Bitdeer Technologies Q3 2026 Earnings Call Transcript
Earnings Transcripts 2 hours ago
MarketsTrading IdeasNews
Bitdeer Technologies (NASDAQ:BTDR) reported third-quarter financial results on Tuesday. The transcript from the company’s third-quarter earnings call has been provided below.Benzinga APIs provide real-time access to
Bitdeer Technologies (NASDAQ:BTDR) reported third-quarter financial results on Tuesday. The transcript from the company’s third-quarter earnings call has been provided below.
Benzinga APIs provide real-time access to earnings call transcripts and financial data. Visit https://www.benzinga.com/apis/ to learn more.
View the webcast at https://edge.media-server.com/mmc/p/pm4ikhxc
Summary
Bitdeer Technologies reported a record third quarter 2026 with an 89% year-over-year revenue increase to $320.7 million and a positive gross profit.
The company’s AI platform, including AI cloud services and HPC infrastructure, has a strong revenue backlog of approximately $7.6 billion, with significant projects underway in Malaysia and Norway.
Bitcoin mining continues to be a major revenue driver, with a record production of 3,855 Bitcoin and ongoing improvements in mining efficiency.
Strategic initiatives include expanding AI cloud and HPC infrastructure, with major projects like Tito in Norway and new developments in Malaysia, aiming for a balanced growth between AI and crypto platforms.
Management discussed a cautious approach to financing, focusing on project-level funding and customer prepayments, while limiting equity financing to early-stage investments.
Full Transcript
OPERATOR
Good day and thank you for standing by. Welcome to the Bitdeer Technologies third quarter 2026 earnings call. At this time, all participants are in a listen-only mode. Please be advised that today’s conference is being recorded. After the speaker’s presentation, there will be a question-and-answer session. To ask a question, please press star 11 on your telephone and wait for your name to be announced. To withdraw your question, please press star 11 again.
I would now like to hand the conference over to your speaker today, Tesh Daya, Head of Investor Relations.
Tesh Daya (Head of Investor Relations)
Thank you, Operator, and good morning, everyone. Welcome to Bitdeer Technologies’ third quarter 2026 earnings conference call. Joining me today are Jihan Wu, Founder, Chairman and Chief Executive Officer; Haris Fozan Basit, Chief Strategy Officer; and Michael G. Potter, Chief Financial Officer. Today’s call will begin with Haris, who will provide a high-level review of our strategy. Michael will then review our operating highlights and third quarter financial results before we open up the lines for questions.
To accompany today’s call, we have provided a supplemental investor presentation available on Bitdeer Technologies’ investor relations website under Webcasts and Presentations. Going forward, our infrastructure metrics previously provided in our monthly production updates will be included in our quarterly earnings release. Before management begins their formal remarks, I would like to remind everyone that during today’s call we may make certain forward-looking statements. These statements are based on management’s current expectations and are subject to risks and uncertainties which may cause actual results to differ materially. For a more complete discussion on forward-looking statements and the risks and uncertainties related to Bitdeer Technologies’ business and industry, please refer to the Company’s Annual Report on Form 20-F and other filings with the U.S. Securities and Exchange Commission.
I also want to note that we will be discussing certain non-GAAP financial measures, including adjusted EBITDA and adjusted profit or loss, and operating metrics. A reconciliation of these measures to the most direct comparable GAAP measures is included in our earnings release issued earlier today. These measures should not be considered in isolation from or as a substitute for the most directly comparable GAAP measures. As a reminder, changes in the fair value of our digital assets and certain borrowings are reflected in GAAP net income or loss and may introduce non-cash volatility into our reported results.
A replay of this call will be available on our investor relations website shortly after the call concludes. With that, I will now turn the call over to Haris.
