Hội nghị báo cáo kết quả kinh doanh quý 3 năm tài chính 2026 của Core Scientific (CORZ): Doanh thu Co-Location và EBITDA tăng vọt

Kết quả kinh doanh quý 3 năm tài chính 2026 của Core Scientific phản ánh sự mở rộng liên tục của mảng cho thuê chỗ đặt máy chủ (co-location) mật độ cao. Công suất bổ sung từ CoreWeave đã thúc đẩy doanh thu và EBITDA điều chỉnh tăng mạnh so với quý trước, trong khi công ty tiếp tục đẩy mạnh đợt triển khai 530 megawatt đầu tiên cho AMD và tiếp tục thu hẹp hoạt động khai thác Bitcoin.

Các điểm nhấn chính

• Doanh thu cho thuê chỗ đặt máy chủ theo GAAP tăng gần 60% so với quý trước lên 215 triệu USD, nhờ có thêm 90 megawatt bắt đầu tính phí trong quý và sự đóng góp cả quý từ các đợt bàn giao trước đó.
• EBITDA điều chỉnh tăng lên 134 triệu USD từ mức 41 triệu USD trong quý 2 năm 2026. Biên EBITDA điều chỉnh mở rộng lên 57% từ mức 25%.
• Core Scientific hiện đang tính phí CoreWeave cho 485 megawatt tại 5 cụm cơ sở, so với 120 megawatt vào đầu năm 2026. Khoảng 100 megawatt còn lại dự kiến sẽ đi vào hoạt động trong vài tháng tới.
• Công ty đã ký hợp đồng cung cấp khoảng 1,1 gigawatt cho CoreWeave và AMD. Ban lãnh đạo kỳ vọng danh mục này, sau khi được bàn giao đầy đủ và tính phí, sẽ tạo ra 1,8 tỷ USD doanh thu định kỳ quy năm trung bình và 1,6 tỷ USD NOI.
• Tiến độ bàn giao cam kết 530 megawatt ban đầu của AMD vẫn được lên kế hoạch bắt đầu tại Pecos vào nửa đầu năm 2027. Khoảng một nửa dự kiến sẽ đi vào hoạt động trong năm 2027, phần còn lại sẽ được bàn giao trong suốt năm 2028.
• Core Scientific dự kiến sẽ hoàn tất việc rút khỏi mảng khai thác Bitcoin vào cuối năm 2026 và bước sang năm 2027 với tư cách là nhà cung cấp dịch vụ cho thuê chỗ đặt máy chủ mật độ cao thuần túy.

Dữ liệu tài chính cốt lõi

Theo chuẩn GAAP, doanh thu từ các hợp đồng với CoreWeave được ghi nhận theo phương pháp đường thẳng trong thời hạn thuê 12 năm, qua đó trên thực tế phản ánh sớm các điều khoản tăng giá theo hợp đồng.

Hiệu quả kinh doanh và hoạt động

Core Scientific đã đưa thêm 90 megawatt vào tính phí trong quý 3. CoreWeave hiện đang được tính phí cho 485 megawatt trong tổng số 590 megawatt đã ký hợp đồng trên 5 cụm cơ sở. Cơ sở Dalton Giai đoạn 2 là cơ sở CoreWeave duy nhất vẫn đang được xây dựng, với khoảng 100 megawatt còn lại dự kiến sẽ được bàn giao vào đầu năm 2027.

Ban lãnh đạo cho biết CoreWeave đã tự tài trợ phần lớn chi phí xây dựng cơ sở hạ tầng của mình. Các thỏa thuận cũng bao gồm những biện pháp bảo vệ theo hợp đồng tại một số cơ sở nhất định, bao gồm các khoản thanh toán thông qua pháp nhân mục đích đặc biệt (SPV) được đảm bảo bằng cam kết từ khách hàng cuối, các khoản bảo lãnh từ CoreWeave và quyền bảo đảm của Core Scientific đối với cơ sở hạ tầng trung tâm dữ liệu cho đến khi quyền sở hữu được chuyển giao vào thời điểm kết thúc hợp đồng.

Hợp đồng ban đầu với AMD bao gồm 530 megawatt tại 5 cụm cơ sở. Ban lãnh đạo cho biết chi phí xây dựng ước tính vẫn duy trì ở mức 11 triệu USD đến 12 triệu USD mỗi megawatt, trong đó các cam kết về thiết bị và các cam kết khác đã được thực hiện trước khi ký hợp đồng nhằm chốt chi phí và giảm thiểu rủi ro triển khai.

AMD cũng nắm giữ quyền ưu tiên giữ chỗ độc quyền đối với khoảng 1,9 gigawatt công suất bổ sung tại Muskogee, Hunt và Pecos. Phần lớn các cam kết ứng trước sẽ cần được đưa ra trong năm 2027, với hạn chót là vào cuối năm 2028. Ban lãnh đạo cho biết mức giá cho công suất cam kết bổ sung sẽ phản ánh thiết kế đã thỏa thuận, giá thị trường hiện hành và chi phí phát triển, với mục tiêu duy trì tỷ suất sinh lời gần như tương đương.

Core Scientific đã đảm bảo được khoảng 385 megawatt điện nối lưới bổ sung, được chia đều giữa Muskogee và Quận Hunt, dự kiến sẽ sẵn sàng vào giữa năm 2028. Công ty cũng đang đánh giá phương án tự phát điện sau công tơ (behind-the-meter), bao gồm đề xuất kết nối đường ống dẫn khí tự nhiên dài 10,6 dặm tại Pecos. Ban lãnh đạo cho biết hệ thống lưu trữ năng lượng bằng pin sẽ là một phần của bất kỳ giải pháp sau công tơ nào nhằm làm mượt tải IT biến thiên.

Định hướng của ban lãnh đạo

Ban lãnh đạo kỳ vọng doanh thu quý 4 năm 2026 sẽ tăng trưởng hai chữ số so với quý trước khi có thêm các megawatt bắt đầu được tính phí và công suất bàn giao trước đó đóng góp trọn vẹn trong cả quý.

