Cipher Mining (CIFR.US) Q3 conference call: Barber Lake expands lease to $9.1 billion, CEO says it wants to build a highway for AI and collect stable toll fees for decades
The Zhitong Finance App learned that Cipher Mining (CIFR.US)'s revenue for the third quarter of fiscal year 2026 was $27 million, up from $25 million in the previous quarter. For the first time, it confirmed high performance computing (HPC) rental revenue; GAAP net profit was 32 million US dollars, compared to GAAP net loss of 268 million US dollars in the previous quarter. The biggest highlight of the season was the Barber Lake park leasing expansion — contract revenue increased to about $9.1 billion, and the lease period was extended from 10 to 20 years. To this end, the company issued $850 million zero-interest convertible bonds and additional investment. In response to rising construction costs across the industry, CEO Tyler Page responded that the company had switched to a more prudent “design before signing a lease” strategy, and emphasized that “we are building highways and charging long-term toll fees for AI”, maintaining a high level of confidence in demand prospects.
Performance performance
Cipher Mining released its financial results for the third quarter of fiscal year 2026 and held a conference call on Monday. This quarter's revenue was $27 million, up from $25 million last quarter. To confirm HPC rental revenue for the first time, the company also began disclosing it by sector. GAAP net profit of $32 million for the quarter was a significant loss compared to the GAAP net loss of $268 million in the previous quarter, mainly driven by non-cash income of $127 million from warrant liabilities ($151 million in non-cash losses in the previous quarter). Adjusted EBITDA losses narrowed to $10 million, compared to $30 million in the previous quarter. The company sold 632 bitcoins this season, receiving $45.5 million in cash and achieving revenue of $500,000.
Core Focus and Management Response
The central focus of this conference is the trade-off between investment and revenue in the company's projects under construction under construction cost inflation, and the future capital expenditure and return issues extended by this. Around the Barber Lake expansion deal, the company paid an incremental investment of approximately $130 million in exchange for a 10-year lease extension, an increase in contract revenue of approximately $5.2 billion, and improvements to the project's NPV and MOIC.
In response to analysts' questions about cost lock-up and future single megawatt capital expenditure, CEO Tyler Page responded: “Tenants are definitely aware that costs are rising, and they know that if they partner with a company like Cipher Mining, we will finance construction on their behalf”, adding that what they have seen so far is that tenants are “willing to raise rents as financing costs rise.”
CFO Greg Mumford added that there are about 20 to 25 high-yield bonds on the market to finance this type of construction, and “three of the four with the tightest price differences are our bonds”, confirming the value of the company's strategy to target high-credit tenants.
Barber Lake Expansion — From “Recovering to Order” to Exchanging Long Orders
The most important commercial move this season was revising and expanding the Barber Lake lease. The new lease doubles the basic lease period from 10 to 20 years, increasing contract revenue to about $9.1 billion and adding about $5.2 billion in rent; the company also introduced a cost sharing mechanism, which brought in approximately $500 million in incremental revenue over the 20-year lease period.
The cost is the company's additional capital investment: tenant demand evolves in real time, generating about US$409 million change orders (change orders), of which tenants reimbursed about $50 million, and Cipher invested the remaining US$359 million in exchange for extended leases; in addition, due to the evolution of scope, inflation, and rising labor costs, the company also shared the construction costs equally with the tenants.
CFO Greg Mumford settled the bill: without expansion, the company's cost would reach $12.3 million per MW of IT, and the average cost-benefit ratio would drop to about 13%; after expansion, the cost would rise to about $13 million/IT megawatt, but incremental rent would restore the cost-benefit ratio to about 15%, increase the 10-year lease period, and improve credit, NPV, and MOIC. He characterized it as “an incremental investment in exchange for a significantly higher value contract”.
The CEO emphasized that the most valuable part is the direct relationship established with the “world's leading AI lab,” and the company aims to develop multiple data centers for the next few years.
Progress of the three signed parks and disposal in Odessa
The three major contracted parks total 700 megawatts, representing contract revenue of approximately US$16.6 billion over the next 20 years (including total extension of approximately US$28.4 billion), with a weighted average rental period of 17.3 years.
Black Pearl delivered the first batch of capacity ahead of schedule in August, confirmed the first HPC rental revenue this season, and has so far delivered 18 MW of critical IT capacity; the second phase of the foundation and steel structure have been completed, and electrical construction is underway; the first batch of Barber Lake data halls will be delivered in stages from Q4 2026 to Q1 2027; Stingray earthwork and foundation are nearing completion, steel construction is in progress, and the shell is expected to be completed in Q4.
As the last Bitcoin mining farm, Odessa produced around 296 bitcoins this season, and the company continues to evaluate monetizing its “accelerated power supply” value into HPC development land, but it is clear that “no equity will be injected into the site.”
5.9 GW near-end pipeline and new on-site power generation engine
The development pipeline contains approximately 5.9 GW of near-end opportunities, and the total combined capacity is approximately 6.8 GW. The Texas pipeline obtained multiple “Batch 0” classifications. Among them, 1 GW of Colcus received conditional baseload classification. It is expected that it will be finally classified after the December data center audit, and will be called the “Pearl of the Crown” by the CEO.
Reveille and Ulysses have been approved to be connected to the grid and are not part of the ERCOT batch process. The company began construction of lateral pipelines (lateral pipelines) that can support up to 2.5 GW of on-site power generation this season, and will be able to supply electricity as early as 2027. Early commercial negotiations have now been held on a data center lease for about 1 GW on-site power generation. The Mceska, Apollo, and Stingray extensions received PCLR classification (only 11 of about 150 conditional study loads were obtained in the first batch). A new 70 MW Horizon campus has been acquired, a grid-connected agreement has been signed, and electricity is expected in 2029.
Full Meeting Minutes
OPERATOR (Moderator): Hello everyone, thank you for your patience. Welcome to the Q3 2026 Business Update Conference Call. At the moment, all participants are in listen-only mode. Following the speaker's presentation, a question-and-answer session will proceed. To ask a question in this section, press an asterisk and add 11 on your phone. You'll then hear an automated voice prompt telling you've raised your hand. To withdraw the question, press the asterisk again and add 11.
