Adam Sullivan; and Chief Financial Officer, Ms. Denise Sterling. We will now begin with remarks from Adam Sullivan.

[Adam Sullivan, executive]

Thanks, Steve. Today, I'll summarize the incredible progress we've made this year before quickly highlighting our second quarter financial performance. Denise will then review our second quarter financials before I discuss what distinguishes Core Scientific from other companies in our industry and provide a look ahead at the key catalysts and milestones that will demonstrate continued progress towards achieving our goals. We will then take your questions.



We have achieved exciting milestones this year, illustrated on Slide 4. We announced a contract with CoreWeave to lease a 16-megawatt data center in Austin for HPC hosting, and we delivered that data center more than 30 days ahead of schedule and began generating revenue from it in the second quarter. We successfully executed our plans for the April halving, maintaining strong operational performance and favorable cash cost to mine. We have now signed HPC hosting contracts with CoreWeave for a total of 382 megawatts in aggregate total potential revenue of $6.7 billion over the 12 years beginning in 2025 and 2026.



We completed 72 megawatts of partially built infrastructure at our Denton, Texas site, bringing our total to about 830 megawatts of operational infrastructure highlighted on Slide 5. We began build-out a partially completed 100-megawatt facility in Pecos, Texas. Stock price appreciation triggered mandatory conversion of secured convertible notes, eliminating $260 million in debt from our balance sheet. We signed an agreement with Block to procure 15x of their new 3-nanometer ASIC chip to refresh and expand our self mining fleet beyond 2024. Stock price appreciation triggered the ability to exercise our tranche to warrants, and we are now preparing some of our sites for HBC hosting, migrating miners from HPC designated sites and redeploying them to bitcoin mining sites.



Our ability to achieve all of this is a testament to the dedication and tenacity of our best-in-class team. We are executing on our strategy and remain focused on making continued progress while creating a new category in the digital infrastructure industry. While we expand our hosting business for HPC, we remain one of the largest and most efficient bitcoin mining companies as illustrated on Slide 6. Our Bitcoin mining business continues to produce operating cash flow and will become even stronger as we migrate our miners to dedicated sites with more favorable power pricing.



Slide 7 highlights our strategic, operational and financial achievements in the second quarter. Note that for the first time, we are now reporting HPC hosting as a separate segment in our financial statements to provide greater visibility to this growth driver for our business. We earned 1,680 bitcoin in the second quarter and generated total revenue of $141 million, including $5.5 million in HPC hosting revenue from our Austin data center. Gross profit of $39 million increased by 5%, while operating income of $7 million declined by 31%.



Our $805 million net loss was mainly driven by the significant quarter-over-quarter appreciation in share and warrant prices that required us to make mark-to-market adjustments to the value of our equity. Denise will explain this in more detail shortly. Adjusted EBITDA of $46 million in the second quarter increased by 2% over the prior year and reflects continued strong cash generation from our core business. Our team has worked exceptionally hard to deliver these strong results. And while we are proud of what we have done, we are just getting started. Now I'd like to turn the call over to Denise Sterling, our CFO, for more details on our financial performance and positioning. Denise?

[Denise Sterling, executive]

Thank you, Adam. We are pleased with our second quarter operational execution and financial performance, particularly given the halving. As Adam mentioned a moment ago, total second quarter revenue was $141 million, consisting of $111 million in digital asset self-mining revenue, $25 million from digital asset hosted mining and $5.5 million from HPC hosting.



Digital asset self-mining gross profit for the quarter was $30.7 million. Self mining revenue increased by $13.7 million or 14% year-over-year, primarily from a 134% increase in the price of bitcoin and an increase of 28% in our self-mining hash rate, which was due to the deployment of approximately 19,000 additional new generation self-mining units.



Although our self-mining hash rate increased 28%, the April 2024 halving and a 68% increase in the network hash rate led to a 52% decrease in the number of bitcoin earned during the quarter. Digital asset self mining cost of revenue increased by $13.2 million for the fiscal second quarter of 2024. The increase was primarily driven by an increase in depreciation expense resulting from the deployment of our new self-mining units, an increase in payroll and benefits costs associated with merit and market adjustments made during the quarter and higher stock-based compensation.



Segment gross margin was 28% in the quarter. HPC hosting revenue of $5.5 million exceeded HPC hosting cost of revenue of $4.9 million for the fiscal second quarter of 2024 by $0.6 million for a GAAP gross margin of 11%. Revenue at the Austin site ramped up over the course of the quarter based on the original delivery schedule. While GAAP lease expenses were accelerated as a result of delivering the data center capacity more than 30 days ahead of schedule.



