Cipher Digital Inc. Q3 FY2025 Earnings Call

Good day, and welcome to the Cipher Mining Third Quarter 2025 Business Update Conference Call. As a reminder, this call may be recorded. I would now like to turn the call over to Courtney Knight, Head of Investor Relations. Please go ahead.

Good morning, and thank you for joining us on this conference call to address Cipher Mining's business update for the third quarter of 2025. Joining me on the call today are Tyler Page, Chief Executive Officer; Greg Mumford, Chief Financial Officer; and Edward Farrell, Senior Adviser and former Chief Financial Officer. Please note that our press release and presentation can be found on the Investor Relations section of the company's website, where this conference call will also be simultaneously webcast. Please also note that this conference call is the property of Cipher Mining and any taping or other reproduction is expressly prohibited without prior consent. Before we start, I'd like to remind you that the following discussion as well as our press release and presentation contain forward-looking statements. These statements include, but are not limited to, Cipher's financial outlook, business plans and objectives and other future events and developments, including statements about the market potential of our business operations, potential competition and our goals and strategies. Forward-looking statements and risks in this conference call, including responses to your questions, are based on current expectations as of today, and Cipher assumes no obligation to update or revise them, whether as a result of new developments or otherwise, except as required by law. Additionally, the following discussion may contain non-GAAP financial measures. We may use non-GAAP measures to describe the way in which we manage and operate our business. We reconcile non-GAAP measures to the most directly comparable GAAP measures, and you are encouraged to examine those reconciliations, which are filed at the end of our earnings release issued earlier this morning. I will now turn the call over to our CEO, Tyler Page.