Please note that today's meeting is being recorded. I'll now leave the conference to today's speaker, Courtney Knight, Head of Investor Relations. Please get started.
Courtney Knight (Head of Investor Relations): Good morning, and thanks for joining us on this conference call to discuss Cipher Mining's Q3 2026 business update. I'm attending today with CEO Tyler Page and CFO Greg Mumford. Please note that our press releases and presentations can be found in the Investor Relations section of the company's website, where the conference call will also be webcast simultaneously. Please also note that this conference call is owned by Cipher Mining, and any form of recording or other reproduction without prior consent is strictly prohibited.
Before we begin, I'd like to remind you that the following discussion, as well as our press releases and presentations, include forward-looking statements. These statements include, but are not limited to, Cipher Mining's financial outlook, business plans and goals, and other future events and developments, including statements about the market potential of our business operations, potential competition, and our goals and strategies. The forward-looking statements and risks in this conference call, including responses to your questions, are based on current expectations as of today, and Cipher is under no obligation to update or amend these statements except as required by law.
Additionally, the following discussion may include non-GAAP financial measures. We may use non-GAAP metrics to describe how we manage and operate our business. We have reconciled non-GAAP metrics with the most directly comparable GAAP metrics and recommend that you review these reconciliations, which were submitted along with the performance report issued earlier this morning. Now I'm going to give the call to our CEO, Tyler Page. Tyler.
Tyler Page (CEO): Thank you, Courtney. Good morning everyone, and thank you all for participating today. I'm Tyler Page, CEO of Cipher Mining, and I'm excited to welcome everyone to our Q3 2026 business update call. The past few months have tested our industry, but I wanted to be clear right from the start: Cipher Mining was made for this market. Recent regulatory developments and market fluctuations have spawned quite a few headlines questioning the direction of the industry, but underlying fundamentals, such as demand for computing power, are now only continuously improving.
Tenant demand in our pipeline has never been stronger, and our strategy — which may be easily overlooked in quieter markets — will continue to prove its worth. Our early decisions on partner selection, project structuring, how to navigate the dynamic regulatory environment, and how to deal with our communities were based on these circumstances. Faced with a more demanding financing market and rising construction costs, we have adjusted our style of play accordingly.
Today I'm going to go over these key points one by one, but first I'd like to take a moment at the beginning to remind everyone who we are and what we do. Fundamentally, Cipher Mining identifies undervalued electrified land and invests capital, expertise, and execution to unlock its value through leasing and development. Simply put, Cipher Mining creates value by turning scarce electricity into long-term contract infrastructure. Our model is built on six pillars. We acquire sites by creatively and inexpensively identifying and obtaining land and power contracts. We rent out these spaces by signing long-term contract cash flow agreements with the highest credit quality HPC tenants — with whom we continue to deepen.
We finance the deal — Greg will discuss this in greater depth — to support sustainable growth through pioneering capital markets solutions. We have successfully completed three project-level financings, and the cost of financing each subsequent project is lower. We then use this capital to develop and deliver mission-critical infrastructure, and build the park around the dynamic needs of tenants, with in-house senior engineering, procurement, and construction teams.
We own and operate these sites while generating long-term recurring cash flow; finally, our financing structure allows us to release project equity from stable sites, recover this capital to the parent company, and reinvest in future development. Greg described this flywheel in an earlier conference call, and it's the core of our compound value over time. We've proven this model in the first HPC data centers. We are one of the first companies in the industry to sign leases and work closely with each tenant to refine their design bases (bases of design), which deepens our construction expertise and strengthens our customer relationships. These efforts have translated into three data center campuses, totaling 700 megawatts, all nearing completion and leasing to the world's most advanced technology companies.
Together, they represent $16.6 billion in contract revenue over the next 20 years, equivalent to an estimated average annual net operating income (NOI) of approximately $864 million. All three projects are in the construction phase, Black Pearl has delivered the first batch of capacity, and the remaining capacity of the three campuses is expected to be phased out over the next few months. As the first HPC data centers are nearing completion, we're building on the decisions we've made before and building on them to optimize our future style of play. I'd like to expand on a few of those points now. First and foremost, we always focus on working with the most trustworthy clients, whose strength underpins our ability to finance every project. We continue to deepen these relationships and have conversations with every leading hyperscaler (hyperscaler) and AI lab about our vast site pipeline.
Second, we insist on prudently pursuing leases that are reasonable in terms of risk adjustment and return. This means a strong focus on our long-term co-hosting (colocation) model. We're not Neo Cloud. Neo cloud providers give customers direct access to chips. This model also provides greater opportunities for apparent income, but with it comes a significant increase in risk and financing obligations. Revenue visibility is lower, risk exposure is higher, and profit margins are highly sensitive to changes in spot pricing of capital costs and computing power.
As a long-term co-construction hosting provider, we deliver turnkey data center capacity and let the tenants do the in-house decoration themselves. Our capital is focused on the data center itself, not the chips in it. In return, we sign long-term contracts with reputable counterparties, which gives us strong revenue visibility and more limited risk exposure. In a market where capital may be more difficult to obtain, this difference is critical. Lenders and investors are putting a premium on predictable contract cash flow, and this is exactly what our model is designed to produce.
We believe this has enabled Cipher Mining to deliver one of the most attractive risk-adjusted returns in the industry, and we intend to continue to focus on this strategy. So what are our evolutions? The first change is the way we think about pricing new leases. We evaluate each project by yield on cost (yield on cost) — the ratio of the stable annualized revenue generated by a site to the total cost of its construction. As the cost of capital rises across the market, we are focused on ensuring that returns on new leases keep pace, and we believe this is within reach given the continued growth in demand and the scarcity of usable electricity over the next few years.