As a reminder, the HPC hosting costs for the Austin data center consists primarily of lease and power expenses. Our power expenses are a direct pass-through to our client with no added margin. Overall, this resulted in a lower-than-anticipated GAAP margin in the initial quarter of operations. We anticipate the Austin data center margin to improve over time to between 35% and 40%, excluding power pass-through as the additional lease expense associated with the acceleration of HPC capacity will be fully recognized by the first half of 2025 resulting in more normalized margins. It is important to note that the terms of our hosting contract for our Austin data center, which is leased from a third party, very significantly from those of our larger HPC hosting contracts, where we are modifying our owned infrastructure and therefore, do not incur lease expenses.



A summary of our segment economics can be found on Slide 8. Gross margins in the quarter were 28%, 30% and 11%, respectively, for digital asset self-mining, digital asset hosting and HPC hosting. Our power costs were favorable in the quarter, declining to $0.0402 from $0.0425 per kilowatt hour for the same period in the prior year. Operating expenses for the fiscal second quarter of 2024 totaled $31.4 million as compared to $27.1 million for the same period in the prior year. The increase of $4.3 million was primarily attributable to $7.2 million increase in personnel and related expenses, and $4.6 million of HPC hosting segment start-up costs incurred during the current period, partially offset by lower stock-based compensation of $6.5 million due to cancellations and forfeitures of equity-based awards and a decrease of $1.6 million in bankruptcy advisory cost.



Net loss for the fiscal second quarter of 2024 was $804.9 million as compared to a net loss of $9.3 million for the same period in the prior year. The increase in net loss of $795.6 million was primarily due to a net $796 million noncash mark-to-market adjustments to our warrant and contingent value right liabilities required as a result of significant quarter-over-quarter increase in the value of our equity.



Also contributing to the increase in net loss was a $14.8 million increase in interest expense, partially offset by an $18.5 million decrease in reorganization items net with no comparable activity for the same period in fiscal 2024 due to the company's emergence from bankruptcy during the first quarter.



Non-GAAP adjusted EBITDA for the second fiscal quarter of 2024 was $46 million or 33% of revenue, a year-over-year increase of $1 million that included several offsetting adjustments. Our power contracts vary in price in terms. As I mentioned previously, our fleet wide power cost averaged $0.0402 per kilowatt hour in the fiscal second quarter. We now expect average cost in 2024 to be between $0.042 and $0.044 per kilowatt hour.



At the end of the second quarter, our self mined to hosted mining mix was 79% to 21%, respectively. As our hosted mining contracts sunset this year and we expand our self mining fleet, we expect our hosted mining percentage to decline. As of June 30, 2024, we operated approximately 163,500 miners in our self mining fleet. We have included the composition of our self mining fleet, including the model mix and efficiency on Slide 9.



And now I'd like to discuss actions we are taking to strengthen our balance sheet. We've enhanced our liquidity in the quarter, ending with $96 million in cash and cash equivalents, up from $50 million at the end of 2023. As illustrated on Slide 10, during the second quarter, our debt decreased by an additional $56 million to $552 million. This decrease was primarily driven by a reduction of $26.4 million in secured convertible notes that were voluntarily converted and the payment of approximately $19.2 million for a mechanics lien related to the additional 72 megawatts of infrastructure completed at our Denton, Texas data center. And shortly after the end of the second quarter, stock price appreciation triggered the mandatory conversion of our secured convertible notes, resulting in the equitization of the remaining $233.6 million, which resulted in a further decrease in our debt to $318 million.



The average interest rate of our remaining debt after mandatory conversion of the secured convertible notes is approximately 12%. We continue to evaluate potential options that would further reduce our debt service and strengthen our balance sheet.



Now I'll turn to our CapEx plans. We plan to acquire an additional 10,000 to 15,000 bitcoin miners in 2024 to continue refreshing our self-mining fleet and to achieve our 21.8 exahash self-mining hash rate goal. We anticipate the CapEx for this purchase would range between $13 million and $19 million, which is included in our 2024 CapEx plan. We also anticipate investing $13 million on the 100 megawatt expansion of our Pecos, Texas site, which is also included in our CapEx plan. CapEx associated with our contracted 382 megawatts of infrastructure to host high-performance computing is funded by our existing HPC client. With the exception of the portion, we will repay through CapEx credits and therefore, is not included in our plan.



I'll now turn to a review of our mining economics summarized on Slide 11. Our direct cash cost to self line of Bitcoin in the second quarter was $29,879. This consists of power cost of $24,533 and direct cash-based facilities operations cost of $5,346, allocated based on the 79% of our fleet dedicated to self-mining and divided by total Bitcoin self-mine in the second quarter of $1,680. Another way to evaluate our mining cost is by calculating the cash-based hash cost of these same items, which represents the same cost expressed as a cost per terahash per day. Our total cash-based cash cost in the second quarter was $0.03 per terahash.