Thanks, Courtney. Good morning, everyone, and thank you for joining us today. I'm Tyler Page, CEO of Cipher Mining, and I'm pleased to welcome you to our third quarter 2025 business update call. The third quarter was truly transformative for Cipher as we made huge strides on our strategic pivot into the high-performance computing space and set the stage for what is, without question, the most exciting earnings update in our company's history. This quarter, we executed a pivotal transaction with Fluidstack and Google, which firmly established our credibility in the HPC space. Following that groundbreaking transaction and leveraging that success, we've now taken another major step forward. I'm thrilled to announce today that we've executed a second landmark HPC transaction, this time with Amazon Web Services. Partnering directly with one of the largest and most innovative companies in the world underscores Cipher's emergence as a trusted leader in next-generation compute infrastructure and confirms our full-scale transformation into an HPC data center developer. Our first HPC deal with Fluidstack and Google established not only Cipher's credibility as a data center developer for the world's most demanding tenants, but also the desirability of more remote areas of Texas for next-generation data centers. We have been talking to investors for over a year about this thesis and saying that we thought the market would evolve in our direction. Our second long-term lease this time with Amazon proves that neither we nor West Texas are one-hit wonders. Our second lease space is the world's largest hyperscaler directly on a 15-year lease at very attractive terms. This is not a fluke and will not be our last HPC deal. Under the agreement, we contracted 300 megawatts of gross capacity, and the project carries approximately $5.5 billion in contract revenue over the initial 15-year term. The capacity will be delivered in two phases beginning in July 2026 and completing in Q4 2026, with rent commencing in August of 2026. Given the strength of the lease we have secured, we believe that we will utilize debt financing to fund the majority of construction costs at the site, and any remaining construction obligations will be funded from cash on hand with no need for further equity fundraising. With these milestones, Cipher has officially arrived as a leader in the HPC revolution, harnessing our sourcing expertise, energy assets, best-in-class team, and operational excellence to power the world's most advanced computing workloads. Continuing with that momentum, we're proud to announce today that we've secured ownership in a joint venture to develop a 1-gigawatt site in West Texas. We expect to own approximately 95% of the JV once a turnkey HPC lease is executed, assuming standard lease and development terms. We are calling the site Colchis, which refers to the mythical home of the Golden Fleece and was a land of legendary wells located at the edge of the known world. For the past 1.5 years, conventional knowledge in the traditional data center industry has been that hyperscalers would not venture outside of major metropolitan areas and that our sites were at the edge of the world. But we have now conclusively proven those incumbents wrong. We will continue to do so at Colchis. This is the most significant addition to our development pipeline to date. This site features a fully executed 1-gigawatt Direct Connect Agreement with American Electric Power, providing dual interconnection capability and targeted power availability in 2028. The transaction also includes options to purchase up to 620 acres of land adjacent to the existing substation. The Colchis site checks every box for a premier HPC development opportunity, ample acreage, large-scale power capacity, availability of diverse fiber routes, and dual interconnection capability. We have already begun to have early-stage discussions with potential tenants for the site. The execution of this transaction once again demonstrates our team's sourcing expertise and ability to secure some of the most attractive large-scale sites in the world. Cipher is one of the few companies in the world that can combine boots-on-the-ground expertise working directly with landowners to source best-in-class sites with a deep technical sophistication needed to serve hyperscalers. This unique and powerful combination makes Cipher exceptionally well positioned to bridge the growing gap between the limited supply of suitable sites and surging large-scale tenant demand. The announcements we shared today are the results of years of hard work and the strong execution and momentum built over the past quarter. I'd like to take a moment to reflect on some of our third-quarter successes. At the forefront of these highlights is our recent transaction with Fluidstack and Google, a transformative 10-year 168 critical IT megawatt AI Hosting Agreement that first positioned Cipher as a major developer in the HPC space. Under this agreement, Cipher will deliver 168 megawatts of critical IT load at our Barber Lake site in Colorado City, Texas, supported by up to 244 megawatts of total capacity. This project represents approximately $3 billion in contracted revenue over the initial 10-year term with options that could extend total contract value to roughly $7 billion over 20 years. Notably, construction is already underway at the site, and we are on track to deliver the full 168 megawatts of critical IT capacity by September 30, 2026. Importantly, Google is backstopping $1.4 billion of Fluidstack's obligations to support project financing and will receive warrants representing roughly a 5.4% pro forma equity stake in Cipher. Cipher will retain full ownership of the site and is in the process of securing debt to fund construction. We will provide more details around that construction financing in the near future. We believe and have now proven that Barber Lake was just the beginning, the first of several projects to capitalize on our team's sourcing expertise, proven development capabilities, strong industry relationships, and unmatched construction track record. We look forward to continuing to partner with leading technology companies to secure HPC leases at our growing pipeline of sites. This expansion is well supported by our successful $1.3 billion convertible offering completed this quarter. This was the largest digital infrastructure convertible issuance to date and was roughly 7x oversubscribed, demonstrating investor confidence in our strategy and pipeline. The strong demand allowed us to take advantage of favorable market conditions, securing a 0% coupon and further strengthening our balance sheet. Greg will discuss the convertible offering in further depth later on the call. The Amazon transaction, the Fluidstack and Google transaction at Barber Lake, the addition of significant new capacity at Colchis, and our successful convertible offering all represent major milestones in advancing our HPC strategy. Together, these achievements expand our business model, secure substantial future capacity, and strengthen our balance sheet, all positioning Cipher to capture the tremendous demand we're seeing and play a critical role in building the next generation of AI infrastructure. As we scale and expand our business model, our bitcoin mining business continues to generate meaningful cash flow. The company surpassed expectations this quarter and is now