Công ty nhận định vùng thấp của khoảng 30 triệu USD là mức cơ sở hàng quý hợp lý cho chi phí SG&A bằng tiền mặt, có thể thay đổi tùy thuộc vào các khoản đầu tư tăng trưởng có mục tiêu.

Core Scientific dự kiến sẽ thu về khoảng 650 triệu USD tiền mặt từ việc thực quyền chứng quyền Đợt 1 vào khoảng thời điểm đáo hạn tháng 1 năm 2027. Ban lãnh đạo dự định ưu tiên nguồn vốn này để góp vốn cổ phần cho các đợt phát triển trong tương lai khi các quyền giữ chỗ của AMD được thực hiện.

Công ty không có kế hoạch phát hành thêm cổ phiếu hoặc chứng khoán liên kết vốn cổ phần để tài trợ cho giai đoạn đầu của việc xây dựng cho AMD. Trái phiếu dự án vẫn là phương thức huy động vốn dự kiến của công ty, mặc dù ban lãnh đạo cũng đang đánh giá các phương án nợ khác nhằm cân bằng giữa chi phí, tính linh hoạt và mức độ chắc chắn khi triển khai.

Rủi ro và các lĩnh vực trọng tâm

Khả năng huy động vốn vẫn là rào cản trọng tâm đối với việc phát triển cơ sở hạ tầng AI. Core Scientific vẫn cần nguồn tài trợ dự án để hoàn thành các dự án ban đầu của AMD, mặc dù ban lãnh đạo cho biết thị trường trái phiếu dự án vẫn rộng mở và công ty chưa ghi nhận thay đổi nào về các điều khoản hoặc cam kết bảo đảm. Hoạt động thẩm định chuyên sâu (due diligence) đã trở nên toàn diện hơn.

Các quy trình pháp lý và cấp phép tại Texas vẫn là trọng tâm theo dõi của nhà đầu tư. Ban lãnh đạo cho biết Pecos vẫn đúng tiến độ để bàn giao đợt đầu vào nửa đầu năm 2027, trong khi Hunt đã bước vào giai đoạn đánh giá độ ổn định hàng quý. Giấy phép môi trường không khí cho các máy phát điện dự phòng tại cả hai địa điểm vẫn đang được xử lý theo khung quy định cấp phép chuẩn (permit-by-rule).

Việc mở rộng hệ thống sau công tơ cũng đòi hỏi thêm các quyết định và phê duyệt. Core Scientific vẫn chưa nộp hồ sơ xin cấp phép môi trường không khí cho dự án phát điện dự kiến tại Pecos và dự kiến sẽ thực hiện việc này cùng với nhà sản xuất điện độc lập được lựa chọn sau cùng.

Ban lãnh đạo cũng giải trình về sự giám sát chặt chẽ đối với CoreWeave và mô hình neocloud. Công ty bày tỏ sự tin tưởng vào vị thế dài hạn của CoreWeave, dẫn chứng các khoản đầu tư ban đầu đã hoàn thành, hiệu quả kinh tế thuận lợi của các cơ sở và các biện pháp bảo vệ theo hợp đồng. Tuy nhiên, các điều kiện tài trợ vốn và uy tín tín dụng của khách hàng vẫn là những lĩnh vực then chốt được các nhà đầu tư theo dõi.

Điểm nhấn phiên hỏi đáp với các nhà phân tích

• Rủi ro phát triển dự án tại Texas: Ban lãnh đạo nhận định Pecos và Hunt là các dự án có rủi ro thấp nhờ các phê duyệt cấp điện trước đó, tiến độ xây dựng và tình trạng cấp phép hiện tại. Pecos trước đây là một địa điểm khai thác Bitcoin đã được cấp điện.
• Huy động vốn dự án AMD: Core Scientific đang cân nhắc cách thức đóng gói và sắp xếp trình tự tài trợ vốn trên khắp 5 địa điểm và 5 hợp đồng thuê riêng biệt. Ban lãnh đạo cho biết hồ sơ tín dụng của AMD rất thuận lợi để thu hút nhu cầu từ các bên cho vay.
• Phân bổ vốn: Nguồn tiền dự kiến thu được từ chứng quyền chủ yếu nhằm mục đích góp vốn cổ phần cho các đợt phát triển trong tương lai thay vì mua lại cổ phiếu.
• Công nghệ sau công tơ: Công ty đang đánh giá pin nhiên liệu, động cơ pít-tông, máy phát điện tuyến tính và các công nghệ khác dựa trên các cân nhắc về cấp phép, môi trường và chuỗi cung ứng đặc thù theo từng địa điểm. Ban lãnh đạo kỳ vọng các hệ thống pin lưu trữ sẽ đi kèm với bất kỳ giải pháp nào được lựa chọn.
• Cơ hội mua lại: Ban lãnh đạo cho biết ngày càng có nhiều quỹ đất đã sẵn nguồn điện, các cơ sở đang xây dựng dở dang và các dự án đã ký hợp đồng được chào bán do một số nhà phát triển gặp hạn chế về vốn. Định giá đất đã có nguồn điện nhìn chung vẫn tương đương so với 12 tháng trước, trong khi số lượng bên tham gia đấu thầu đã giảm.
• Danh mục dự án dài hạn: Core Scientific đang đánh giá khoảng 2 gigawatt công suất tiềm năng tại các địa điểm mới, dự kiến đóng điện từ năm 2029 trở đi. Công ty dự định đa dạng hóa giữa các bang và các công ty điện lực, đồng thời duy trì tiêu chuẩn cao cho việc phân bổ vốn.

Toàn văn biên bản cuộc họp công bố kết quả kinh doanh

Toàn văn cuộc gọi công bố kết quả kinh doanh

Phần trình bày của ban lãnh đạo

Operator

Greetings and welcome to the Core Scientific Fiscal Third Quarter 2026 Earnings Call.