On Slide 12, we provide an updated illustrative financial view of our HPC contracts with our existing clients. Based on the total 382 megawatts now contracted for HPC hosting, we project aggregate total revenue over the 12-year contracts of approximately $6.7 billion with a profit margin of up to 80%. Power and utilities costs are direct pass-through to our clients and will be grossed up and included in both revenue and cost of revenue.



As a reminder, our client is paying for all CapEx, and we are responsible for repaying $1.5 million per HPC megawatt through a CapEx credit once the data center is energized. Slide 13 illustrates the flow of CapEx related to the HPC data center build-out cost and their impact to our financial statements over time. Funds are deposited by CoreWeave in a joint escrow account to pay for build-out costs as they are incurred. During construction, expenses incurred to develop the HPC data centers will be recorded in construction in progress. Once the data center is placed into service, the cost will be reclassified to property, plant and equipment and begin to depreciate.



And now I'd like to share a few modeling details. We continue to model a statutory effective tax rate of 23% for 2024. We also retain more than $300 million in net operating loss carryforwards, which will reduce future cash taxes. Our share count was approximately 183 million shares as of June 30, 2024, and approximately 258 million shares as of August 2, 2024.



Note that approximately 40 million shares associated with the mandatory conversion of our convertible notes have now been issued along with approximately 35 million shares from exercise warrants. Our updated share count will be reflected in our third quarter filings. And now I'll turn the call back to Adam. Adam?

[Denise Sterling, executive]

Thanks, Denise. Core Scientific is a leader in digital infrastructure for bitcoin mining and HPC hosting. We now have approximately 830 megawatts of operational infrastructure within our total contracted power of 1,200 megawatts. This contracted power represents power that is ready for us to use when we complete the associated data center infrastructure.



Our significant portfolio of contracted power is only part of what differentiates Core Scientific and positions us to become one of the largest publicly traded data center companies in the United States. Core Scientific is creating a new category in digital infrastructure, and we are executing on our vision for the future of application-specific data centers.



We understand the differing requirements between GPUs and bitcoin miners because our team has considerable experience in managing both as well as deep industry expertise across the design, development and operation of data centers. Now that we have the potential to generate $6.7 billion in cumulative revenue over the next 12 years from HPC hosting, viewing us as a data center company is more aligned with the future of our business.



As we convert a portion of our digital infrastructure to HPC hosting, we expect more investors to value this growing part of our business using data center multiples. We'll value our bitcoin mining business using mining industry multiples. We believe that this approach will yield a more accurate representation of the true value we are creating.



Now I'd like to discuss the main value creation catalysts for the remainder of this year, described on Slide 14. Executing on our remaining option agreement, expanding our power portfolio and diversifying our HPC hosting client base. First, yesterday, we announced another HPC hosting option exercise under our contract with CoreWeave for an additional 112 megawatts, increasing our total contracted infrastructure for HPC hosting to 382 megawatts.



We are in discussion with CoreWeave and a number of other potential clients to contract the remaining 118 megawatts of our infrastructure available to support additional HPC hosting. Our clients' option for additional infrastructure extends through the beginning of September, providing the ability to contract additional capacity at the original contract terms.



Second, we are also building a pipeline of opportunities for new assets to expand our infrastructure capacity. These opportunities range from 25 megawatts to several hundred megawatts, including brownfield and greenfield projects and extend over a large geographic area. We have developed specific site selection criteria to guide our efforts and optimize our resources as we seek to expand our site portfolio for HPC hosting and bitcoin mining.



Third, we seek to diversify our client base and are working to achieve that either with the remaining 118 megawatts or with new asset opportunities. Of course, delivering the completed infrastructure to our client is what will trigger revenue generation, and we are now at the early stage of the process, modifying our sites to accommodate HPC hosting involves developing detailed designs, performing selective demolition and construction, installing new equipment and supporting infrastructure and delivering the completed site to our client. We will provide updates over time as we progress through this modification process.



We remain on track to deliver the associated HPC hosting infrastructure for the first 200 megawatts in the first half of 2025, the next 70 megawatts in the second half of 2025 and deliver the latest 112 megawatts in the first half of 2026 as illustrated on Slide 15.



I want to make it clear how integral our team and our experience is to the process of converting our sites to HPC hosting. Without a deep understanding of the entire process, we would not have been able to create this opportunity to expand our hosting business. This is an important differentiator favoring Core Scientific.



Shifting now to our bitcoin mining business. April marked the latest halving event in which bitcoin block rewards declined by 50%, which should have resulted in a 100% increase in our cash cost to mine, assuming all other factors held constant. However, our cash cost to mine in the second quarter was about $29,900, an increase of approximately 60% over the first quarter.