operating approximately 23.6 exahash per second of self-mining capacity. The same disciplined foundation we established in the bitcoin mining space, delivering 5 data centers on time and on budget will fuel our successful expansion into HPC. I'd now like to provide a brief overview of our energy portfolio, which highlights our execution across business lines and the strength of our pipeline going forward. On the mining side of the business, this quarter, we brought Black Pearl fully online, which grew our operational mining capacity from 423 megawatts to 477 megawatts across Odessa, Alborz, Bear, Chief, and Black Pearl. In doing so, we exceeded our previous hash rate projections and achieved a total self-mining hash rate of approximately 23.6 exahash per second. In addition, our fleet efficiency stands at an extremely impressive 16.8 joules per terahash, making us among the most efficient miners in the industry. Our proprietary software, which allows us to dynamically curtail our data centers, has proven to be a critical advantage in optimizing for profitability, maintaining low power prices, and monetizing older rigs. This area of expertise is expected to remain a key competitive advantage in the future and, in fact, may be an increasingly valuable aspect of the business as the HPC landscape continues to evolve. Importantly, our current mining operations are fully funded, and we do not anticipate further investment in that side of the business as we prioritize our pipeline toward HPC. As discussed, this was a monumental quarter for Cipher in that we grew our contracted AI hosting capacity from 0 last quarter to 544 gross megawatts this quarter across 2 transactions with world-class partners. Behind that, we have a robust pipeline of 3.2 gigawatts of future capacity that spans from 2025 to 2029 and beyond. While we are extremely proud of our mining production, market dynamics, scarcity of energy capacity, and frenzied demand from tenants has made it clear that the best use of our extensive pipeline of sites is for HPC workloads. We are in ongoing discussions on our pipeline with leading partners and look forward to prioritizing all of these sites for HPC development. Let’s now turn to a review of our current operations on both sides of the business. At Barber Lake, we are constructing a data center for our industry-leading partners, Fluidstack and Google. Construction at the site is well underway. Ground has been broken, and both engineering and procurement are progressing smoothly. We've secured the necessary labor force and locked in most of the long lead time equipment, putting us in a strong position to meet all key construction milestones on schedule. We are firmly on track to deliver the full 168 megawatts of critical IT capacity by September 30, 2026. The lease is anticipated to commence the following month in October 2026. Note that we still retain 56 megawatts of current capacity at Barber Lake. These additional megawatts allow us to pursue an additional colocation agreement, potentially prioritizing different deal elements or to deploy our own compute at the site. Our team is carefully assessing the merits of all potential options to maximize the value of the remaining 56 megawatts in Phase 1. In addition, we maintained an MOU on an additional 500-megawatt upsize at the site, which would come online in 2029 to 2030. Given the site's ongoing development potential and live deal discussions, we look forward to providing further updates as things progress. Turning to our current mining operations. As of September, the current operating hash rate at the site is approximately 11.3 exahash per second using approximately 207 megawatts. Odessa's fleet efficiency stands at roughly 17.6 joules per terahash. On this page, we also provide the observed all-in electricity cost per bitcoin at our 5 sites. Black Pearl began contributing significant cash flow to the business in the third quarter. The first 150 megawatts at the 300-megawatt site are currently mining approximately 10.1 exahash per second, exceeding prior guidance and contributing approximately 36% of production this quarter. Fleet efficiency at the site stands at an extremely impressive 13.9 joules per terahash. Lastly, we provided a combined overview of our joint venture data centers of Alborz, Bear, and Chief. The 3 sites have a total power capacity of 120 megawatts and generate approximately 4.4 exahash per second. We own 49% of the JV sites and our portion recently generated roughly 9% of our overall bitcoin production in the third quarter. Let's now shift to an update on our development portfolio. Slide 11 provides an overview of our next to energize site in Andrews County, Texas called Stingray. The site features 100 megawatts of front-of-the-meter capacity, all necessary regulatory approvals, and 250 acres of land adjacent to the transmission assets. In the third quarter, we continued the development of the substation for the site and secured long lead time items, including transformers and high-voltage breakers. The site is on track to energize in the fourth quarter of 2026. Slide 12 outlines additional capacity spanning 2027 and beyond. Reveille located in Cotulla, Texas is on track to energize in Q2 2027. The site is fully approved for 70 megawatts, and we have initiated development of the substation. Given both Stingray and Reveille have secured interconnect approvals and established energization time lines, we've engaged with multiple prospective tenants and are in ongoing discussions to secure the most attractive lease agreements for these locations. Our 3Ms, Mikeska, Milsing, and McLennan are all currently undergoing final interconnection approval processes and load studies have been completed at all 3 sites. The interim Oncor FEAs have been signed with Oncor for Mikeska and McLennan, and the required deposits have been paid. We're targeting up to 500 megawatts of capacity at each of these sites. In addition to interconnection rights, our purchase options also include significant land parcels at each location, all of which are well suited for HPC data center development. We are confident these sites will be in high demand as development progresses. Last on this page is Colchis, which, as mentioned, is our latest site acquisition and the most substantial addition to our pipeline to date. The site features a fully executed 1-gigawatt Direct Connect Agreement with American Electric Power, providing dual interconnection capability and targeted power availability in 2028. The site is roughly 80 miles southwest of Abilene and around 80 miles southeast of our Barber Lake facility. As mentioned, the site is extremely well suited for HPC given its ample acreage, large-scale power capacity, availability of diverse fiber routes, and dual interconnection capability. Last quarter, we discussed our strategy to position Cipher ahead of the curve in anticipation of the evolving AI data center landscape. Since then, we have executed 2 landmark HPC transactions as well as our most significant pipeline addition to date. With the industry moving even faster than we had anticipated, we are more confident than ever that Cipher is among the best positioned companies in the world to seize the near-term opportunities created by the growing power shortfall. Simply put, we are just getting started. I will now turn it over to our new CFO, Greg Mumford, for a review of our third-quarter financials.