The spread between our cost-benefit ratio and the cost of capital is one of the ways we create value for our shareholders, and we focus on protecting this spread. Additionally, we are now investing more development capital at the earliest stages of each project. In the early stages of our transition to HPC data center development, we and our tenants were willing to sign leases before design benchmarks were fully defined to prioritize speed. We are now taking a more prudent approach to aligning design requirements with tenants before signing leases to reduce the risk of major change orders and construction schedule impacts as construction progresses.
In our near-end pipeline, we have been working hard to advance design perfection to 30% to 60% commercial completion before executing leases. This upfront engineering and design work gave us greater visibility into the supply chain and construction costs before committing to the economic terms of the lease. We combined this work with a selective, forward-looking procurement approach to pre-target critical equipment to protect our development schedule and protect Cipher Mining from supply chain availability challenges and cost inflation.
Additionally, we are evaluating various new structures for future leases, including enforcing construction reimbursement agreements with tenants prior to final lease execution. In short, the fundamentals of our business haven't changed. We remain focused on creditworthy tenants, long-term co-built escrow leases, and returns that can justify the capital we have invested. What has changed is the level of rigor we invest in each new project, from how we price our leases to how early we lock in design and procurement. We believe this evolved style of play allows Cipher Mining to grow confidently, running a near-end opportunity pipeline of approximately 5.9 gigawatts.
Now let's talk about our three contracted parks. In these leases, we partnered with leading tenants, and the weighted average tenancy was 17.3 years, representing approximately $16.6 billion in contract revenue; approximately $28.4 billion in contract revenue if all extensions were exercised. These cash flows are long-term, highly visible, and supported by investment-grade credit. As Black Pearl's existing capacity is launched and Barber Lake's first data rooms are gradually being rolled out, these projected cash flows are increasingly being converted into realized revenue.
Now let's take a closer look and take a look at the full picture we've built in our operations, contracts, and future pipeline development, including grid-connected and on-site power generation opportunities. The portfolio currently totals approximately 6.8 gigawatts of capacity. Part of this reflects the HPC capacity we've signed up for, as well as our legacy Bitcoin mining capacity in Odessa. But the vast majority of them are in our pipeline, representing substantial growth beyond existing contracts.
With ongoing site acquisitions and existing site expansions, as well as increasing on-site power generation — our site is particularly suited to this — we are in an excellent position to seize multiple future growth levers. Moving on to the page 7 slide, I'd like to highlight a few key developments this quarter. First, and most importantly, we've revised and expanded our Barber Lake lease. The new lease added a 10-year lease, bringing the site's total contract revenue to $9.1 billion. We've also established a phased delivery plan, and each data lobby will be launched from the fourth quarter of 2026 to the first quarter of 2027. Second, we're pleased to report that we confirmed our first HPC rental revenue in the third quarter at Black Pearl, thanks in part to capacity delivered early in response to tenant requests.
Revenue will continue to rise as we deliver more capacity to Black Pearl, Barber Lake, and Stingray over the next few months. Third, we have made substantial progress on the Texas pipeline, and several pipeline sites have received Batch 0 classification, including Colcus's 1 gigawatt conditional baseload classification. We expect to receive the final classification in December after the data center audit is complete. Fourth, we have begun construction of horizontal gas pipelines capable of supporting up to 2.5 gigawatts of on-site power generation.
These pipelines will connect multiple sites to gas supplies, and may supply electricity as early as the end of 2027. This on-site power generation strategy provides us with an alternative path to electricity beyond the grid and allows us to expand the usable capacity of our existing site. Finally, we're proud to support our communities. Together with Fluidstack and Anthropic, we are committing a total of $10 million to repair and upgrade Colorado City's water infrastructure.
The funds will be used to upgrade municipal infrastructure determined by the city to improve the water supply for residents after last summer's shortages. Moving on to the page 8 slide, I'd like to take a moment to talk about our recent lease expansion to Barber Lake, why we're moving forward with it, and why it's a significant development for Cipher Mining and our shareholders. First, the numbers. The revision doubled the base lease from 10 to 20 years and increased contract revenue to approximately $9.1 billion, adding $5.2 billion in incremental rents. Importantly, every one of Barber Lake's key project metrics, including IRR, MOIC, and net present value, achieved substantial improvements even after factoring in the incremental capital investment required.
While these improved economic conditions are significant, the most valuable part of this deal is our direct and budding relationship with one of the world's leading AI labs. We expect this tenant to become the world's most important co-managed tenant in the next few years, and we aim to develop many data centers for it over the long term. The changing nature of the lease also shows the AI lab's long-term commitment to this particular site, which we see as a strong candidate for future expansion.
Having Barber Lake grow with this tenant is at the core of our long-term ambition for this site. Importantly, this revision also establishes a phased delivery framework linked to our tenant computing power deployment, and we are progressing smoothly according to this plan. The data lobby will be delivered from the fourth quarter of 2026 to the first quarter of 2027, and our tenants are currently actively deploying computing power. Before introducing our portfolio and pipeline, I'd like to take the call to our Chief Financial Officer Greg and ask him to provide more perspective on our expanding opportunities at Barber Lake.
Greg Mumford (CFO): Thanks, Tyler. The deal was a major commercial success. At Barber Lake, we increased revenue by $5.2 billion, extended our lease by 10 years, directly faced a leading AI lab, and implemented a cost sharing mechanism, which resulted in approximately $500 million in incremental revenue over the 20-year lease. In exchange for these improved economic conditions, Cipher Mining chose to invest additional capital into the project. There are two main components to this investment.
The first component is the change order. Our tenants' needs evolved in real time during construction, generating approximately $409 million in lease change orders. These costs should have been borne by the tenant and are reimbursed to Cipher Mining. As part of the revisions, tenants will be reimbursed approximately $50 million, while Cipher Mining agreed to invest the remaining $359 million in exchange for a 10-year lease extension and significant incremental revenue that followed.