Lower power prices in the second quarter mitigated the effects of the having asset carefully managing the operations of our self mining fleet relative to hash and power prices based on the detailed plans we developed well before that happened. As Denise mentioned, we plan to procure 10,000 to 15,000 traditional miners this year to achieve our fleet refresh and cash wrinkles. Beyond 2024, we expect to begin acquiring the new 3-nanometer block ASIC chips when they are available in deploying them in our own chassis to our dedicated bitcoin mining sites to boost our hash rate and efficiency.



When deployed and combined with our proprietary firmware, fleet management and energy management software, we expect to have greater operational control over these miners to respond more dynamically to changes in hash price, power price and environmental conditions. We expect the new block chips, combined with our proprietary tech stack to improve our operating efficiency and profitability. We expect the majority of our bitcoin hosting contracts to Sunset this year, making way for us to focus on self-mining and HPC hosting.



To summarize our guidance for 2024 highlighted on Slide 16. We expect 21.8x as of self mining hash rate. Approximately 800 megawatts of total operational infrastructure, understanding that this number will vary significantly as we temporarily power down sites for HPC conversion and power upsides for bitcoin mining. 16 megawatts of revenue-generating HPC hosting infrastructure and average Bitcoin mining fleet power price of [ $0.42 to $0.44 ] per kilowatt hour.



We have been deliberate in our planning and communication, having established our strategy, described its evolution and now executing against it with significant HPC hosting contracts and progress in the conversion process. There is much work to do, but we are the best equipped team in our industry to succeed, and we are confident that we are best positioned to translate that work into shareholder value. Thank you for your engagement and attention, and thank you to all our clients, industry partners and all our teammates for your ongoing efforts and support. We will now take your questions.

[Operator]

[Operator Instructions] Our first question comes from Greg Lewis with BTIG.

[Gregory Lewis, analyst]

Adam, clearly, you continue to execute on transitioning your existing power to HPC. So congrats on that. I was hoping to get a little bit more color. Obviously, there's inherent value in the data center contracts. Kind of curious, as we think about opportunities ahead to kind of site and source new locations, whether for HPC or traditional bitcoin mining, how you're thinking about funding that? And is there an opportunity to take advantage of these 12-year contracts with CoreWeave to help you kind of finance further growth.

[Adam Sullivan, executive]

Yes. Thanks, Greg. On the capital process, or capital structure process and how we're thinking about that going forward, the nice part, as you mentioned, related to our contracts, it does open up a number of different doors for us. Our credit story is rapidly improving as we move towards 2025. And that's not only exciting to us, but also exciting to our shareholders.



And I think something that's going to open the door on is our ability to continue to access different parts of the capital markets that we haven't been able to necessarily access more specifically as just a bitcoin mining company.



So right now, our focus is on the site selection process, targeting sites anywhere ranging from 25 megawatts to a few hundred megawatts. And really, we're targeting sites that aren't necessarily going through the traditional brokered processes. And that will allow us to go after sites that are less expensive, but very attractive to us. And so right now, our plan is to continue to fund purchases out of either our operational cash flow, cash on balance sheet.



And there are other methods for us to look at right now in the capital markets that could provide us a lot more optionality in terms of going out and acquiring sites. So we have a number of sites in our target range right now, and we're definitely hopeful as we mentioned, to exercise on some of those over the course of 2024.

[Gregory Lewis, analyst]

And then just one more for me on the -- you kind of guide to that fleet average price of $0.42 to $0.44. Just kind of curious, I don't believe we have been fixing any power. So as we think about that number, is that just based on forward curves or kind of like any kind of thoughts around how we're kind of focused in on that. That's a pretty narrow attractive power cost. So just kind of any color around that.

[Adam Sullivan, executive]

Absolutely. The way we're looking at the rest of the remainder of this year is based on not only our fleet efficiency and machines that we'll be acquiring and placing at each of our sites. But that also takes into account the forward curve at certain facilities. When you look at our site portfolio today, we are operating in a number of regulated markets. And so we have much greater insight into where those power prices will come in at. And then in the deregulated markets such as Texas, those are based on our current efficiency based on where the forward curve is, we feel very confident in our ability to execute on that $0.42 to $0.44.

[Operator]

Our next question comes from Lucas Pipes with B. Riley Securities.

[Lucas Pipes, analyst]

Adam and team, great job on the execution this year. My first question is also on the expansion of the HPC site portfolio. And I wondered, Adam, if you could maybe comment just on the general kind of competitive dynamics in that market today after what you have done. Who is competing with you in this -- in the site selection. And in terms of like the next couple of sites, how quickly do you think you might be able to execute something.