Haris Fozan Basit (Chief Strategy Officer)
Thank you, Tesh, and good morning, everyone. The third quarter was an important period in Bitdeer Technologies’ evolution. We continue to build a diversified technology and digital infrastructure platform with a growing presence in AI, cloud services, and high-performance computing infrastructure. The market continues to evolve quickly. Bitcoin markets remain cyclical. Demand for AI compute continues to outpace available supply while competition for power, infrastructure, and customers is intensifying.
Against this backdrop, our advantage is our ability to develop, control, and monetize power infrastructure. Power is the foundation of our platform, and we aim to deploy it where it generates the greatest long-term economic value. We do that across two platforms, AI and crypto. Our AI platform spans AI cloud services and HPC infrastructure. In AI cloud, we deploy GPU capacity alongside customer commitments with disciplined utilization and capital investment.
In HPC infrastructure, we develop and lease large-scale data center capacity to credit-supported customers under long-term agreements. The economics are driven by contracted revenue, not Bitcoin prices or GPU markets. Our title project in Norway is a significant milestone as we advance construction toward initial delivery. Our crypto platform spans Bitcoin self-mining and our SEALMINER ASIC design. Self-mining generates revenue from our existing power while providing flexibility as we develop sites for longer-term HPC opportunities. Our proprietary SEALMINER technology gives us control over the efficiency and cost of our mining hardware. We deploy it across our own facilities and with our co-mining partners.
Since both platforms run on a common foundation of power infrastructure and technical expertise, we gain both schedule and capital flexibility. We can mine to monetize available power today while we develop infrastructure for long-term HPC leases, and we can invest in GPU capacity when and where we see attractive contracted demand. We are beginning to see that strategy translate into long-term revenue visibility. Our AI platform now represents approximately $7.6 billion of expected contracted revenue. That includes approximately $4.7 billion from the base term of our Tito lease and approximately $2.9 billion of AI cloud backlog. These carry different revenue recognition profiles and capital requirements, but together they show the scale of what we are building.
Our priorities are deliver title, expand AI cloud and HPC infrastructure in line with demand, pursue disciplined and financeable growth, and keep improving the efficiency of our mining operations. We do not want to depend on any single market cycle. Our objective is to build a platform that generates durable cash flows from the power and technology assets we control. With that, I will turn the call over to Michael.
Michael G Potter (Chief Financial Officer)
Thank you, Harris, and good morning, everyone. The third quarter was a record quarter for Bitdeer Technologies. Revenue grew 89% year over year to $320.7 million. Gross profit turned positive and adjusted EBITDA more than doubled quarter over quarter. Bitcoin mining continues to generate most of our revenue in the near term, while our AI platform scales. Before the details, a quick scorecard against what we told you last quarter: Tito Phase 1 remains on schedule for delivery at year end in Malaysia. A102 is sold out and A201 is more than 70% contracted. Our first GPU financing for A102 is in final stages. All 221 megawatts at Massillon are now online and equity was a much smaller part of our funding. We raised $58 million under the ATM, a continuation of and compared with $457 million in the second quarter.
Our focus is on execution. Let me start with our AI platform, beginning with AI Cloud Services. AI Cloud revenue was $20.4 million, up 46% from the second quarter. That is more than 11 times the $1.8 million we reported in the third quarter of 2025. Our deployed GPU count was roughly flat quarter over quarter. The sequential growth came mainly from pricing and mix as utilization of our higher-value GB-class systems increased. Our AI Cloud ARR was approximately $86 million at the end of September.
We ended the quarter with 4,328 GPUs deployed at approximately 95% utilization. AI Cloud generated $3.1 million of gross profit in the quarter, a gross margin of approximately 15%. That compares with gross losses of $2.3 million in the second quarter and $0.8 million in the third quarter of 2025. Contracted demand continues to build. During the third quarter we signed significant multi-year GPU cloud agreements with leading enterprise technology companies. Deliveries of Nvidia GB300 systems under several of these agreements are expected to begin in November. Malaysia is where most of this demand is landing, and the third quarter was a breakout period for our Malaysia footprint. At our A102 facility, we sold out all 9.5 megawatts ahead of energization.