We think this is an attractive exchange for recent reimbursements for significantly greater long-term asset value. The second component is higher construction costs, driven by evolving scope, inflation, and rising labor costs. According to the amendments, we split these costs equally with our tenants and recoup our share by leasing at the agreed rate of return throughout the lease term. The right side of the slide explains why we think this is the right capital allocation decision. When Barber Lake was initially underwritten, our guideline was $9 million to $11 million per megawatt of IT.
We ended up financing the construction of the project at $10.4 million per megawatt of IT, supporting an average cost-benefit ratio of approximately 15%. Without this revision, reimbursed tenant-led change orders combined with overall scope evolution and inflationary pressure would have brought Cipher Mining's cost to $12.3 million per MW of IT and reduced our average cost-benefit ratio to around 13%. Instead, we decided to add capital to cover a portion of the reimbursable change orders and additional costs required to complete the project in exchange for improved lease economics and a stronger relationship with the end users of our site's computing power.
With this deal, Cipher Mining's cost increased to around $13 million per MW of IT. But the additional rent restored the project's average cost-benefit ratio to around 15%. At the same time, we added a 10-year lease, improved credit, increased contract revenue by approximately $5.2 billion, and improved the project's net present value and MOIC. This is not just additional capital required to complete the project, but an incremental investment in exchange for a significantly higher value contract.
We believe the additional long-term value we received at Barber Lake significantly outweighed the incremental equity we invested in acquiring it. In that regard, I returned the call to Tyler and asked him to introduce our current portfolio and pipeline.
Tyler Page (CEO): Thanks, Greg. Let's talk about current portfolios, starting with Barber Lake. Construction is progressing as scheduled, and our tenants have begun deploying computing power throughout the facility. We look forward to sharing further progress as we deliver the first data lobbies in the next few months, and the last data lobby for the grid-connected facility is expected to be delivered in the first quarter of 2027. Over the past few months, we have acquired 685 acres of adjoining land to support the campus's expansion opportunities.
Next up are Black Pearl and Stingray. At Black Pearl, we delivered our first batch of capacity early in August in response to tenant requests and confirmed our first HPC rental revenue. To date, we have delivered 18 megawatts of critical IT capacity, and the remaining data hall in phase I is undergoing electromechanical (MEP) renovation. In Phase II, the foundation and steel structure of the building have been completed, and electrical construction is ongoing. We are working closely with tenants to adjust the phased delivery schedule around their computing power deployment plans, and construction is progressing according to these dates.
Additionally, we have acquired 256 acres of adjoining land to support the future growth of the park. Next up is Stingray. Earthwork, leveling, and cushion preparation are nearing completion, steel structure hoisting is underway, and shell construction is expected to be completed in the fourth quarter. We are also continuing site preparation to support future stages of the site expansion project and look forward to providing Stingray's updates as construction accelerates over the next few quarters.
Odessa, our last Bitcoin mining farm still in operation, performed well in the third quarter, producing around 296 bitcoins. We continue to evaluate the possibility of monetizing Odessa's “accelerated power delivery” value into an HPC development site, which comes from two sources: our long-term equipment and site infrastructure, and the remaining term of our PPA. Together, these provide any future developer with significant power delivery time advantages. We look forward to sharing updates as these conversations progress.
Now on to updates to our development pipeline. Both Reveille and Ulysses are fully approved to be connected to the grid and are not part of ERCOT's batch process. Demand for usable 2027 sites is unusually strong, but we're focused on securing the right lease for Cipher Mining, not just one that can be signed in first. As I mentioned earlier, we are working with potential tenants to advance design elements, which will reduce the risk of cost and schedule impacts caused by changing orders once construction begins.
We believe that in a market facing inflation and supply chain challenges, this prudent approach is correct, and spending more time designing and aligning tenants before the lease is executed will create more value for shareholders in the long run. Looking ahead to 2028, our 1 gigawatt site, Colcus, received conditional baseload certification. We have fully cooperated with the data center audit and expect to receive the final classification in December. Given its size and grid-connected status, we see Colcus as the cornerstone of our future portfolio, and the market's interest in this site is extremely strong. We are negotiating with a number of top partners and, as always, are pre-investing in engineering and design throughout the pipeline.
We've expanded the site to 622 acres, making Colcus fully ready to rent out once it has been finally sorted. Several other sites in our portfolio have received valuable PCLR certifications in the research load (research load) category, and we expect related updates next year, but I'd like to take a moment to talk about on-site power generation, which we think is Cipher Mining's next major value driver in the near future. In the third quarter, we began construction of horizontal gas pipelines capable of supporting up to 2.5 gigawatts of power generation, and the first batch of electricity could be delivered as early as 2027. This timeline has made these capacities highly sought after in both our contracted and pipeline sites, and we are currently in early but active commercial negotiations on a data center lease of approximately 1 GW to be powered by on-site generation.
The slide on page 16 shows the current overview of our portfolio, including what was added this quarter. I'm not going to go through each element one by one, but I would like to highlight a few points. This quarter we acquired Horizon, a 70 MW site in Texas. Horizon has signed a grid-connection agreement with the utility company and is expected to be electrified in 2029. Our site acquisition team continues to expand their pipeline, and Horizon is a great example of how they unlocked an opportunity: a site we purchased at an attractive price, and we believe our expertise can unlock significant value as an HPC campus.
I'd also like to point out the PCLR certification for the Mceska, Apollo, and Stingray extensions. Of the approximately 150 conditional study load sites, only 11 initial sites received this certification, and these three ranked among them. PCLR status allows these sites to receive full electricity prior to local transmission upgrades, meaning we expect to receive all the megawatts we requested within the submitted schedule, some of which are limited by curtailment (curtailment). We expect to get more clear information on power restrictions next year, but we've seen a variety of ways to design backup power for this section, which will allow us to commercialize these sites even before local transmission upgrades lock in all of the electricity.
Coupled with our investment in horizontal gas pipelines of up to 2.5 gigawatts of potential on-site generation capacity, our total investment portfolio today amounts to 6.8 GW. We believe hyperscale cloud vendors and AI labs will have strong demand for our pipeline sites, and we will be careful about how we bring each site to market, prioritizing tenants with good credit, attractive long-term contracts, and returns that create lasting value for shareholders. In this regard, I handed over the phone to Greg and asked him to explain our financing activities, capital structure, and third quarter financial results.