[Adam Sullivan, executive]

Yes. Thanks, Lucas. Thank you for joining. As we go through processes on some of the larger sites and some of the broker processes, the folks that we're competing with are the likes of Equinix, Digital Realty, et cetera. And so those processes can get expensive at certain times, and that's really why we're focused on, I would call it, a different set of assets. What we're focused on are both greenfield and brownfield opportunities. For certain data center companies, they are not interested in touching brownfield opportunities.



We have experience converting brownfield opportunities not only from a data center perspective, but also a bitcoin mining perspective. You look at our portfolio today, we have 3 sites in there that are brownfield conversions. So we have the experience, and we believe we are carving out a niche in the site selection process that will provide our company very interesting opportunities.



And those opportunities aren't only great for us, but these are opportunities that are very interesting to the potential clients that we're speaking with.

[Lucas Pipes, analyst]

My second question is regarding leverage. Adam, Denise, how do you think about kind of the proper leverage ratio of this business after what happened? You touched on it a little bit earlier, but I would appreciate how you think about that? And then is there any sense of urgency to make changes to the balance sheet as it stands today? I would appreciate your thoughts.

[Adam Sullivan, executive]

Absolutely. Our objective with our capital structure is to create something that balances the risk of the business with the future growth of the business. And right now, we essentially have bitcoin credit risk. And as we look forward to 2025, it's a much more stable credit risk business. And so we feel very confident that there will be opportunities for us. Obviously, the $260 million conversion of the convertible notes, that's an incredible milestone for us as a company. And we've almost cut our debt in half since emergence. And we've covered that by over 60% since the beginning of the year.



So we're in a process right now of making sure we are going to have the right securities on our balance sheet. There's no urgency around that. We feel very comfortable with where we're at from a capital structure perspective today. But as we look forward, that perspective obviously changes. And so we're evaluating all opportunities that are available to us. And those opportunities are going to continue to expand as we start to see more of our sites come online.

[Operator]

Our next question comes from Joe Flynn with Compass Point Research.

[Joseph Flynn, analyst]

I was hoping you could comment on the remaining options for the 112 megawatts of data center capacity. Just like what ultimately has been achieved to get those signed, whether it be like vitalizing designs or securing items in the supply chain? And could you maybe just comment on continuing conversations with diversifying the customer base?

[Adam Sullivan, executive]

Thanks for the question, Joe. What we're looking at today and the one point I'll make a note on first is we're obviously very close discussions with CoreWeave, just recently executing 118 megawatts this week, which is another great milestone for us as a company. The next 112 we're working through finalized designs, working through all the necessary items related to that.



And we're still in discussions with other clients, and I think that's really a key point here is that all of the work we're doing to prepare and negotiate with CoreWeave on these last 12 or 118 megawatts. That all directly relates to the same type of design and diligence that would be required for us to sell that site to somebody else as well.



I think on the client diversification point, that's really important. That's really why we're so focused on our site selection right now. We want to be able to make an offering to a number of clients who have different needs, different size needs, different location needs. And those -- and part of what we're going through right now is ensuring that we are building a portfolio that is attractive to these end clients. And so that's why we're very aggressively pursuing sites today, and that's going to provide us a really interesting opportunity for the remainder of this year to diversify our client base, which is one of our key catalysts.

[Joseph Flynn, analyst]

That's helpful. And I was wondering if you could comment on reports about delays in black well shipments. And ultimately, it sounds like the contract starts on delivery of the data center, but if that could ultimately impact kind of timing of deliveries?

[Adam Sullivan, executive]

We're obviously keeping a pause on where black well pods are in the delivery process. This is something that we're working very closely with our end clients and even potential clients on to ensure that we're delivering facilities alongside of schedules of when there's going to be the deliveries of those GPUs. So it's hard to comment today about where the deliveries will end up in 2025 and 2026. And really, right now, we're focused on continuing to execute and deliver these facilities as quickly as we possibly can for our end clients to have the optionality to turn on as quickly as those GPUs are delivered.

[Operator]

Our next question comes from Brett Knoblauch with Cantor Fitzgerald.

[Brett Knoblauch, analyst]

Maybe to start on sourcing new sites and locations. I guess can you maybe just provide some color on the supply and demand imbalances out there for energy? And I guess, what you're seeing in terms of, not competition for the same sites, but going out and adding additional sites, and how to view that HPC demand is only going to continue to increase. So I guess, how tight is the market out there? I don't know if you can quantify it or anything qualitative would be helpful. But I guess how tight is the market out there for power?