Thanks, Tyler, and good morning to everyone on the call. I'm excited to join today's call as Cipher's new Chief Financial Officer. It's a privilege to be part of such an innovative company that's playing a key role in the evolution of digital infrastructure and high-performance computing. I want to start by expressing my gratitude to Ed Farrell for his leadership and many contributions over the past 5 years. Ed has built a world-class finance organization and leaves behind a strong foundation that positions Cipher well for its next phase of growth. The company is fortunate to have his continued guidance as a senior adviser during this transition period. As I step into this role, my focus will be on maintaining a disciplined approach to our financial strategy, broadening access to new funding sources, and optimizing our overall cost of capital. We'll continue to take a thoughtful approach to capital allocation, ensuring we're maximizing sustainable long-term growth and driving value for our shareholders. I'm excited to work with Tyler, the leadership team, and our talented finance organization to build on Cipher's strong momentum. To begin, I'd like to remind everyone that today I will be discussing our performance for the third quarter of 2025, which ended on September 30. I'd like to highlight that this quarter was marked not only by strong execution as we officially expanded into our HPC hosting and grew our pipeline, but also by disciplined capital raising that positions us to sustain and accelerate that momentum moving forward. During the quarter, we completed our second convertible offering, an upsized private placement of $1.3 billion of 0% convertible senior notes due 2031. This transaction reflected strong investor demand and confidence in Cipher's long-term strategy. The notes were issued with an initial conversion premium of approximately $16.03 per share, representing a 37.5% premium to our stock price at issuance. We also entered capped call transactions that increased the effect of conversion price to approximately $23.32 per share, substantially reducing potential dilution to our shareholders. The net proceeds from the offering were used to fund the cost of entering into the capped call transactions and will be used for construction at our 2 currently contracted HPC sites to advance our HPC strategy across our now 3.2 gigawatt development pipeline and for working capital and general corporate purposes. Importantly, this financing bolsters our balance sheet and reflects our disciplined approach to growth. We're very pleased with the market reception and believe this transaction positions Cipher well to capture the significant opportunities ahead in HPC and digital infrastructure. Let's now turn to a review of our financials, beginning with our sequential financial performance. In the third quarter, our hash rate increased by 40%, driven by the energization and ramp-up of our Black Pearl facility, where Phase 1 of the 150-megawatt front-of-the-meter site came online in June. Black Pearl began the quarter contributing approximately 3.4 exahash per second and ramped up to approximately 10.1 exahash per second during the quarter. This led to a 35% increase in production as well as an increase in our electricity cost per bitcoin given Black Pearl is a front-of-the-meter site. The higher cost per bitcoin was also driven by an increase in network hash rate over the quarter. Moving down the slide, we reported $72 million in revenue, up 65% from $44 million in the prior quarter. This growth was driven primarily by the increase in bitcoin price and the increased production from Black Pearl. For the quarter, we reported a GAAP net loss of $3 million, or $0.01 per share compared to a net loss of $46 million, or $0.12 per share in the prior quarter. We are proud of the substantial quarter-over-quarter improvement in our results, particularly given that bottom line performance was impacted by higher depreciation expense. This depreciation expense reflects the assets placed into service at Black Pearl, including the deployment of the latest generation rigs, as well as the upgrade at Odessa completed in Q4 2024. Additionally, the bottom line continues to be influenced by changes in the fair value of our power purchase agreement at Odessa. These expected fluctuations reflect movements in forward power prices and the decaying time value of the remaining contract term, which extends through July 2027. As Ed has previously noted, the true benefit of this contract lies in its provision of long-term, low-cost fixed price power for our Odessa operations. This quarter, as part of the execution of our HPC lease at Barber Lake, we granted Google warrants as compensation for their commitment to backstop the lease payments from our tenant Fluidstack. These warrants are recorded at fair value and as a result, this quarter, we recognized a $32 million gain in change in fair value of the warrant liability. Excluding noncash expenses, such as the change in fair value of our power purchase agreement, share-based compensation, depreciation and amortization, deferred income taxes, the change in the fair value of the warrant liability, and nonrecurring losses, we reported a third quarter adjusted earnings of $41 million or $0.10 per share, up roughly 34% from $30 million last quarter. Cash and cash equivalents increased significantly, driven by the $1.2 billion of net proceeds from our most recent convertible financing. Let's move on to Slide 15 and take a deeper look at the results of our operations. For the quarter, we mined 383 bitcoin at Odessa and 246 at Black Pearl, bringing our total production to 629 bitcoin mined in total across our wholly owned sites. This production generated $72 million in revenue at an average price of roughly $114,400 per bitcoin. This compares to the 434 bitcoin mined in Q2 2025 at an average price of $99,700 per bitcoin, resulting in $44 million in revenue. G&A expenses, which include IT, corporate insurance, professional fees, and other public company costs decreased slightly both sequentially quarter-over-quarter and year-over-year. Depreciation and amortization expense totaled $60 million, up from prior periods, driven by the deployment of the new mining rigs over the last 12 months. Our oldest rigs in the fleet will be fully depreciated in Q4, but those rigs can remain productive and continue to generate attractive returns when deployed strategically. We recognized a small unrealized gain on our bitcoin holdings this quarter compared to a $17 million gain in Q2, reflecting a modest increase in the spot price at quarter-end. We finished the quarter holding approximately 1,500 bitcoin in treasury. On our non-GAAP reconciliation, we reported a GAAP net loss of $3 million. Adjusting for $44 million in noncash and one-time items results in adjusted earnings of $41 million for the quarter, up from $30 million in the previous quarter. Now let's turn our attention to the balance sheet. On Slide 17, total current assets at quarter-end were $1.4 billion, up from $220 million last quarter, driven primarily by the net proceeds of the $1.3 billion we received from our convertible offering. In addition, we held $170 million of bitcoin. As we have discussed in depth on our previous earnings calls, we actively manage our treasury, neither selling nor holding every bitcoin mined, and we remain disciplined in our approach to capital management. I'll quickly cover some additional balance sheet line items as of September 30. PP&E totaled $650 million, up 37% from $474 million. This increase is primarily related to equipment deployed at Black Pearl. Deposits on equipment of $8 million, down from $183 million last quarter, is primarily related to the reclassification of rigs at Black Pearl from deposits to in-use property and equipment. At the end of the third quarter, our equity interest in the Alborz, Bear, and Chief JVs stood at $42 million. Moving down the balance sheet. Derivative assets were up primarily due to the inclusion of $90 million of capped calls associated with the new convertible note, which raises the effective conversion price of the convertible debt and effectively minimizes potential dilution to shareholders. Current liabilities increased this quarter due to the short-term classification of the Google warrants associated with the Fluidstack lease at Barber Lake. Lastly and importantly, I want to highlight that short-term borrowings remain at 0. We continue to manage the balance sheet conservatively, ensuring we're well positioned to meet any capital needs. Before we conclude, I'd like to thank everyone for joining today's call. We're proud of the tremendous progress we've made this quarter and the transformative growth we've achieved as we continue to expand our business lines, grow our pipeline, and strengthen our balance sheet to support that growth. As always, we remain firmly committed to disciplined execution, capital efficiency, and delivering long-term value for our shareholders. Thank you for your continued support, and we look forward to updating you on our progress in the next quarter. At this time, I will pause, and Tyler and I would be pleased to take any questions.