Greg Mumford (CFO): Thanks, Tyler. As Tyler stated, demand for our capacity remains strong, and we believe Cipher Mining is well-positioned for this market. This is evident in our project-level financing performance. Despite tightening access to capital across the industry, our project bonds significantly outperformed their peers, thanks directly to our priority selection of tenants with good credit and financeable lease terms from the outset. At the core of our strategy is to efficiently finance each project with a flexible structure that allows us to invest capital recovery into new developments over time.
We also recognize that how to finance the next phase of growth is critical to our shareholders, and we are reviewing every financing decision from this point of view. Today, I'll introduce the current state of our capital structure, details of our recent convertible bond issuance, and our thoughts on how to finance future growth. Then, I'll review financial results for the third quarter. First, take a look at our current capital structure and liquidity situation. As discussed in the previous conference call, each of our three contracted parks is funded through special priority guarantee bonds at the project level, and the parent company provides equity at the same time. This structure allows project debts to be segregated (ring-fenced), paid by long-term contract rents from investment-grade counterparties, and amortized over the underlying lease period. Our pricing has steadily improved over time, from 7.8% for Barber Lake to 6.8% for Black Pearl to 6% for Stingray.
I would like to briefly highlight how well these project bonds are performing in such a more challenging credit market. The broader AI infrastructure credit market has been heavily repriced over the past few months, while Cipher Mining's project bonds have been significantly more stable. We think this relative performance is important because it validates the decisions we made long before the market became more challenging. Right from the start, we intended to prioritize the most creditworthy tenants and tenancy terms aimed at stabilizing the market.
These choices can easily be overlooked in an environment like today. But they have become clear differentiators. We believe that the relative performance of our bonds reflects the strength of our contractual cash flow, the quality of our counterparties, and our growing execution record. This credit situation directly affects the availability and cost of capital for our future projects. After the end of the quarter, when we finalized the Barber Lake deal and quantified the costs with a high degree of certainty, we issued $850 million, 0% convertible senior notes due 2032, at a conversion price of $15.31, and used part of the proceeds to invest in the Barber Lake lease amendments. We chose the convertible bond market because it allowed us to lock in the company's capital for a period of 6 years with 0% coupon interest and a 35% conversion premium.
To further protect shareholders and meet our growth expectations, we also invested approximately $90 million in capped calls (capped calls) to economically dilute the hedging to $28.35 per share, or 150% of the reference share price. At the same time, we exchanged approximately US$166 million in 1.75% convertible bonds due in 2030, repaid more than 90% of these notes, and basically eliminated all cash interest expenses associated with this offering.
Overall, total debt is approximately $6.7 billion. These include $4.5 billion in non-recoverable project-level guaranteed notes and $2.2 billion in parent company-level convertible bonds, almost all of which are 0% interest. After deducting cash and cash equivalents, the net debt is approximately $3 billion. After considering the October launch and Barber Lake's funding, we have approximately $724 million in unrestricted cash and cash equivalents. Importantly, our contracted projects are funded until completion.
At Barber Lake, part of the convertible bond proceeds is used for incremental investments related to lease revisions, while cost protections built into Black Pearl and Stingray leases limit Cipher Mining's exposure above the agreed capital expenditure threshold. We have sufficient working capital to meet the challenging market environment while continuing to make forward-looking investments in our development pipeline. In short, we have a capital structure that matches our contract cash flow well, a fully funded project, and a flexible parent company balance sheet that supports our pipeline.
Let's now review our financial results for the third quarter of 2026. Third-quarter revenue was $27 million, up from $25 million in the second quarter. As Tyler mentioned, this quarter included our first HPC rental revenue and introduced segmented disclosures. We reported GAAP net profit of $32 million for the quarter, compared to GAAP net loss of $268 million for the previous quarter. This change was mainly driven by $127 million in non-cash gains on warrants liabilities, compared to $151 million in non-cash losses in the previous quarter.
Adjusted EBITDA also improved to a loss of $10 million, with a loss of $30 million in the second quarter. Compensation and benefits fell to $39 million, mainly reflecting a decline in equity incentives. General and administrative expenses increased during the quarter, driven by transaction-related consulting fees and a $3 million charitable donation to Colorado City. In the third quarter, we sold 632 bitcoins, received $45.5 million in cash, and realized a profit of $500,000, compared to a realized loss of $24 million in the second quarter.
This quarter, sales used lower-cost production from 2026 and are in the midst of a recovering market. Also, below the business line, interest income was $33 million, less than $36 million because limited cash was invested in construction. Interest charges fell from $67 million to $59 million. Capitalized interest rose from $16 million to $37 million as construction expenses for Barber Lake, Black Pearl, and Stingray accelerated, which was enough to offset the first full quarterly interest on Stingray notes.
Now look at the balance sheet. As of September 30, 2026, we ended the third quarter with total assets of $7.7 billion, an increase of $3.4 billion compared to December 31, which is roughly the same as June 30. In the third quarter, we invested significant capital in three projects. Restricted cash declined by $1.5 billion this quarter to $2.2 billion. Over the same period, property and equipment increased by $1.8 billion to $3.9 billion, driven by continued construction on our site.
We are also continuing to invest in our pipelines. Intangible assets increased by $48 million, and utility companies' safety guarantees increased by $62 million, driven by site acquisition activity. In terms of liabilities, accounts payable, accrued expenses and other current liabilities continue to reflect normal construction bills and payment schedules. Deferred revenue increased to $46 million, reflecting advance tenant payments and change orders billed in advance prior to confirmation. Before I open the question, I'd like to leave you with a few key points. Barber Lake's revisions more than doubled the site's contract revenue and improved project returns on every key metric. We raised long-term capital with 0% coupon and capped options to protect shareholders from dilution. Now our contracted projects are funded until completion, and additional liquidity is retained at the parent company level. Finally, our design and procurement measures enable us to bring greater cost and time certainty to every future lease.