[Adam Sullivan, executive]

Thanks, Brett. The market right now is tight and it's requiring, I would say, expertise in the site selection process. The interesting part about the process we've gone through over the past 7 years, locating sites for Bitcoin mining is we're using a very similar criteria in terms of what we're looking for on the HPC side.



So we've been able to locate sites, and we're in the process of working through those negotiations on sites that will provide us -- or that are, I would say, a more interesting opportunity for us from a pricing perspective because it is competitive. On these large sites, even on behind the meter, that's another part that is extraordinarily competitive today amongst the hyperscalers. And so those are some processes that we're involved in, but isn't necessarily the main focus for our site selection.



And we're looking at a range of different sites, a range of different states of those sites. And I think that's really going to provide us the most amount of value here. And that's really just given our experience working with utilities, working with local governments. And all of those things over the past 7 years that we've worked through in the bitcoin mine side are aiding us greatly in this process that we're going through on the HPC side.

[Brett Knoblauch, analyst]

No, that's helpful. And then I guess when you're talking about different locations and regions across the U.S., are you looking at a site with, I guess, thinking this site could be used for training or the site could be used for inferencing? Or is it all one of the same for you guys? Or are you like primarily looking at maybe like very low 1 to millisecond latency sites?

[Adam Sullivan, executive]

Our focus is mainly on low latency sites. What we're seeing right now is a pretty major shift or we believe there's going to be a major shift going on in training versus inference over the next few years. And so when we are locating sites, we want to ensure that they have the opportunity to do both. We think that's important not only for the longevity of the site, but also for the longevity of the client health as well, having the optionality to switch between either application that they're utilizing the site for.

[Operator]

Our next question comes from Joseph Vafi with Canaccord Genuity.

[Joseph Vafi, analyst]

Everyone. Nice to see all the progress and thanks for taking my questions. I guess first one is on Austin, you're generating HPC-related revenue there now. I know it's still early days in this journey. But interesting, if you could share some of your learnings to date out of that facility in this new line of business and potentially what and how it may be kind of helping you plan and get ready for larger deployments. And then I'll have a quick follow-up.

[Adam Sullivan, executive]

The interesting part about our team is many of our team members actually sat in this facility when they work for HP. This exact facility, sit in very similar places in the office. So we have high familiarity with this site. Some of the learnings that we're going through, I would say, both on the client side as well as on the newest generation of GPUs that are coming out. Some of those requirements are different than the historical data centers that have been developed over the course of the past 20 years.



And we've been integral in the design process and working with not only our end client but also NVIDIA on what the design is of these facilities. And that's aiding us in the process as we work through the designs on the future conversions that we're working on today, the next 382 megawatts. So you're absolutely right, there are learnings with this newest generation of GPUs, both from a cooling perspective as well as a latency perspective. And all of those things are going to aid us in this conversion process going forward.

[Joseph Vafi, analyst]

That's good color. And then also, I know it's early days here on the block 3-nanometer chipsets. But maybe you could remind us here on why you find these attractive. I mean there's clearly a lot of good minor solutions in the market and maybe a quick hint into how you see them performing relative to the rest of your fleet when they're deployed.

[Adam Sullivan, executive]

Our objective is to build our future facilities with lower CapEx, have lower maintenance costs, and we're really focused on developing the next generation of Bitcoin mining infrastructure. And so as we are evaluating the type of opportunities that were in the market, we realized it was much easier to design a facility around a hash board than it was to design a facility around a machine. And so as we look forward to what the future of bitcoin mining looks like, we recognized that form factors were going to continue to evolve over time, and we're going to be at the forefront of that evolution.



And that's something very unique to Core Scientific. We've operated over 16 different manufacturers in our facilities, over 700,000 machines. That provides us very unique insight into not only the operational process for machines but also the repair process. Our repair center would be one of the largest repair centers in the country if we were just a repair business.



And so that's something that's very unique to our business. We have a very unique insight, and we're trying to be on the forefront of this industry. And I think people are going to be fast followers to where we're going right now.

[Operator]

Next question comes from Darren Aftahi with Roth Capital Partners.

[Darren Aftahi, analyst]

First, I guess, just diving into site selection a little bit more. Can you talk a little bit about your source of referrals. I know you said local governments and utilities. But I'm curious about that. And then would you guys ever look at maybe subscale legacy data centers as an option? And then I'm also curious if this is primarily domestic exercise or if you guys would look international as well?

[Adam Sullivan, executive]

The way we look at it right now is there's really 2 avenues for growth. One is brownfield and greenfield opportunities. The second is something that you mentioned, the subscale data centers. I'm going to touch on the second one first here. We have been looking at a number of different existing data center opportunities similar to the deal that we did in Hibbett's. I think going forward, one of the things that we're looking at more closely is acquiring those sites and finding data centers that have excess megawatt capacity on the substation that will allow us to continue to increase the capacity of that facility. Just given that power densities have increased so rapidly, we have an opportunity to bring a lot more megawatts into a traditional data center.