These five-year commitments represent more than $800 million in expected revenue. At A201, we secured offtake commitments for more than 70% of the 21.7 megawatts. That represents more than $1.7 billion in expected revenue over five years. We also ordered more than 100 racks of GB300 systems for the site. We expect revenue to begin at A102 in the first quarter of 2027. A201 will follow in the second quarter of 2027. Together, these commitments lifted our expected AI Cloud revenue backlog to approximately $2.9 billion.
These contracts carry attractive unit economics on average over the contract terms. A102 implies approximately $17 million of revenue per megawatt per year and A201 more than $21 million per contracted megawatt. Revenue in any given year may vary with contract pricing schedules and deployment timings. Once both sites are fully deployed, they represent more than $500 million of annualized revenue. We also added meaningful AI Cloud capacity since the end of the second quarter. In September we signed a 10-year agreement for A202A, a 65.1 megawatt facility at our Johor campus. Energization is expected in the third quarter of 2027. In October we added A901, a 67 megawatt campus expected to be ready for service in the second quarter of 2027. We also signed an eight-year agreement for A103A, a 60 megawatt Vera Rubin–ready facility at our Cyber Gaia campus.
Energization is expected in the first quarter of 2028. Including these sites, our secured AI cloud data center capacity is now approximately 332 megawatts. That is about 95% of our 350 megawatts target for the first quarter of 2028. I want to be precise about what these numbers mean. Secured capacity means sites we own or hold under an executed data center services agreement. It’s not the same as customer contracted capacity. The combined value of our signed AI Cloud contracts and AI Cloud opportunities under negotiation now exceeds $10 billion.
This includes our approximately $2.9 billion backlog. The amount under negotiation is not a binding commitment. Turning now to our HPC infrastructure, the second part of our AI platform, our focus remains on Tito. Tito has leased the Volta for 16 years, subject to the tenant’s no-fee termination right at 10 years. The lease covers 121 megawatts IT or 133 gross megawatts. The base term value is approximately $4.7 billion. If the eight-year renewal option is exercised, it would rise to approximately $8 billion over the base term. Average annual revenue, including rent and service fees, is approximately $293 million with an expected NOI margin of about 90%.
This implies roughly $264 million of asset management. Average annual NOI. Tito is a conversion of one of our existing Bitcoin mining sites to HPC infrastructure. Grid power and much of the electrical infrastructure are already in place. This significantly reduces the interconnection and energization risk that many new-build data center projects face. The site workforce has grown to more than 300, working two shifts. Long-lead equipment, including the battery energy storage systems, was ordered in advance of requirements. Key electrical equipment has also begun arriving on site. The 22 kV transformer houses and UPS systems are now installed. The battery storage buildings are nearly complete and dry coolers are being mounted. The remaining work is focused on three areas: completing the high-voltage connection to the new data halls, commissioning, and integrated systems testing.
Phase 1 delivery remains on schedule for year end. Customer installations are planned to begin in January. Phase 2 data hall deliveries are expected to begin at the end of the first quarter of 2027. Rent payments for the first eight years are expected to be backed by dedicated credit support. For years one to six, that support is anticipated to be approximately $1.3 billion of irrevocable letters of credit. They are expected to be issued by two highly rated global financial institutions. For years seven and eight, we anticipate an approximate $475 million dedicated cash reserve.
Our portfolio has approximately 3.1 gigawatts of global power capacity with about 1.75 gigawatts online. Most of our near-term AI capacity is in Norway and Malaysia, where the power is secured, and many of our largest U.S. sites are already energized. Turning to our crypto platform, beginning with self-mining, we mined 3,855 Bitcoin in the quarter, up 43% from the second quarter. September was a record month at 1,355 Bitcoin. Self-mining hash rate continued to expand during the quarter. It reached approximately 82.3 exahash per second at quarter end. Co-mining revenue was $59.9 million, up 140% quarter over quarter. Co-mining hash rate was 28.7. Our power cost per Bitcoin mined was approximately $36,000 in the third quarter, in line with the second quarter and down from $49,000 a year ago.