Today's market environment is harsher, and we think this is beneficial to Cipher Mining's advantage. Our vast pipeline of scarce electricity opportunities, deepening relationships with leading hyperscale cloud vendors and AI labs, and capital flexibility allows us to selectively decide which opportunities to advance. Our goal is not to grow for growth, but rather to turn these opportunities into long-term contract cash flow to create lasting value per share.
Thank you all for your continued support. Tyler and I would love to answer any questions you may have.
Q & A session
OPERATOR (Moderator): Thank you. As a reminder, if you need to ask a question, press the asterisk on the phone plus 11 and wait for your name to be broadcast. To withdraw the question, press the asterisk again and add 11. For the sake of fairness, we ask that each person only ask one question and one question. Please wait while we put together a list of questions and answers. The first question came from UBS's Ryan Gravett. Your line is already open, please talk.
Ryan Gravett (UBS UBS): OK, thank you. Given the Barber Lake update and the industry's generally seeing inflation in construction costs, I'd like to ask how should we view future capital expenditure requirements and single megawatt spending? Do you expect future projects to increase rents in line with inflation and construction costs while maintaining development yields? Thank you.
Tyler Page (CEO): Thank you for asking. Let me try to answer first, then Greg, you can add some more content because you're probably more aware of how the cost of financing changes in the market from day to day. Overall, I'd like to say that we're seeing an environment of pressure on the supply chain and an inflation/deflation environment where labor coexists, and we think this trend will continue in 2027. It's hard to answer your question specifically because so many things depend on design requirements. I'd like to add that tenants are definitely aware that costs are rising, and they know that if they partner with a company like Cipher Mining, we'll finance construction on their behalf.
So I'd like to say what we've seen so far is that tenants are willing to raise rents as financing costs rise. I think what we're also seeing is an evolution in the structure. I mean, tenants are increasingly willing to accept arrangements such as capital expense reimbursement agreements in advance, so they can lock in our long-term equipment by providing funds in advance to preserve the construction period and then finance it later. So I also think tenants are looking for ways to cut costs from design, and everything is still changing as thinking around these data centers continues to evolve.
So I'd say what we're seeing now is that they're willing to respond to that. Greg, would you like to add a little more about the state of the financing market?
Greg Mumford (CFO): Okay, sure. Thanks for the question, Ryan. We touched on this slightly in the formal speaking part of the call, but I think it's worth emphasizing again. There are approximately 20 to 25 high-yield bonds issued on the market, providing financing for various types of tenants and various underwriting arrangements, across the entire spectrum. Of the four bonds with the tightest spreads in this group, only three are our bonds.
I think this is important because it explains the work we've done in history — targeting the best quality opponents — but it's also about ensuring our continuous and efficient access to capital markets. So we're very focused on ensuring access to these capital markets. We believe that with the current trading level of our bonds, as long as we stick to our duty and focus on really high-quality tenants, we can continue to go out and finance projects.
Ryan Gravett (UBS UBS): Understood, thanks everyone.
OPERATOR (Moderator): Thank you. Please wait until the next person asks a question. The next question comes from Wells Fargo's Eric Lube Choi. Your line is already open, please talk.
Eric Lube Choi (Wells Fargo, Wells Fargo, Wells Fargo, Research): OK, thanks for answering the questions. I want to talk about the Odessa project. I know the Odessa site, I know that New Era Energy signed a PPA with Luminant for a standalone project, and I'd like to know how this (if any) would change the site's monetization opportunities.
Tyler Page (CEO): Of course. So I think the opportunities for us to monetize it for HPC at this venue are still very strong. I'm saying this because we're familiar with the PPA they signed. I mean, that kind of structure we, as developers, wouldn't sign. What we're looking for is a behind-the-meter arrangement to allow those trillion-dollar tenants to take on the heavy obligations of any PPA. So it's up to them. However, any tenant interested in the site needs to be connected to the site as soon as possible, and is generally eager to obtain hard-to-obtain long-cycle equipment, which we just own.
So we insist that our ongoing dialogue is still moving forward. I think we have an extremely valuable piece of the puzzle in our hands to unlock this puzzle. I don't think we'll inject any equity into this venue. If things continue the way I expected, all we need to do is earn rewards by unlocking the interest of this venue's tenants. But these are ongoing conversations, and we are still very confident.
Eric Lube Choi (Wells Fargo, Wells Fargo, Research Division): OK, thanks. Just one more question, Tyler. Given your exposure in Texas and some of the regulatory complexities associated with ERCOT, how do you view the relationship between geographic concentration and finding a venue outside of Texas in the long run? I know you have a site or a pipeline site in Ohio, but what are you considering in the long run? How focused would you like to be?
Tyler Page (CEO): OK, of course, that's a great question. I think we've spent a lot of time thinking about this. I think the truth is that in an evolving regulatory environment, Texas is essentially the same as almost every other data center location in the US, and is in a constantly evolving environment. Whether it's regulatory approvals, taxation, community relationships, etc. The reason it's getting the most attention is obviously because there's so much construction planned there, and we obviously have plenty of exposure there.
I think in the long run, our view is that the construction of data centers is related to the future of the US economy. Right now we're in the election cycle, and there's a lot of noise. In the long run, it doesn't matter who is in charge. It's a critical part of the country's success and that of Texas. So I think that although Texas is currently receiving close attention from investors — everyone is asking what that means — we don't think Texas is substantially different from other regions in terms of these risk factors.
We'll have to wait and see. We have been operating at ERCOT for a long time. Overall, we think this is a system that works very, very well. We're excited to be building in Ohio. We're looking at sites in other locations, but I'd like to emphasize that everywhere else, you'll see situations such as extremely long power supply timelines and heavier security deposit requirements. So, although in theory we always like diversification, if you look at a venue where you have to pay hundreds of millions of dollars in advance and lock in for five years to connect to it, it's not that appealing.