[Operator Instructions]

As a reminder, this conference is being recorded. I would now like to turn the call over to your host, John Charbonneau, Senior Vice President, Investor Relations. Please go ahead.

Jon Charbonneau

Good morning and welcome to Core Scientific's third quarter 2026 earnings call. Before we begin, statements made on this call other than historical facts are forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. These statements are based on our expectations. Words such as anticipates, estimates, expects, intends, and believes, and similar expressions identify forward-looking statements. These statements are subject to risks and uncertainties that could cause actual results to differ substantially. For more information, please review the risk factors discussed in the company's reports on Forms 10-Q and 8-K filed today with the SEC, along with the press release and slide presentation included in those filings. Our forward-looking statements speak only as of today.

We undertake no obligation to update them to reflect subsequent events or circumstances. Today's presentation is available on our website, investors.corescientific.com. The information on this call is accurate only as of November 5, 2026. Joining me today are Core Scientific's Chief Executive Officer, Adam Sullivan, and Chief Financial Officer, Jim Nygaard. We will hold a question and answer session after management's remarks. Adam, over to you.

Adam Sullivan

Thanks, John, and good morning, everyone. Thank you for joining our Q3 2026 earnings call. Beginning approximately 2 years ago, we announced a series of co-location agreements with CoreWeave, providing for the development at CoreWeave's expense of 590 megawatts. While many questioned the strength of this arrangement and its potential to support our future growth, we executed. Today, we have built 5 data center campuses, and we are billing our customer for a total of 485 megawatts. The remaining 100 megawatts is currently in commissioning, and we will begin recognizing additional revenue over the course of the next few months. In addition, these contracts enabled us to raise over $3 billion to fund our next phase of growth and has established a reputation for delivery that helped us win our latest customer contract.

At the same time, uncertainty around new data center development in Texas, Georgia, and elsewhere makes our completed operating sites already serving customers and our sites in advanced development with secured power and near-term capacity even more valuable. Our execution has delivered tangible results. GAAP co-location revenue increased nearly 60% sequentially, while adjusted EBITDA reached over $130 million, up from $41 million in the second quarter. We expect further growth as the remaining CoreWeave capacity comes online in the coming months. As we approach completion of that buildout, we are also advancing the initial 530 megawatts contracted by AMD. With over 1.1 gigawatts contracted between CoreWeave and AMD and 1.9 gigawatts under reservation with AMD, we have tremendous confidence in our ability to scale.

Investors have had much to assess in recent months, from risks in Texas to concerns about financing and access to capital across the AI industry. Today, I will address those topics directly and explain where Core Scientific stands. I'll begin with this week's election results and their implications for Core Scientific, as regulatory developments have understandably been a focus for investors. Our outlook for our currently contracted capacity is unchanged. We do not expect the election results to affect our existing operations or delivery timelines for capacity under development.

Our confidence rests on our execution, development progress, and our community relationships. As we have brought hundreds of megawatts online under our first customer contract, we have worked closely with local officials, utilities, and community stakeholders. 2 of the campuses supporting our AMD contract already have operating data center capacity for our other customers, giving us development certainty and affording us the opportunity to build on established operations and relationships. We have also advanced our projects through the increased permitting and regulatory review introduced this year. Our sites received conditional classifications consistent with our expectations under ERCOT's batch 0 process, and we continue to work through the remaining requirements. We are confident that our operating facilities and the remaining capacity left to be delivered within the next few months will continue to operate in this new landscape without interruption.

In Oklahoma and Alabama, we expect continued support for responsible data center investment. In Muskogee, Governor Stitt and Governor-elect Ivey have supported our development and expansion plans, recognizing the jobs, investment, and economic activity these projects bring to their communities. In Alabama, Governor-elect Tuberville is supportive of data center development with a focus on protecting prices for local citizens. Core Scientific is fully aligned with their positioning as we funded more than $12 million in infrastructure investments, including transmission upgrades that provide grid hardening protections for the Auburn region.

In Georgia and Texas, we expect the incoming administrations to emphasize responsible data center development. We have operated in both states for years and have established relationships with local officials and utilities. Our approach remains consistent: address community concerns directly and fully fund the infrastructure our projects require. At Dalton, we have delivered over 70 megawatts of billing capacity to our customer and expect approximately another 100 megawatts to come online in the coming months. Construction for our first customer is well advanced, and we continue to coordinate closely with the local utilities and community stakeholders as we complete the remaining capacity and begin to build for our second customer.

In Texas, Pecos remains on schedule for initial delivery in the first half of 2027. Both Pecos and Hunt have secured their stormwater permits, and we expect their backup generators to qualify for authorization under the applicable permit-by-rule framework. These milestones support our development schedules as we advance the remaining requirements. At Hunt, our campus has been included in the November Quarterly Stability Assessment, an important and final milestone on its path to energization.

Responsible development means making tangible investments in our communities. In Dalton, we have invested more than $700,000 in road improvements, and in Muskogee, contributed $13 million towards the utility-owned substation and switchyard supporting our project. These investments reflect our commitment to funding the infrastructure our developments require and supporting the communities where we operate. Regulatory requirements will continue to evolve, and our experience navigating those requirements, together with the progress already made across our campuses, gives us confidence in our ability to execute.

Next, let me turn to financing. Sustained demand for AI infrastructure creates both a significant opportunity and an extraordinary need for capital. Facilities require substantial investment before they generate revenue, so access to financing will increasingly shape the pace of development. We believe securing capital on attractive terms will be an important competitive differentiator. Notably, we have already invested from cash on hand all the equity capital required from Core Scientific for the initial 530 megawatts under the AMD contract. And going into early 2027, our focus is on securing the remaining project financing to support delivery of our commitments. We are progressing through due diligence for potential project bonds and working with leading financial institutions to determine the best financing path.