That reflects continued fleet upgrades which strengthen our underlying cost position. Our self-mining fleet efficiency reached 15.4 joules per terahash by the end of the third quarter, from 20.1 a year ago. As we complete our SEALMINER upgrades, we expect fleet efficiency to continue to improve. Maintaining a leading fleet efficiency position is particularly important during periods of lower Bitcoin prices, helping preserve margins and operating flexibility.
On ASIC design, our A4 Ultra Hydro operates at 9.45 joules per terahash at the mining rig level. We continue to deploy SEALMINER into our own sites and into co-mining partner sites as opportunities arise. Across both platforms, we aim to develop our sites as a good neighbor in partnership with local communities and governments. Our approach is to secure our own power without drawing on existing ratepayers. We pay for the grid infrastructure our facilities require, and we invest locally through jobs, training, and long-term partnerships.
In September, we signed the White House Ratepayer Protection Pledge. It reaffirms our commitment to responsible digital infrastructure development. Now to our financial results. Third quarter revenue was $320.7 million. That is an increase of approximately 89% year over year and 40% sequentially. Most of the sequential growth came from self-mining and co-mining. Self-mining contributed approximately $49 million of the increase and co-mining approximately $35 million. AI Cloud added approximately $6 million. Gross profit was $10.8 million for a gross margin of 3%. That is a $19.4 million improvement from the second quarter.
Higher Bitcoin production, improved fleet efficiency, increased co-mining activity, and AI Cloud growth drove the gain. These benefits more than offset higher average power costs. Our average power cost rose to 4.7 cents per kilowatt hour from 4.4 cents in the second quarter. Turning to operating expenses, G&A declined to $32.8 million from $34.3 million in the second quarter. R&D was $28.2 million, compared with $36.1 million in the second quarter, which included a one-time development expense. Selling expense was $4.2 million, up from $2.2 million, driven by higher advertising and marketing expenses. Interest expense was $34.3 million. Net loss was $125.5 million, or $0.46 per share. The result reflected non-cash items.
These included $125.8 million of depreciation and amortization. They also included a $54.6 million fair value loss in Bitcoin-denominated borrowings as Bitcoin prices increased. These were partially offset by a fair value gain on our Norway power purchase agreement. Because we repay our Bitcoin-denominated borrowings with mined Bitcoin, the fair value loss on these borrowings is a non-cash timing-related impact. Adjusted EBITDA was $69.6 million, up approximately 173% year over year and 124% sequentially. That represented an adjusted EBITDA margin of 21.7%. As a reminder, we transitioned to U.S. GAAP on January 1, 2026, and prior periods have been recast on that basis. Third quarter 2025 adjusted EBITDA was $25.5 million. Sequentially, adjusted EBITDA rose about $38.5 million. Revenue grew about $92 million, led by Bitcoin production and co-mining. That was partially offset by about $40 million of additional electricity costs as our expanded fleet consumed more power.
Other cash costs rose about $14 million. This reflects a larger self-mining fleet, co-mining growth, and AI Cloud. Moving to the balance sheet, we ended the quarter with $163.7 million of cash and cash equivalents. We also held $50 million of restricted cash. That compares with $456.8 million of cash and $39.5 million of restricted cash at the end of the second quarter. Digital assets were approximately $43 million, and digital asset receivables, mainly collateral for our Bitcoin-denominated borrowings, was about $344 million. We invested approximately $288 million in data center construction and land, including significant construction spending at Tito and the land purchase at Rockdale.
We also continue to invest in GPU capacity to support AI Cloud demand. In total, capital expenditures in the quarter were $478 million, including approximately $110 million for SEALMINER production. We made net repayments of approximately $92 million on our related party loan. Total debt was $1.95 billion, up approximately $107 million from the second quarter. The increase reflects additional Bitcoin-denominated borrowings from a related party and a higher Bitcoin price.