Our first question comes from Paul Golding with Macquarie.

Congrats on the announcement and all the progress on HPC. I wanted to start off with a question around the deal itself. 300 gross megawatts, Stingray, you have on track for energization, 100 megawatts in '26 and Barber Lake, you have 56 megawatts after the Fluidstack deal. How should we think about the distribution of power to deliver the 300 megawatts as well as maybe pricing across liquid and air cooled since you're delivering both? It looks like averaging out the deal is about $1.7 million per critical megawatt on my back of the envelope math. If you could just talk through some of those deal points on pricing as well as how you plan to deliver that capacity across your fleet. And then I have a follow-up.

Sure. Thanks, Paul, for your question. To begin, we're still finalizing the details of the deal we signed with AWS. This involves determining the specific numbers that will be represented. They are taking 300 gross, and we are modifying an existing air-cooled 150 megawatts, which allows for a quick time to market for the first phase of that build. For the second phase, we are still working on the design, and there are discussions about the trade-offs between speed to market and optimizing for the highest critical IT load. This is not finalized yet. Generally, if the entire site ends up being air-cooled, the Power Usage Effectiveness (PUE) will align with our design at Barber Lake, which is around 1.4 to 1.45. If we incorporate more liquid cooling in the second phase, we could achieve a more efficient PUE. We are still clarifying those numbers. Regarding costs, as mentioned, we expect the cost per critical IT megawatt to align with what we've seen in the past at Barber Lake.

And you...

If not better, because we do have some infrastructure in place already that was bought in a cheaper market.

Got it. Appreciate that color. And then you also mentioned debt financing as a majority of CapEx sourcing and then cash on hand. Are you able to give any more detail around financing plans in terms of you're already developing the Fluidstack capacity. So is this cash on hand from prepayment deposits? And is there any kind of backstop here to help support project financing and going to market for that?

Let me outline a high-level framework for that, and then Greg can add any specifics if he wishes. There are two different structures. The first deal is with Fluidstack and Google, which resembles existing market structures. We plan to explore our debt financing options for this in the near future, and I expect the structures to be quite similar depending on market conditions. Regarding the AWS lease, it's a direct-facing hyperscaler lease and likely the first of its kind for a conversion from bitcoin mining to a long-term 15-year lease. This should be very financeable. We have recently increased the convertible debt offering to $1.3 billion due to favorable market conditions, providing us with significant excess cash. This should support the equity component needed for the AWS lease financing. Greg, do you have anything else to add, or is that sufficient?

Yes. I mean, Tyler, I think you said it right earlier, is that we're not prepared to give specifics on the financing that we're looking at for the Google, Fluidstack deal, but we are exploring opportunities, and we'll be hopefully updating the market in short order. As it relates to the AWS deal, we think that there's going to be a lot of opportunities in front of us to explore different types of project or construction level financing, and we're going to work through those options and make sure that we're making the right decision.

Our next question comes from Greg Lewis with BTIG.

I guess the first question is about additional sourcing of power. Congratulations on that. It seems pretty challenging. Tyler, as we think about things accelerating and what's possible, could you provide an estimate of how things are progressing and what you're seeing at ERCOT? You mentioned the 500 megawatts. When did those enter the queue? We have some power coming online in 2026. Just an overall update on how we should consider the availability of power from your growth pipeline.

Sure. Let me provide some details regarding the sites that are pending final ERCOT approval. These projects have been in the queue for some time, and all load studies and related documents have been submitted. The timing largely depends on the specific business operating model of the transmission distribution service provider involved. For Colchis, we're expecting to energize by 2028. We have already made a deposit to support the work required by our TDSP, AEP, which is confident about progressing with construction in anticipation of having that site energized by 2028. Regarding Mikeska and McLennan, we have signed interim FEAs, which are required by the TDSP, Oncor. We have also made a deposit, and construction on the Oncor side is expected to begin once we receive the final ERCOT approval, which we are eagerly awaiting. For Milsing, we have not paid a deposit yet as we are collaborating with a different TDSP, and their process is somewhat different. Overall, while it's challenging to predict specifics with ERCOT, we are optimistic about the timelines based on the feedback from the TDSPs and our assessment of when those sites should become available.

Okay. Super helpful. And then on the optionality of the 56 megawatts, I think you mentioned potentially maybe offering your own AI cloud services. Could you talk a little bit about how we're thinking about that in terms of just bringing on another customer, maybe there's an option that could be extended? Just how we should think about that 56 megawatts? And maybe around the timing, is this something we want to kind of have buttoned up in the next 12, 18, 24 months? Or hey, it's out there and time is on our side?