While project bonds remain our expected approach, we are also evaluating other debt financing options to balance cost of capital, flexibility, and execution certainty as market conditions continue to evolve. We believe our execution on the CoreWeave contract has helped establish credibility with potential lenders and financial institutions, positioning us well to compete for capital. Turning to CoreWeave. The increased focus on financing across the AI industry has brought greater scrutiny to CoreWeave and the neocloud business model. We remain confident in CoreWeave's long-term position in the AI ecosystem. For our facilities, CoreWeave has already funded the overwhelming majority of the buildout, which is nearing completion. We are now billing for 485 megawatts of the 590 megawatts contracted across 5 campuses, up from 120 megawatts at the beginning of the year. Dalton Phase 2 is the only facility still under construction, and we remain on track to deliver the remaining approximately 100 megawatts by early next year.

The substantial investment required to build these facilities has therefore already been made. That upfront investment also brought down CoreWeave's ongoing lease rate, which we believe makes our facilities among the most economically attractive in its portfolio. With its build costs already spent, we believe retaining this capacity allows CoreWeave to preserve favorable economics while continuing to serve end customers whose workloads are already running within our campuses. In addition to those economic incentives, our agreements include contractual protections. At certain campuses, payments are made through special purpose vehicles supported by end customer commitments, with CoreWeave remaining the ultimate guarantor. We also retain a security interest perfected by UCC filings on our data center infrastructure until ownership of the infrastructure transfers to Core Scientific at the end of the contract.

We believe CoreWeave's scale, expanding customer base, and strategic relationships position it well for the long term. Our focus remains on delivering and operating the infrastructure that supports its growth. Our execution on the CoreWeave buildouts played a central role in earning AMD's business and securing the initial 530 megawatts contracted across 5 campuses. AMD could see the results firsthand, operating facilities across multiple locations and a team that had delivered complex infrastructure at scale. That track record, together with the quality of our power and real estate portfolio, gave AMD confidence in our ability to support its deployment plans.

AMD's initial commitment builds on our established design, and we continue to work closely with its team on co-design initiatives to support future requirements as its technology roadmap continues to evolve. Though they represent less than half of our total contracted capacity today, the AMD contract directly or through credit support represents over half of our contracted stabilized NOI. Delivery of that 530-megawatt commitment remains on track to begin at Pecos in the first half of 2027, with approximately half of the 530 megawatts expected to come online next year and the balance throughout 2028.

Beyond the initial 530 megawatts, AMD holds exclusive reservation rights for approximately 1.9 gigawatts of additional capacity across Muskogee, Hunt, and Pecos. Commitment deadlines generally fall on a quarterly basis, with the overwhelming majority of these advance commitments required to be made in 2027 and the final deadline at the end of 2028. Under our agreement with AMD, we are not disclosing individual deadlines. For any additional capacity AMD commits to, we expect pricing to reflect the agreed-upon basis of design, prevailing rates, and development costs at that time, with the objective of maintaining substantially similar returns.

We expect to support those commitments through a combination of incremental grid-connected power, capacity progressing through load studies, and behind-the-meter solutions. This includes approximately 385 megawatts of incremental grid-connected power already secured, split equally between Muskogee and Hunt County, expected to be available by mid-2028. Since signing the initial 530 megawatts, we have worked with AEP to conduct due diligence on multiple vendors for behind-the-meter solutions. In parallel, we have advanced development of a dedicated natural gas pipeline interconnection at Pecos.

The proposed 10.6-mile pipeline would connect the site to the existing natural gas transmission infrastructure, supporting potential on-site power generation and unlocking additional capacity for future expansion. This work positions us to move forward as conditions become more certain and our plans align with AMD's needs. Before I turn the call over to Jim, let me put our growth opportunity in perspective. We have approximately 1.1 gigawatts contracted across CoreWeave and AMD. Once fully delivered and billing, we expect that portfolio alone to generate $1.8 billion in average annual run-rate revenue and $1.6 billion in NOI. Delivering that capacity remains our priority. We are also simplifying the business to focus our resources on more attractive, long-term growth opportunities. We expect to complete our exit from Bitcoin mining by year-end.

Beyond our existing contracts and AMD's reservation rights, we continue to evaluate approximately 2 gigawatts of potential capacity at new sites, with power energization expected in 2029 and beyond. The current environment is also creating a broader range of acquisition opportunities. In recent months, we have seen more assets offered for sale, from powered land to facilities under construction to projects with customer contracts already in place. The opportunities right now are abundant. Our operating scale and demonstrated execution give us an advantage in evaluating those opportunities. With a business already generating substantial revenue and adjusted EBITDA, we can assess what an acquisition would add to our portfolio and how effectively we could develop, operate, and finance it.

The right opportunity could accelerate delivery, reduce development costs, or add contracted revenue. We intend to be selective, pursuing acquisitions that meet our return requirements while preserving the capital and management attention needed to deliver our existing commitments. Our immediate focus is on completing the CoreWeave buildout and delivering the AMD program while positioning the business to expand our site portfolio and our customer base. We have substantial growth already under contract through 2028, the capabilities to execute on it, and opportunities to extend it even further. We are excited about where Core Scientific is today and believe the broader market backdrop will open up new opportunities for our business. With that, I'll turn the call over to Jim.

Jim Nygaard

Thanks, Adam. We continue to execute on our co-location buildouts in the third quarter, bringing another 90 megawatts into billing. Those deliveries, together with a full quarter contribution from capacity delivered in the second quarter, drove a significant increase in our financial results. GAAP co-location revenue grew nearly 60% sequentially to $215 million, while adjusted EBITDA more than tripled to $134 million from $41 million in the second quarter. This also drove a meaningful expansion in adjusted EBITDA margin to 57% from 25% last quarter. Looking ahead, we are expecting a double-digit sequential increase in fourth-quarter revenue as we continue to bill for additional megawatts and benefit from a full quarter of revenue from prior deliveries. As a reminder, under GAAP, revenue from the CoreWeave contracts is recognized on a straight-line basis over the 12-year lease terms, effectively pulling forward contractual escalators.