And so that's something that's very attractive to us right now given our experience already retrofitting existing data centers. When we talk about working closely with the utilities, local governments, that's just something that we've done over the past 7 years in the bitcoin mining side, and we formed very close relationships with all of our utilities.



And we think that's really important. I would say, over the past 7 years, bitcoin mining has come in and out of favor with governments, with local communities. And really, one of the things that we've tried to highlight is that we try to be great community citizens in every jurisdiction that we operate in. We try to be great partners to our utility companies.



And all of that has paid dividends as we've started to make the shift in the HPC and we look towards expansion. That's something that's opened a lot of doors for us, just given how good of an operator and how good of a community citizen we've been in each of our locations.

[Darren Aftahi, analyst]

Great. And then just as a follow-up. On the time line you guys laid out on Slide 15, I'm just kind of curious, obviously, you laid it out there, so you feel confident in it. But if there was to be slippage in that, is there any penalty to Core Scientific if you don't hit those bogeys?

[Adam Sullivan, executive]

This is something we work very closely with CoreWeave on the time lines. We share in our execution risk in terms of getting the facilities and the materials on the site as quickly as we possibly can. So it's less of an issue from a perspective of penalties. It's more of an issue of us ensuring that working together with CoreWeave, we're able to source all of the necessary items. But we've worked really hard to ensure that we not only have our plan A on supply chain, but also a Plan B and a Plan C. And so we have all of those lined up, and I think we're excited and confident in the time lines that we've laid out today.

[Operator]

[Operator Instructions] Our next question comes from Rosemarie Sison with Odeon Capital.

[Rosemarie Sison, analyst]

Just had one thought on your expansion, which sounds really interesting from the HPC idea with brownfield and greenfield expansion. The question comes to mind is that I'm just curious as to whether you feel that you have enough infrastructure in your current state to be able to support that. It sounds like it would take a lot of people with expertise in order to effectuate that kind of a strategy. And I don't know, I think the other caller asked something about whether you might be thinking about going globally or not. You didn't address that, so I'm assuming no, but just the question arises, can you -- first of all, your expertise in-house is very important to keep and to keep happy. But also how do you look at expanding that in order to facilitate your growth?

[Adam Sullivan, executive]

Thanks for the question, Rosemarie. And just to go back, we are looking at international opportunities as well, it's not just domestically. But as we look at our team today, we're set up to execute on additional projects in 2025 and 2026 above what is currently contracted. So we feel very confident in that. Now that doesn't mean that we're not looking to add very specific talent to our bench. And so that's a process that we're going through today. And what we've been very lucky in is the fact that we have a great story right now.



We're on the leading edge of both bitcoin mining and HPC infrastructure, 2 of the highest growth industries in America right now. And so we've been very fortunate to be able to attract talent to our company who want to be a part of our story. And so that's something that we've been finding, I would say, easier to attract talent to the company. And it's something that really excites us is the fact that we've been able to add additional bench strength even over the course of the past 3 months, to just help aid in the execution process, not only in the first 382, but also in the future sites that we're looking to acquire.

[Rosemarie Sison, analyst]

That's great. And internationally, where would you think about expanding?

[Adam Sullivan, executive]

Our objective right now is to work with the potential clients that we're negotiating with domestically and to locate in areas where they'll find -- or they have attractive opportunities for themselves with their end customers. So it's hard to speculate on where we may go, but we're looking at a number of international markets that we think will be very attractive long term, not only for existing potential clients that we were in discussions with, but also attracting and refining new potential clients as well.

[Operator]

Our next question comes from Kevin Dede with H.C. Wainright.

[Kevin Dede, analyst]

Adam, thanks for having me. Hi Denise. I'm not sure that you really peeled back the onion on the locations that you're working to convert. And I understand the first -- well, actually, it's a second 200 megawatt extension with Core. You committed 80 megawatts to power that. I was wondering if you could offer a little detail on location and additional power you might need in support of HPC.

[Adam Sullivan, executive]

Thanks, Kevin. One of the points I want to emphasize first is there is a competitive advantage for us right now. We're leading the market in terms of contracts signed this year for HPC for AI. And we're looking at potential expansion opportunities at some of our sites as well. So where we believe we have a competitive advantage. We're not necessarily trying to give that edge away. And so we're working diligently with local utilities. We're also looking at potential expansions within the markets we're currently in as well, given the existing relationships that we have.



So we feel like we're in a very good position. And when the time is right, it's hard to speculate on when, but we will be announcing the locations of the places that we are converting from bitcoin mining to HPC.