So the answer is it depends. I think we've had a lot of questions and interest around the idea of owning and operating our own GPUs and then selling the compute to an offtaker. I think in general, we have been progressing slowly on that front because we want to make sure we're getting the best risk-adjusted returns for the megawatts we've got. So obviously, you can produce numbers that are higher on the revenue side if you're selling compute, but you're taking on a whole bunch of risks, much larger financing risks, GPU life cycle obsolescence risk, etc. I do think a key to making that business very attractive would be to lock up a long-term offtake with a highly credible counterparty for the compute. So we've seen those deals. We're looking at them. Candidly, I think the numbers we signed on our lease with AWS are better. I think we will probably both make more in terms of profits and with much, much, much less risk. So it still remains to be seen from our perspective what the best use of a megawatt is to make the most money, but we're in very active discussions in exploring all available business models. And obviously, as we sign up new 1-gigawatt sites, we're going to have a lot of optionality as things progress. As it relates to the specific 56 megawatts, I'm highly confident we will have some sort of deal there pretty soon. There is a lot of interest, both on using that capacity to operate our own GPUs and sell compute as well as have it leased on a colocation basis. It's fair to say that this market is literally getting more frenzied by certainly by the week, if not the day. So rental rates on leases are going up rapidly. The level of interest is overwhelming. And so from our perspective, we're spoilt for choice. We've put ourselves in a very advantageous position. And so depending on which deals we think will produce the best risk-adjusted returns, that's how we'll proceed. But I do think the 56 megawatts there as well as the 100 megawatts at Stingray, the 70 megawatts at Reveille will all be taken up. If this market level of interest continues, we will not have an available megawatt. We have multiple parties interested in all those sites and locking them up as soon as possible.

Okay. Congrats on the AWS announcement.

Our next question comes from Andrew Beale with Arete Research.

Can I just ask, what are you thinking about the design of Colchis? And what do you think the likely CapEx of that as the greenfield will be per megawatt? And just thinking about ERCOT approval, can you talk about what getting the Google, Fluidstack and AWS leases does in helping your approvals at the other sites, such as the 3Ms? And how much difference partnering with AEP makes on that approval front?

Thank you for the question. Predicting the budgetary cost at Colchis is somewhat challenging at this stage, especially since we are still in the early stages of the acquisition. We're starting to engage in exploratory discussions with potential partners interested in colocation, given the scale of the project. In the meantime, we will deploy capital expenditures for the minimum requirements at the site, including fiber, substations, land, and water sourcing. I anticipate our build costs will align with what we've previously experienced at other sites, which typically range from $9 million to $11 million per critical IT megawatt for similar colocation setups. However, we must consider the potential for inflation, price fluctuations, and supply chain issues. I don't foresee any significant changes in cost per megawatt aside from the usual factors over time. We plan to provide more details in the coming months and quarters. With power availability expected in 2028 and the scale of Colchis, we aim to secure a partner soon, as this project entails a significant construction timeline and obligation. I don't have any reason to believe costs will vary. Could you please remind me of your second question? I got a bit lost.

Just about signing these leases with Google and AWS. How does it help your 3Ms and other approvals?

Thank you. Yes, there's huge benefits to these partnerships. I think, again, up until a few months ago, I can't tell you how many times we heard no one's ever going to sign at those sites. No one's ever going to sign with a former bitcoin miner, at least not a traditional hyperscaler. That discussion is now over, obviously. And it probably won't be for us. It will be for others as well as other deals get signed across the ecosystem. I think every deal adds incremental credibility. We deserve a lot of credibility, anyone that got to know the quality of our team, their experience, the things they've built in the past. And just looking at Cipher's own track record, if you took the word bitcoin out and just said our team has delivered 5 data centers on time and on budget in this exact geographical region, there would be no reason to doubt what we say. It's just the traditional bias from incumbent industries against the word bitcoin. So I think every deal adds credibility with everyone, deals beget deals. And I talked about this a fair amount about striking our initial deals focusing on the quality of the counterparty and setting our business up as a franchise such that we can extract the most value from the entire pipeline we've got. And I'm happy to say that that's exactly what we're seeing. So every conversation gets a little bit easier, and we have a lot more credibility on new leases with regulators, with transmission distribution service providers. And truthfully, that feedback I mentioned in the case of Colchis, which is actually scheduled to go out shortly, that's what matters to ERCOT, right? So having more credibility and having money invested in the space and being a credible counterparty makes a transmission distribution service partner want to move forward on your project and spend their own money because they're more likely to get paid and the same on the ERCOT side. So all these things beget more success, and that's probably the biggest reason for optimism around here these days.

Our next question comes from Michael Donovan with Compass Point.

And congrats on the progress. I guess just in terms of supply chain, what are you seeing in terms of constraints for long lead assets?