Moving on to our Bitcoin mining business, we continue to wind down operations during the quarter, with nearly 50% fewer miners online at the end of September than at year-end 2025. We expect to complete the wind-down during the fourth quarter and enter 2027 as a pure-play high-density co-location provider. On the expense side, third quarter cash SG&A was approximately $28 million, slightly below the second quarter, which included 1-time professional fees related to our bond issuance. While we are not providing formal SG&A guidance, we continue to view the low-$30 million range as a reasonable quarterly baseline, with some variability as we make targeted investments to support growth.

Turning to liquidity and financing, we ended the third quarter with approximately $1.7 billion of liquidity, comprised of approximately $800 million in cash, approximately $840 million in restricted cash and debt service reserves, and $100 million of revolver availability. We included restricted cash in this measure, which consists of funds held in escrow for certain utility and development costs. Liquidity declined approximately $1 billion from the second quarter, reflecting $540 million in capital expenditures primarily related to AMD sites and approximately $440 million for the Polaris acquisition. With our equity contribution to the initial 530-megawatt buildout already invested, our financing efforts are focused on the remaining capital needed to complete these projects.

Project bonds remain our expected approach, and we continue to advance that process with our financing partners as we look ahead to 2027. We are also evaluating other debt financing options to secure an attractive cost of capital as market conditions evolve. We do not expect to issue additional equity or equity-linked securities to fund the first phase of the AMD buildout. We also currently expect approximately $650 million in cash proceeds from the exercise of our Tranche 1 warrants around their January 2027 expiration, further supporting our liquidity.

Beyond our existing CoreWeave and AMD commitments, we will maintain a high bar for capital deployment, including new site acquisitions and development of uncontracted capacity. We intend to preserve our financial flexibility to pursue attractive opportunities while prioritizing delivery of the projects already underway. From the $1 billion financing facility announced in March to our $3.3 billion senior secured bond issuance in April, we have consistently demonstrated our ability to secure capital to support our growth. Our focus today continues to be on execution at every level in our organization to ensure the efficient deployment of that capital for the long term. With that, I'll hand the call back to the operator for Q&A.

Operator

[Operator Instructions]

Our first question comes from the line of John Todaro with Needham & Company. Please proceed with your question.

Jon Charbonneau

Hey, guys. Thanks for taking my question and congrats on all the progress thus far. I guess the first one, just following the election, I mean, it sounds like there's a lot of confidence, but just I guess if we could drill into a bit more, is there kind of any risk to the PUC in December, any delays to kind of that final baseload review or any expectations there?

Adam Sullivan

Yes, I appreciate the question, John. I mean, I think that the differentiating factor for our sites is really a few fold. The first is just the status of where those contracts are, of where those buildings currently sit, and the status of our development. You know, Pecos, I'm sure you've seen the videos, you know, we have a building in place, chillers set, we're working on the interior there. That's the site that's going to come online in early 2027. And as a reminder, that was a Bitcoin mining site, so a site that had already been energized.

So an extraordinarily low-risk proposition in Pecos. Hunt, we feel very similar. A site that was approved for power back in 2024, looking at what's coming up in November related to the upcoming study that ERCOT will be performing, you know, that's really the final stages. You know, it generally happens 1 to 2 quarters before energization. So we have a lot of confidence in the delivery of that site as well, also making tremendous progress on the construction side. We view our 2 sites that are in Texas today that are under development as very low risk.

Jon Charbonneau

Great. Exactly what I was hoping to hear. And then second question, just on the debt financing you guys talked about for the AMD leases, is there just any gating items to maybe call out to getting that done?

Jim Nygaard

So we're in the process, as we've mentioned in prior comments, of optimizing the package of financing that we are intending to bring to market. Keep in mind that we have 5 different locations that we're building at. There are technically 5 separate leases, so we do have some options as to how we package and sequence. And that's what we're trying to do as we speak. So as you can imagine, we've been working with our financing partners and the rating agencies, and the liquidity that we've been able to establish on our balance sheet has afforded us the flexibility to do that in a very disciplined manner.

Phần hỏi đáp

John Todaro

Great, understood. Thank you, gentlemen. I appreciate it.

Adam Sullivan

Thanks, John.

Operator

Thank you. Our next question comes from the line of Darren Aftahi with Lucid Capital Markets. Please proceed with your question.

Darren Paul Aftahi

Yes, good morning. Thanks for taking the question. Question on AMD and the reservation rights. If, you know, politically the wind blows a different way, are there adjustments on those commitment deadlines in 2027 and 2028? And then I guess just back to Texas with your 2 sites, and I appreciate all the color, you know, if the governor was to kind of put in legislation after the fact when she takes office, is there anything that could be sort of clawed back that would delay the process? And I guess when we were at your site visit, you had your lobbyist there. I mean, any kind of real-time thoughts on what the political landscape in Texas might mean in a more draconian scenario? Thanks.

Adam Sullivan

Yes, no, thanks, Darren. Appreciate the questions. As it relates to the reservation rights, I think from our expectation, you know, based on the pathway over the course of 2027, there are some chunkier deadlines over the course of 2027 that we expect to see AMD exercising for additional capacity. And, you know, I think from the standpoint of any potential delays as it relates to potential political headwinds, obviously we're in daily dialogue with the AMD team. So we're working hand in hand across any headwinds that we're facing anywhere across our portfolio.

In Texas, I think the great part for us is, you know, in Texas, things are never done retroactively. You know, it's a business-friendly state. And given where we stand with our permits and where we are at with construction and energization, we have very strong confidence in our ability to deliver those sites on time based on our existing schedules.

Darren Paul Aftahi

Thank you. Appreciate it. Thanks, Adam.

Operator

Thank you. Our next question comes from the line of Brett Knoblauch with Cantor Fitzgerald. Please proceed with your question.

Brett Knoblauch

Hi, guys. Thank you for taking my question. On the AEP expansion, obviously there's a lot of power and a lot of that's coming or is expected to come from behind the meter. Have you guys filed for the necessary permits for the behind-the-meter generation yet? And is that something that you think you can do once maybe the permit freeze ends post completion of the audit? And then as a follow-up question, we get it kind of out of the way here, the warrants that exercise in January, what is the current game plan for the cash influx you're about to receive there? Would you consider repurchasing shares so it's non-dilutive, or would you use that to maybe go fund out as an equity investment for additional data center expansion? Thank you.