[Kevin Dede, analyst]

And the power required to supplement the additional contract extensions that you've announced since the second one, I guess, the best way to phrase it, Adam. Is there any more detail on that you could offer.

[Adam Sullivan, executive]

Kevin, I just want to make sure I understand the question. You're just saying what the total power necessary to service the additional 112 and the additional 70?

[Kevin Dede, analyst]

Exactly. Yes. I would assume that the 16 in Austin is all set, you don't need additional power there. But in developing these other locations, obviously, you're going to need the power to support them.

[Adam Sullivan, executive]

Yes. A great way to think about that is when we talked about in our Investor Day about the 700 megawatts needed to service the 500 megawatts to the GPUs. So a great way to think about it is about 1.35 to 1.4 POE that's necessary above what's contracted from a GPU megawatt perspective. And so if you're trying to calculate kind of back the envelope map in terms of what the total megawatt need is, that's a great metric to use kind of multiplying the number by 1.35 to 1.4, depending on the location.

[Kevin Dede, analyst]

Fair enough. Could Denise or yourself help me with understanding the cash generated from the warrant conversion? What sort of trends have you seen? And what would be fair to expect?

[Denise Sterling, executive]

Yes. Kevin, so thank you for the question, Kevin. We saw a very minimal impact to Q2 from the actual exercise of our tranche 1 warrants. And so you'll actually be able to see that coming through from a balance sheet perspective. But we've actually, in the last several weeks, seeing the conversion of about $35 million of our tranche 2 warrants. And as you remember, those are penny shares. So no significant cash flow. So we really haven't seen the $670 million that we talked about associated with our Tranche 1 warrants to date. But we do have some movements, and we will continue to see some of those roll in, but they're not significant as of yet.

[Kevin Dede, analyst]

You mentioned $30 million and $35 million added to the share count.

[Denise Sterling, executive]

Yes. Just to kind of put things in perspective, Kevin. We ended the quarter with about 183 million shares. Where we sit today as of the end of last week is at 258 million shares, as we mentioned in our prepared remarks. And that's primarily driven by the additional conversion of our convertible notes for the mandatory piece. It was about 40 million shares. And as I mentioned, the exercise of our tranche 2 warrants added about 35 million shares.

[Operator]

Our next question comes from Jack Chan with Imperial Capital.

[Jack Chan, analyst]

I wanted to ask on the expansion opportunities. Will you be looking to do deals with these future customers at the same economics as your CoreWeave deal? Or are different structures being contemplated? And if so, what those could look like?

[Adam Sullivan, executive]

The contract is something that's actually something that we're working through today in terms of how do we structure these and have the right give and take between the buy down of CapEx versus the rate that they pay. And so we're working through that based on both the credit risk of the potential customer as well as what the length and tenor is of that end contract. So we're working through that today.



So obviously, the CoreWeave deal was extraordinarily unique. It was something that completely changed the data center industry from a term perspective. So we're obviously using that as a starting point. But I would say we're expecting the contracts to change based on where we are going forward, our ability to access the capital markets in different ways to make the potential new contracts that we're signing more efficient. And so there's a number of opportunities for us, and we're exploring all of them.

[Operator]

Our next question comes from Lucas Pipes with B. Riley Securities.

[Lucas Pipes, analyst]

Just really quickly on Slide 15, you show application-specific data centers kind of 1, 2, 3, 4, 5. Would those be all at different sites or might there be an overlap? And should we think of those as kind of, call it, ASDC1 an 80-megawatt cluster at one of the facilities. Just wanted to clarify that.

[Adam Sullivan, executive]

Thanks, Lucas. A great way to think about this is more on a facility level basis, not necessarily on a site level basis. So it doesn't necessarily mean that it's spread across all different sites. It could potentially be across different facilities at specific sites.

[Lucas Pipes, analyst]

Okay. So essentially, it's a different -- 1, 2, 3, 4, 5 is at different sites, but it may not be within one building, it might be several buildings.

[Adam Sullivan, executive]

Yes, absolutely right. And we do have multiple sites at some of our sites. And so it's a great way to think about it's more on a facility level, those ASICs.

[Operator]

Thank you. And there are no further questions at this time. I'll turn the floor back over to Steve Gitlin for closing comments.

[Steven Gitlin, executive]

Thank you, Diego, and thanks, everybody, for your questions and for your attention today. An archived version of this call, all SEC filings and relevant company and industry news can be found on our website, corescientific.com. We wish you a good day, and we look forward to speaking with you again following next quarter's results.

[Operator]

Thank you. This concludes today's conference. You may disconnect your lines at this time. Thank you all for your participation.