So, on the one hand, we are always looking for and continuously evaluating sites. But frankly, no matter how this process evolves, we're still confident in Texas.
Eric Lube Choi (Wells Fargo, Wells Fargo, Wells Fargo, Research): Thanks, Tyler.
OPERATOR (Moderator): Thank you. Please wait until the next person asks a question. The next question comes from BTIG's Greg Lewis. Your line is already open, please talk.
Greg Lewis (BTIG): OK, hello, thank you, good morning, and thank you for answering my questions. Tyler, I'd like you to talk more about the horizontal gas pipeline you're developing. I mean, apparently, the speed at which electricity is marketed will continue to be something that companies that can access it can take advantage of in the next few years. When we consider these pipelines, how do we also need to solve the equipment requirements for converting natural gas delivered through pipelines into electricity?
Tyler Page (CEO): Thanks, Greg. That's a great question. At a high level, I'd like to say that what drives tenants' interest in on-site generators is the “accelerated power supply” story. In other words, grid power is easier in many aspects of operation, but the value of getting data centers up and running faster is so great that tenants are willing to get creative if there are regulatory or other barriers to the potential timeline of grid-connected assets. What I think we've experienced is that tenants prefer a site that can both generate their own electricity and has a connection to the grid.
So our venue was very valuable. The way we think about horizontal pipelines is that we have sites close to large pipelines, we have sites that have been or are about to be connected to the grid for a short period of time, and that kind of expanded capacity may be very attractive to tenants who want to do large-scale construction. As for the device, this is one of the really hard things to get. There simply aren't enough turbines and reciprocating engines in the world.
The arrangement we envision in the deal is for a well-known, reputable, and reputable independent power producer (IPP) to work directly with our tenants to arrange the supply and operation of the equipment, and then the IPP will face the tenant with a power purchase agreement (PPA). So ultimately, Cipher Mining's role will be to finance and build data centers. We also need to clarify how to build and operate it differently because the source of electricity is different from the grid.
But the challenges surrounding equipment and operations all fall between tenants and IPP. This is one of the major difficulties in putting all of this together. We've already, you know, been in deep negotiations on a deal of around 1 gigawatt, so we feel like we're in a very good position there. We have multiple interested tenants at multiple sites, and we are also thinking about how much long-term viability our site can become over time.
Greg Lewis (BTIG): Okay, very helpful. One more question about Horizon, a 70-megawatt site. One problem, I think, is that at that scale, it's below 75, so I'm assuming it's outside of Batch 0. So does that also change the way you think about the long-term expansion of this venue?
Tyler Page (CEO): Yes, I think so. I mean, what I'm saying is, I think ERCOT's rules may evolve. It's not currently within that framework because you're right, it's less than the minimum threshold. But, you know, I wouldn't be surprised if these elements of regulatory scrutiny evolved over time. So that makes it so appealing today. We want it to maintain this appeal from a timeline perspective, that is, to maintain that timeline perspective. I think a site like that, when you take my perspective a lot further, I think what makes Cipher Mining's way of operating and creating value very different from traditional data center developers is that so much equity value that we end up creating comes from an increase in the valuation of what we do at a site. So we found a site in a non-traditional area, took it on extremely cheap terms, and then when we put the site on the market, there was a huge jump in value.
What is common across the market is that sites, you know, available sites are so scarce now that I think the valuation of grid-connected land plots has risen sharply everywhere. Furthermore, the regulatory requirements surrounding security deposits make them very, very expensive. So we're always looking for opportunities to create that kind of extra value. I think it's fair to say that a site of the size of 70 MW has the potential to generate huge value, and we can still buy them cheaply.
The ROI analysis of these sites was very appealing to us.
Greg Lewis (BTIG): Very helpful. Thank you very much.
OPERATOR (Moderator): Thank you. Please wait until the next person asks a question. The next question comes from Cantor Fitzgerald's Brett Knobloch. Your line is already open, please talk.
Brett Knobloch (Cantor Fitzgerald): Good morning, everybody. Thanks for answering the questions. I just wanted to dig a little deeper into the extra cost of Barber Lake. I understand the change order part, but maybe as far as supply chain inflation is concerned, when are large parts of these costs locked in? I know you've been saying that the equipment is covered. I think my opinion is that labor costs are locked in when you sign a general contractor (GC). So maybe I'm just misunderstanding when exactly all the costs on your side are locked in.
If you could elaborate on this and what is driving this growth.
Tyler Page (CEO): Let me add one more thing, then I'll leave the details to Greg. But I want to say at a very high level that the nature of the cost structure of all of our initial sites stems from the fact that our tenants are really anxious and anxious to lock down these sites. And we're also in a hurry to sign those leases because this is a complete repositioning of Cipher Mining. Right. I'm saying this because moving fast means our designs aren't perfect enough.
The result — tenants are willing to cooperate — was that there were more change orders over the past year than we expected. My priority right now is that I think we work with the best long-term tenants in the world. So we work hard to meet the needs of our tenants. But if they make a very significant change order, it will have a knock-on impact on costs and timelines. And when you reprice these changes in an evolving environment, that's the source of that kind of thematic exposure.
Therefore, you may need to hire an additional construction team to handle the adjusted construction schedule to accommodate changes in the cooling infrastructure or other aspects of the tenant's design. This then has a cascading effect throughout the construction process. I emphasized this before handing it over to Greg because as we look ahead to the next batch of leases — we've talked about it on the phone before — now we're very confident the demand exists. It's a very different environment as we go through the vertical climb of corporate transformation. We have the ability to be more careful about next-gen leases. On these next-gen leases, we won't... We've been delaying execution of these leases until we finalize 30% to 60% design perfection.
This would greatly narrow the scope of these cost fluctuations. So structurally, that's where it comes from. Also, the venue is evolving. I mean, before I handed it over to Greg, I mentioned that I just learned this morning that we just issued the first 51 MW RFS notifications for critical IT to Barber Lake. So now we're working with them in a way that makes our tenants happy. Collect rent from multiple sites and multiple tenants. But this first environment has always been dynamic because we have to handle a large number of change orders in one place. We were transforming a Bitcoin mining farm at the time.