Adam Sullivan

Yes, thanks, Brett. I'll take the first one, then I'll hand it over to Jim for your question related to the warrants. You know, as we are going through this process behind the meter, you know, there are a few necessary steps. We obviously talked about bringing the lateral pipe to our site, the 10.6-mile pipe, to allow for the opportunity to build a power plant on site. You know, part 2 of that, which is related to your question related to air permits, we have not filed for the air permits. You know, that's going to be done in conjunction with the IPP that we ultimately choose. We have a narrowed list of IPPs that we're considering here as part of the ongoing process, not only selecting IPPs but also of selecting the generation units that we'll be using on that site. Both of those are important, necessary next steps, and we're working hand in hand with AEP today to choose who that right partner and the right equipment that will be used at that site. As it relates to the warrants, Jim, I'll let you take that question.

Jim Nygaard

We do have liquidity coming in from the exercise of the cash-only warrants in January. It is our expectation that we will prioritize that capital as we discussed in prior public comments, that it's primarily going to be diverted to our proactive builds going forward. We expect the reservations will be exercised over time, and we're going to need that capital to effectively deploy as equity in those sites. So that's going to continue to be the priority as to how we think about capital allocation.

Brett Knoblauch

Perfect. Thank you, guys.

Adam Sullivan

Thanks, Brett.

Operator

Thank you. Our next question comes from the line of Jonathan Petersen with Jefferies. Please proceed with your question.

Jonathan Petersen

Oh, great, thanks. I'm curious if we could, maybe, Jim, get a little more color on what you're seeing in the project financing market to finance some of the AMD builds. What sort of change in terms of covenants that they're looking at, kind of loan-to-cost levels that lenders are willing to lend on today? Are you seeing any of that change, or does it just kind of feel paused until next year and maybe interest rates settle down?

Jim Nygaard

The first thing I would say to address your question is that the project bond market is absolutely open. I would tell you, just in general, as the market has continued to evolve, the amount of diligence and questions that are asked have, I think, become a bit more comprehensive. So there's a strong desire to really nail down a lot of the details that are behind these builds, as everyone knows these are complicated projects. And, you know, everything from the investor standpoint, all of the constituents involved in that process want to make sure that there's clear line of sight on the variables that matter.

There really has not been any change that we've seen by way of terms or covenants. And at the end of the day, what we're really excited about is aligning with AMD. That was one of the key decision-making criteria that we focused on when deciding to go with AMD earlier in late summer is the strength of their credit profile that they bring to this exercise. They're very early in their CapEx deployment cycle. There's not a lot of AMD paper out there for what we're about to do, and the market has a strong appetite for it. So we're very confident in what we're about to bring to market here.

Jonathan Petersen

Okay, that's great. It's great to see the recurring revenue line move up and the adjusted EBITDA margins also trend higher. As we think over the next few years, as you start to stabilize all these projects, where should we think about EBITDA at the corporate level margins trending toward?

Jim Nygaard

Yes, we have not provided specific EBITDA level guidance at this stage. As we continue to start 2027 with a very clean P&L, we're very excited about because we will not have mining in our numbers anymore. And I think it's going to be a great fresh start for investors to understand what a true co-location P&L will look like. What we have guided to is directionally on the revenue and NOI potential that are behind the contracts. And we've also indicated a rough corporate SG&A cash spend that's in that $30-ish million range per quarter. So when you do the math on both of those, you can start to coalesce around what EBITDA would look like for the organization on a consolidated basis.

Jonathan Petersen

Great. Thank you, guys. Appreciate it.

Operator

Thank you. Our next question comes from the line of Nick Giles with B. Riley Securities. Please proceed with your question.

Nick Giles

Yes, thanks, operator. Good morning, guys. One of your peers yesterday announced a deal with 1 of the frontier labs using kind of tenant-brought modules, and I was just curious if this is a structure that you've explored or kind of what shifts you're seeing more broadly in design from a tenant perspective. Thanks.

Adam Sullivan

Yes, I'll answer the first part of the question, then I have Matt Brown, our Chief Operating Officer here with me as well. The first thing I just want to say before handing it over is just that was an incredible deal that Riot announced yesterday. We're incredibly excited about what they're building out in Corsicana, and I think that's a pretty big new high watermark for the industry. But Matt, I'll let you talk about the tenant-brought modules.

Adam Sullivan

Yes, I think from what you can expect from our standpoint is we're working really closely with AMD on future designs. And you should expect that some of those same customers are going to be tenants in those facilities and also at the table kind of working through our design program sort of collaboratively. So in terms of how we view our program, as it relates to tenant-brought modules, it's not something that we've explored a lot. We have a lot of experience in this type of infrastructure as a Bitcoin miner. That was a pretty prevalent infrastructure architecture during that period of the company's history. But we'll continue to just work through with AMD and our customers, and if a customer comes to the table that wants a really purpose-built design that gets to lowest cost per token, it will be something that we entertain, and we'll just kind of continue to work through that with them.

Nick Giles

Makes sense. Thanks, guys.

Operator

Thank you. Our next question comes from the line of Ben Sommers with BTIG. Please proceed with your question.

Benjamin Sommers

Hey, good morning, and thank you for taking my question. So I wanted to ask a little bit about what you're hearing for Calvert City and Grand Forks, you know, given that there are some political headwinds or just political commentary going around, and given that those sites have approved power today, have conversations there changed at all over the past couple months?

Adam Sullivan

No, the conversations have not changed over the past few months. You know, both of those sites have hurdles for us to overcome to build data center capacity there. So we're still working through some of those hurdles as we look to bring new sites to market. But I think both of those sites are interesting, are tight on power, right? Kentucky in particular, with a large power allocation from TVA. That's very rare, and that availability is rare in the market. So we have a lot of confidence in when we're able to bring those sites to market, they'll be attractive.