Yes, we've discussed this over the years, and we typically work backwards to establish a timeline based on long lead time items. Generally, this focuses on getting the substation in place. From there, on the HPC side, it varies a bit depending on site-specific design needs driven by tenant requirements. As a broad rule, if tenants require backup generators for necessary uptime, those usually become the next significant factor in the timeline. I believe we have a strong track record in this area. Our construction team, which includes experienced members from Vantage, Whiting-Turner, Google, and Meta, is skilled at procuring all necessary items for these data centers and maintaining relationships across the supply chain. For instance, regarding Barber Lake, over 85% of the equipment, including all long lead time items, is already secured. This process will remain consistent across each build specification. Currently, the main risk for us, as well as anyone signing these deals, is ensuring construction and financing are delivered on time. Our team has demonstrated excellence in this area, and I fully expect we will outperform others in the industry regarding on-time delivery. While the supply chain situation is constantly evolving, we believe we are well positioned, and we have strong confidence in our aggressive timelines for the projects we have.

That's helpful, Tyler. And then I guess my second question is a bit more esoteric. So I'm hearing discussions about sites being linked up to, let's say, you have a 500-megawatt site here, 500-megawatt site there to link them up to deliver a 1-gigawatt campus for a specific workload. Are you hearing more of these types of discussions? And could we theoretically think of the 3Ms coming together for one large 1.5 megawatt or gigawatt campus?

I believe it depends on how the market progresses. Clearly, many hyperscalers aim for redundancy in data centers located in close geographical proximity. Currently, we have about a dozen data center sites concentrated in West Texas. However, I don't consider the three facilities as being close enough together to link them in the current structure. It's advantageous that they are not too far apart, but I’m not sure people would categorize them that way. While this phenomenon exists, I wouldn't necessarily group our sites in that way. There are significant efficiencies gained from having a workforce in the same area, yet our sites are not typically just 10 miles apart. For example, Colchis is approximately 80 miles from Barber Lake. Overall, our customers express interest in discussing the possibility of building their own availability zone, but we haven't progressed enough to determine which sites would be dedicated to a specific tenant.

Okay. I appreciate that, Tyler. This is my last question, I promise. We are making great progress at the edge of the earth. What should we consider regarding areas outside of Texas?

So great question. We are always looking at opportunities. We just happen to love Texas, and it seems that we always find the best opportunities. I do think that part of it is that there's a lot of things. Business is great in Texas. It's a great place to do business. It also has a history of risk-takers and entrepreneurs that want to speculate on early-stage opportunities. I think it echoes oil and gas somewhat in that there are folks that will speculate on grid interconnections and take a risk on being able to get something. And maybe Cipher's secret sauce, to be honest with you, now that we've originated 12 sites down there is that we have a team that has demonstrated excellence at sourcing these sites from what I'll call kind of grid wildcatters or people that are early-stage investors and an interconnection opportunity but are not prepared to develop the site at a high level that would be ready for an end user like a hyperscaler. I would argue that Cipher is basically the only firm. Maybe we have a handful of competitors, but I think we certainly do it best in that we can speak very credibly with that audience that originates these sites and at the same time, go have an all-day technical meeting with our entire construction and operations team with a hyperscaler and impress them as well. And so bridging that gap between, let's call it, early-stage speculation on grid opportunities and then delivering that to the highest quality end user we have in-house. And honestly, that's why I believe we're a tremendous growth stock opportunity. We're not just a basket of assets. The point being we're not going to stop developing these sites. Now to answer your question more directly, however, because I was just saying how wonderful Texas is, yes, we are looking at sites, particularly in PJM. Historically, we've looked at sites all over the world. Often, the economics haven't gotten to a position we like to be in. We do have a relentless focus on risk-adjusted returns here. And so often, things are either too risky to justify the investment or perhaps the price is too high. They're too mature. There's not enough risk that we feel like we can quantify better than others. So we are looking at PJM. That's a market we would like to expand into and stay tuned. I hope that we'll have announcements in the future.

Our next question comes from Mike Colonnese with H.C. Wainwright & Co.

And congrats on the 2 big HPC deals here. Really great to see. I can appreciate the expected delivery time lines you provided with regards to the 2 contracts. But how should we think about the revenues layering in over the course of 2026 and beyond from the 2 agreements?

Yes. So the full delivery of the Fluidstack, Google deal is expected to be completed at the end of September next year, and so rent begins in October of '26. Amazon is, again, getting finalized, but it begins in August of next year. And then there'll be stages, though. The second stage would be closer to year-end of next year.

Got it. And then more of a high-level question, Tyler. In your view, what has changed for counterparties that has accelerated the pace of deal announcements we've seen in the space over the past month or so? It feels like the level of urgency from hyperscalers, neoclouds, and some others has really picked up from where we were just a few months ago. So it would be great to get your thoughts there.

Yes, I can say that in my 25 years of professional experience, I have never seen anything like what is happening in the market right now. As for the reasons behind it, I’m not certain. I’ve listened to the same podcasts as everyone else and heard the CEOs of major cloud providers discussing a supply shortfall. It seems that for larger, diversified cloud companies, it’s easier to forecast long-term capacity needs for traditional cloud services. However, what has taken everyone by surprise is the rapid surge in demand for AI. Currently, this demand is extraordinary, leading to a scramble among companies that underestimated their quick requirements. There’s a competitive race occurring, and every discussion begins with a request for megawatts that are immediately available. This has evolved to requests for availability in 2026 and even 2027. The tone of these conversations is changing weekly, becoming more enthusiastic and the demand is increasing. Lease rates are rising, which is expected in such a market. I’m pleased with the strategic positions we’ve established to secure the strongest possible clients for our business. We now have pricing advantages which can enhance our economics and improve deal conditions. Our first long-term lease, spanning 15 years, with the largest hyperscaler in our sector indicates the power shift towards those holding scarce assets, which we have carefully managed over recent years. I’m not sure if this level of excitement can last indefinitely, but we feel we have a unique insight into the situation. I've been saying this for some time now: the demand is extremely high and continues to grow.