Therefore, in these first batch of construction projects, the operating environment has always been very dynamic. Greg, I don't know if you want to add anything else, or not.
Greg Mumford (CFO): OK, thanks for asking, Brett. I think it's important to differentiate between two specific groups. One was a change order, and we were eligible for reimbursement; the other was cost inflation across multiple different verticals. So one of the things we're talking about is scope evolution. This is a relatively large “hodgepodge” category. But overall, it refers to some of the things that are happening at these sites, and these large-scale infrastructure projects may result in lower productivity than you might expect.
Some of this may be due to the fact that you know that a tenant might arrive early with their racks and chips, you know, you want to work with them as much as possible, so you find a place to store them, then you settle them, and you're still working around them. So you're losing some productivity. As you know, overall, inflation in owner-supplied (OFE) equipment is not one of the main drivers. Although we did see some inflation and some longer delivery cycles, causing some construction teams to stand there waiting for the equipment to be delivered. But it is true that the labor market has always been tight. I mean, I don't think we're the first — and won't be the last — to hear this. This is a question we know very well and are thinking about in our next construction project.
But when we signed these deals a year and a half ago, the market was very different from what it is now. So, one thing we've talked about a lot about is focusing on engineering, design, and tenant alignment earlier to ensure we can transfer risk with general contractors or subcontractors about these labor costs earlier. Then, back to that point. So without this deal we made, if we were reimbursed for the change order and covered the additional overall cost evolution, say $350 million to $400 million, our cost would have been around $12.3 million per key megawatt. That's just over our range of 9 million to 11 million. And that doesn't take into account that out of the 350 million to 400 million, we could have spent more time dismantling and sharing.
This explains the existence of a cost-sharing mechanism — 50% of which is reimbursed to us, plus return on capital.
Brett Knobloch (Cantor Fitzgerald): Awesome, very, very helpful. Maybe just ask one more question. The December audit is expected to be completed? Are potential tenants or customers you are reaching out to wait for the audit process to be completed? Or if you guys have conditional baseload and are at the top of these sites and everything goes well, can they be rented out even before the audit? Or what should we think of the different categories you guys got?
Tyler Page (CEO): I mean, I'd say Colcus is really the jewel in the crown, because a 1 gigawatt front of the meter (front of the meter) venue is extremely rare. We have several of the world's most respected tenants who are very interested in this site, and I expect it to be a real catalyst for our valuation increase. I think the site is so appealing that people are making due diligence and reviewing the lease framework. They're getting 30% to 60% design perfection in an attempt to make it clear what they really want to do.
I don't expect that we will sign the final lease before getting connected to the grid, or if we do, then there will definitely be a prerequisite that we receive the grid connection. So I think you're getting the same results either way. If someone wants to move forward, and we think we've found the right dance partner and the right structure, we can theoretically move forward with one prerequisite. But my instinct is that you need to get connected to the grid no matter what.
Brett Knobloch (Cantor Fitzgerald): Awesome, I really appreciate it, Tyler. Thank you all.
OPERATOR (Moderator): Thank you. Please wait until the next person asks a question. The next question comes from Michael Donovan of Compass Point. Your line is already open, please talk.
Michael Donovan (Compass Point): Hi, thanks for answering the question. Congratulations on the progress. You mentioned PCLR sites and the use of on-site power generation to support workloads during grid power restrictions. Are you also considering batteries and do potential tenants have preferences?
Tyler Page (CEO): Yes. So we're considering batteries as one of the ideas. I don't think the PCLR certification is fully understood by the market. I mean if I had to sum up the current market, it's all too short-sighted, so that's a potential solution. But the point is, when we look at the focus on supply time availability, I don't think this demand characteristic will change much over the next few months. But hopefully we can make progress through the ERCOT process. Soon we'll be discussing sites for research load certification, probably until next spring, then I think PCLR certification will be a real big deal. So, you know, we'll wait and see. Our feeling is that when we observe the evolution of tenants' construction needs, availability requirements, and cost considerations, the idea that “there is a definite power supply timeline, but it may be necessary to do engineering detours in terms of power restrictions or switching to on-site power generation” is very interesting.
I don't think the market is fully aware of how big these certifications are.
Michael Donovan (Compass Point): Thanks, Tyler. Then at a high level, how do you expect the cost-benefit ratio of an on-site power generation project to compare to a grid-connected project?
Tyler Page (CEO): Greg, would you like to answer this?
Greg Mumford (Chief Financial Officer): OK, I'll try to answer. Judging from our discussions and what we've seen so far, I don't expect it to have substantially different pricing. I think if the tenant has designated a site or opportunity as a high priority, we see cost-benefit discussions in these lease negotiations that are broadly consistent with the grid-connected site in front of our table. I think more lease discussions will focus on SLAs and consider different power generation sources to try to match the conditions that both parties are willing to accept.
Michael Donovan (Compass Point): Understood. Thank you all.
OPERATOR (Moderator): I'd like to return the call to Tyler Page for a closing address.
Tyler Page (CEO): Okay, thanks to everyone who dialed in. I just want to remind everyone that we are really in the vertical-ascending phase of transformation. I know there's a lot of short-term pressure in the market given everything that's happening. But I want to encourage all of our investors to take this perspective: we're building highways for AI listings, and we're charging long-term toll fees. So, as we deal with everyday headlines, we should remember that we are building large-scale sites, and we will continue to collect toll fees for decades to come, ultimately leaving behind a very valuable highway.
So, as we transformed our company, all of the most important fundamentals were there. We mentioned that headline—we're now also starting to collect rent at Barber Lake. A meaningful portion of the site has been delivered and we're very excited about the progress over the next few months. So thank you all for your continued support. cheers.
OPERATOR (moderator): This concludes this conference call. Thank you all for participating, now it's time to hang up. I wish everyone a great day.