Benjamin Sommers

Super helpful. And then just 1 more, if I could. Adam, you made a comment on just seeing more assets being offered for sale over the past few months. Just kind of curious on what you think is driving that and how that impacts your growth capacity plans longer term.

Adam Sullivan

Yes, I appreciate the question. I think the overarching theme right now in the market is capital is the throttle on all new development. Capital will be the throttle to bringing any compute online. And the precursor to that is building new data center infrastructure. And what we're seeing in the market today are not only 1-off data center development, but also some private equity-backed data center companies who are just running out of capital. And I think this is where we talk about before organic growth is always going to be priority #1 until you get into a market where you're facing headwinds. That headwind today is capital availability. And that's the reason why we're seeing powered land sites, as well as even sites with customer contracts, going up for sale today.

It's really just the amount of headwinds as it relates to getting financing in-house and where your scale currently stands. And we're in a great position, significant amount of NOI generated on a quarterly basis, continue to scale our platform, and there might be opportunities for us to execute on that will be accretive to our business.

Benjamin Sommers

Super helpful. Thank you for taking my questions.

Operator

Thank you. Our next question comes from the line of George Sutton with Craig-Hallum Capital Group. Please proceed with your question.

George Sutton

Thank you. Just further to the question on the value of powered land, I mean, the generic thought process is this is going to make powered land more valuable, but obviously the financing constraints kind of are a big offset to that. So as we think about valuation opportunities out there, what would you anticipate?

Adam Sullivan

It's an interesting situation. You know, the demand is incredibly high for these sites, but it's also limited by capital and the companies that have available capital at any given point in time. So for us right now, the valuations that we're looking at on powered land especially are actually generally still in line with where they were over the course of the past 12 months. It's just demand is getting higher, but the amount of bidders on those projects is getting smaller. So it's an interesting dynamic that we're going through right now. But I think for us, the way we're thinking about it is given market headwinds, how can we position ourselves better? This team was built to ride through difficult times. And I think some of that we're seeing as it relates to capital availability for other companies. So we feel good about our opportunity to capitalize on what's in front of us today.

George Sutton

So you've obviously got incredible growth opportunity in the short to midterm, but as we think about the longer term pipeline, how much does the election and some of the challenges out there change your thought process relative to breadth of things you're looking at, or does it become more focused? Just curious to hear there.

Adam Sullivan

Yes, it's a great question, George. I mean, how we're thinking about our long-term development is still along the same lines as it was 3 days ago prior to the election. We have opportunities to build in front of us. We have locations that we are specifically choosing based on the political landscape, local community relationships. And really, we have a lot of experience in this compared to other traditional data center developers. You know, at 1 point, we were the largest publicly traded Bitcoin mining company. We were in a lot of locations, a lot more locations than other folks in the industry.

And we had to deal with the headwinds of being what I'll call all the black sheep coming into a local community. We did a fantastic job at engaging with the local community, building relationships, and building, frankly, in new locations that we were less familiar with. So we feel very good about the long-term prospects as we look at the sites that we have under exclusivity, how we're going to be executing on those going forward. We know that step 1, going into any new location, strong engagement at the local level is priority #1, working with state officials as well, and then making sure that we're hiring local and building a true base in that community. So we've run this playbook before. We feel great about where we stand, and we look forward to building new locations in our portfolio.

George Sutton

Great. Thank you.

Operator

Thank you. Our next question comes from the line of Michael Donovan with Compass Point. Please proceed with your question.

Michael Donovan

Hi, guys. Thanks for taking my question. Congrats on the progress. I thought we could hone in a little bit more on behind the meter in terms of actual tech solutions that you're using. Are you looking more towards recips or batteries? And really how should we think about utilizing 1 or the other, and is there a difference yield on cost that we should expect if 1 or the other is used?

Matt Brown

Yes, I can take the technology aspect of the question here. Every site and portfolio will dictate the project, and the location of the project dictates what the solution ends up being. So whether it's fuel cells or recips or linear generators or components, every project will have its own nuances that drive to the solution. Some of those are going to be sort of related to emissions and what we could get permitted to do. Some of that's going to be related to other environmental aspects of the site. That sort of drives to that. So we want to select the right technology for the right location, and the right project is sort of our approach to it. You can expect that we've evaluated all the technologies overlaid against all of our existing programs and that we understand which ones are optimal for which use cases, including supply chain timing, etc.

So I think we have a pretty good grasp of kind of what directions we would go for each of our sites. And all the solutions, all the behind-the-meter solutions, are using some sort of BESS-type battery. Doesn't matter if it's recips or what. The BESS and the inverters that go with the BESS solution are really required for some of the load smoothing to handle some of the variability in IT workloads, so you expect there's going to be a battery component to anything we would deploy.

Michael Donovan

Great, appreciate that. Best of luck moving forward.

Operator

Thank you. Our next question comes from the line of Stephen Glagola with KBW. Please proceed with your question.

Stephen Glagola

Hi, thanks for the question. I had a TCEQ follow-up from an earlier question. Adam, what TCEQ permits do you have outstanding for Hunt and Pecos? And then assuming the batch 0 data center audits are completed on schedule in December while Abbott's still in office, how do you expect the remaining TCEQ permitting process for Pecos and Hunt to proceed in that month or so period before, I think it's like the January 19 transition to the governor-elect? Thank you.

Adam Sullivan

I appreciate the question, Steven. We're currently working through the air permits at both Pecos and Hunt with TCEQ. Our expectation is that we'll be going through a permit by rule, which is the ability to get a permit essentially by law in the state of Texas for those air permits if we meet certain qualifications. Obviously, if there's any significant headwinds, there's always the availability to do design changes as it relates to your backup generators, which is the only thing that that air permit is for. And so there's always ways for us to potentially mitigate any potential headwinds as it relates to that.

Stephen Glagola

All right, thank you.