Our next question comes from Joseph Vafi with Canaccord Genuity.

Congrats on all this great progress and welcome on board, Greg, and congrats, Ed, on your retirement. Just a couple here. Just maybe just the most updated thoughts here, Tyler, on your behind-the-meter agreement and what comes next here for Black Pearl given that site and its unique power procurement and the expiration of that deal and then overlay on top of that, obviously, everything going on in the HPC environment. And how does that site evolve from here?

Joe, do you mean Odessa? You said Black Pearl, but our behind-the-meter PPA is at Odessa.

Yes. I'm sorry, yes, Odessa...

Yes, I just wanted to clarify. Apologies for the previous call before the market. So at Odessa, we have a power purchase agreement for 207 megawatts at an exceptionally low rate for electricity at our bitcoin mining facility, which is effective until the end of July 2027. This contract holds significant value as it is quite favorable for us, and we reflect that on our balance sheet. The fixed, low price makes it particularly advantageous. In light of the current high demand for power, we receive many inquiries about possibly converting that site. However, given the incredibly low power costs, bitcoin mining remains a highly profitable option there. While high-performance computing might become an interesting opportunity in the future, we are not hurrying to make any changes since our contract provides strong economics for bitcoin mining. It’s worth noting that the site is located next to a natural gas generation facility owned by Vistra. As our reputation in the industry improves, I anticipate that more prominent companies will look to us for data center partnerships. Should opportunities arise, we would need to work closely with our power provider Vistra and potential tenants. Nonetheless, we are in no rush because the favorable economics are secured for another year and three quarters.

Sure. And then just really quickly, I may have missed it, but this deal with Microsoft. Is it going to be at one particular site? Or is it going to be distributed? I just don't know if I saw that in the press release.

So we haven't done a deal with Microsoft yet. I know it's confusing today because I think...

I'm sorry. Yes, I'm getting them all confused today. There was another one, yes, Amazon. Sorry about that.

But that's fine. Amazon is converting one large site from a bitcoin mining facility to HPC.

And our last question comes from John Todaro with Needham.

Congrats on the lease. The time line seems pretty quick on getting that Black Pearl site for AWS delivered. Just wondering kind of if I'm missing something or the confidence in being able to deliver that?

Yes, our confidence is very high. This lease comes after extensive technical discussions with our team. At that site, we have constructed 150 megawatts with exceptional building quality. Unlike our other sites, which had tighter timelines and utilized more temporary solutions, this site was designed with a long-term perspective for future conversions. I’m pleased to share that most of the site can be immediately reused for Phase 1's 150 megawatts, which is why we have such an aggressive timeline for it. As for Phase 2, it relies on our ongoing conversations, completion of a procurement process, and establishing a supply chain timeline. I’m confident we can meet this timeline, especially since we're reconfiguring a site that was built to a very high standard.

Got it. That makes sense. And then my last one, just when you're procuring a site like Colchis, who are you competing with? Like, obviously, you're signing the major hyperscalers. Are they looking to build out some of their own sites at this point, too? Or is it mostly, I guess, maybe other bitcoin miners you're competing with?

Yes. That's a great question. I mentioned this earlier, but this highlights the often overlooked growth equity aspect of our company. Engaging in such deals requires significant local knowledge and understanding. It's akin to working directly with a wildcatter. The hyperscalers are large institutions that aren't as agile as we are. They're typically accustomed to receiving polished presentations from firms like Jones Lang LaSalle for completed data centers or well-prepared sites. They don't delve into local specifics or navigate the complex situations that can arise in these deals. I would argue we are likely the best, if not the only company with a high level of credibility among that group for closing deals, as well as the ability to connect with hyperscalers. There are additional layers of capital in traditional commodities production that simply aren't present here. This reduces competition at that level and is a key part of our value proposition. Thank you all for joining us today. I'd like to recognize Ed Farrell, who has been my right-hand man since the beginning at Cipher. We've expressed our gratitude internally for his hard work as he transitions from Chief Financial Officer to senior adviser. I want to take this moment to extend a special thank you to him as one of our largest shareholders. We truly appreciate everything he's done, and I can confidently say that we wouldn't be where we are today without his contributions. It’s hard to believe I won’t have him around the office to share Godfather references anymore or to remind me about the business challenges we face. However, we are in capable hands with Greg Mumford as our new CFO. As we move forward with capital raising and optimization, I’m confident in his leadership. So thank you to Ed on behalf of all shareholders, and we wish him a wonderful retirement. Thanks, everyone. We'll speak soon.

Thank you for your participation. You may now disconnect. Everyone, have a great day.