Operator

Thank you. Our next question comes from the line of Kevin Deedy with JonesTrading. Please proceed with your question.

Unknown Speaker

Hey, Adam. Thanks for having me on. Nice to talk to you again. Can you just sort of frame up your perspective on load studies across Oklahoma and Pecos 200 megawatts? I know you said that it's a little risk there, but it's a study load, it's not baseload. Frame that up against your expectations that AEP has in the timeline.

Adam Sullivan

Yes, absolutely, and I appreciate the question, Kevin. You know, I would label Pecos as load or new supply that, you know, we look forward to having a few years from now. Our plans have included really the next 2 to 3 buildings in Pecos to be behind-the-meter developments, so we've been progressing that. Obviously, we're building that and not the lateral. You know, 1 of the things as we work through on the Muskogee side is that we do have an expectation. You know, right now we have additional power on grid for us to sell to AMD. We have additional power that's working its way through load study.

You know, we're working very closely with OG&E on getting additional power at that location. And we feel very good about that opportunity. So we're looking forward to building the next few buildings in Muskogee on grid. And as we continue through that development, based on the sizing and the timelines that AMD is looking to develop, we might be putting a behind-the-meter building in between potentially power that we already have versus load study power. But that's really yet to be known based on us receiving feedback on that next load study for 600 megawatts total, which would be 400 megawatts of critical.

Unknown Speaker

How easily would behind-the-meter combustion be feasible in Muskogee? I mean, I know you mentioned an almost 11-mile pipe that you'd need in Pecos. What sort of considerations do you have to work around to manage the same opportunity in Oklahoma?

Adam Sullivan

Yes, I think, you know, from the perspective of Oklahoma, the state and local government both have been extraordinarily supportive. When we announced our Muskogee project and the acquisition of Polaris earlier this year, Governor Stitt, in a quote in our press release, had related to looking forward to us not only continuing to build out in the state of Oklahoma, but also building out behind-the-meter generation on a go-forward basis. You know, it's been part of Governor Stitt's bringing more generation to the state of Oklahoma. We look forward to Governor-elect Mazie continuing on with that promise to being incredibly supportive of behind-the-meter generation.

From a technical standpoint, there is more distance that we have to travel by lateral pipe. You know, the great part is there is plenty of natural gas in the state of Oklahoma. So we have a great opportunity to grow and expand that site. We have plenty of land to grow it well beyond what our current plans are if the opportunity arises to grow it to that level. So from the status of Oklahoma, we feel great about doing behind the meter there. And from a complication standpoint, it's not higher risk than what we have out in Pecos, Texas today.

Unknown Speaker

Yes, I'd probably assume that even less risk given, I guess, less demand for space and combustion there. But thanks, Adam.

Adam Sullivan

Of course. Thank you, Kevin.

Operator

Thank you. Our next question comes from the line of Ryan Gravett with UBS. Please proceed with your question.

Ryan Gravett

Great, thank you. Just any updates in terms of build cost expectations for the current AMD leases, including how much is locked in at this point? And then a separate topic, but just on the 2 gigawatts of new site opportunities that you're diligencing, anything you can share at this point in terms of geographic mix, or if there's any portion of that pipeline that you feel is particularly well positioned in light of the election outcomes? Thanks.

Adam Sullivan

Yes, I appreciate the question, Ryan. You know, on a cost-per-megawatt basis, we're still holding true at the $11 million to $12 million a megawatt. So there's no change there. I think the great part for us is, prior to signing the AMD contract, we were out in market making commitments and spending capital to secure the OFCI. And part of that was really to make sure that we were locking in as much of that build cost as we possibly could ahead of signing a customer contract and really de-risk the entire program. So from that perspective, we feel great about our existing cost per megawatt there.

As it relates to the pipeline, we have a strong geographic mix in that pipeline. We were looking at not only developing some of the existing states where we currently reside, but also adding some new states to our map. I think that's going to be part of how we're thinking about our new developments, to continue to diversify on utilities, continue to diversify on states, and really diversifying that mix across all of these different variables is important to building a diversified portfolio. You know, we're excited about what's in our pipeline. We think they're very saleable assets once we're able to bring them to market. And we look forward to updating folks in 2027 as it's related to that.

Ryan Gravett

Great, thank you.

Operator

Thank you. Our final question this morning comes from the line of Brian Dodson with Clear Street. Please proceed with your question.

Unknown Speaker

Hey, good morning. Thanks very much. You know, as you're engaging existing and potential clients, I guess, what is their reaction to rising building costs and rising borrowing costs? Do you think that general terms on contracts in terms of like a yield on NOI will remain consistent, or I guess, what's your viewpoint there for the next 12 months?

Jim Nygaard

I appreciate the question. I think with respect to prospecting going forward and the cost dynamics that are currently present in the market, I mean, certainly, customers understand that we are in an evolving market. You know, financing has become more expensive. I also think there's a fundamental question around basis of design. I think you always have to start with that question when we start talking about the cost to deploy. Unfortunately in our industry, there's a lot of apples and oranges and grapefruits when we're talking about a build. You've got to be mindful of the geographic region that you're building in, and then fundamentally, what exactly are you building for? And they can have dramatically different impacts on that equation.

So look, I think we're going to continue to see openness on the customer front to find ways to cut costs out of builds. And if that means changing the specifications of what we're delivering with fewer nines of performance because we can tolerate less uptime, that tradeoff for cost of deployment might make sense in certain circumstances, depending on how those workloads are allocated. So it's a great question. It's going to certainly vary as you go from customer to customer and the technology that they're wanting to deploy at that location. So unfortunately, it doesn't have a generalized answer other than when costs are going up, customers are inherently going to look for better ways to put that dollar to work of compute to get to the most efficient outcome for their needs.

Unknown Speaker

Great, thanks. I appreciate the color. Thank you.

Operator

Ladies and gentlemen, this concludes our question and answer session and will conclude our call today. We thank you for your interest and participation. You may now disconnect your lines.

This live transcript is auto-generated without human intervention or review.