Our convertible notes totaled $1.2 billion in principal. Coupons range from 4% to 5.25%. None mature before December 2029. On liquidity, let me be direct: cash declined because we funded a heavier investment quarter, including Tito construction and the Rockdale land purchase, ahead of our expected financings. That was a deliberate choice to protect the construction schedule. Operating cash flow was approximately $6 million. This includes the customer prepayments but not the $221.5 million of mined Bitcoin we sold, which is reported in investing activities. Together, operations and Bitcoin sales generated roughly $227 million of cash during the quarter.
Our funding plan matches each major use with a dedicated source. First, Tito: the remaining capex of the Volta lease of approximately $350 million is expected to be funded with project-level financing secured by contracted rent. Second, AI Cloud GPUs, where customer prepayments are our core source. Deferred revenue, mostly AI Cloud prepayment, rose to approximately $342 million at quarter end. These prepayments fund a large part of the related GPU deployments—50% or more.
Third, our ongoing Bitcoin production, including 1,355 Bitcoin in September, which supports our mining operations and corporate needs. If needed, we can also phase discretionary capital spending such as non-Tito construction. We are already doing this, as I will discuss in our outlook. On financing, we’re in the final stages of GPU financing for our 9.5 megawatt A102 site in Malaysia. Work is underway on a similar financing for our 21.7 megawatt A201 site. Some GPU financings may be shorter term so we pay down that debt faster. Lenders are differentiating sharply by structure and counterparty. We seek to structure our financings around contracted cash flows, customer prepayments, and, where available, credit support from reputable financial institutions.
We have started our own Tito project financing process, and we will share updates as appropriate. We prioritize non-dilutive, project-level financing wherever contracted cash flows can support it. In the near term, our goals are to close the Tito project financing, complete the A102 and A201 GPU financings, and deliver Tito Phase 1. Finally, our outlook for 2026. We now expect crypto mining infrastructure capital expenditures to be at or below $200 million, the low end of our prior range of $200 million to $280 million. We are directing capital towards our contracted AI and HPC projects while we complete our financings. Our long-term mining strategy is unchanged. This guidance excludes spending on co-miner hardware, GPUs, AI Cloud, and HPC infrastructure development. Remaining Tito capital expenditure for the Volta lease of approximately $350 million will be spread across Phase 1 and Phase 2 delivery.
That spend extends into 2027. We expect to provide 2027 capital guidance with our fourth quarter results. We also expect our mining hash rate to continue to grow over time, subject to normal seasonal curtailment. To close, let me step back. The past several months have brought increased volatility. Bitcoin prices moved sharply, credit has tightened across AI infrastructure, and investors are asking harder questions about how these projects get financed and delivered. Those concerns are understandable. Our model is built for this environment. Our AI and crypto platforms share one power foundation and complement each other. That flexibility matters. We do not have to force every site into one model.
We match each asset to the approach that fits it best. A large, well-connected site may suit a long-term lease. A smaller site with strong customer demand may work better for AI Cloud. A site being developed for AI can mine Bitcoin until it is ready. Our objective is simple: maximize the return on every power asset and protect the balance sheet while we do it. That means funding growth with contracted revenue, customer prepayment, and asset-level financing wherever we can.
We believe this is the right path to durable value for our shareholders. Now our focus is execution: delivering Tito on schedule and on budget, bringing Malaysia online, and closing our financings. Thank you, operator. Please open the call for questions.
OPERATOR
Thank you. As a reminder, to ask a question please press star 11 on your telephone and wait for your name to be announced. To withdraw your question, please press star 11 again. Our first question comes from Nick Giles with B. Riley Securities. You may proceed.
Nick Giles (Analyst at B. Riley Securities)
Thanks, operator. Good morning, everyone. Lots of updates here today. I first wanted to ask about the AI cloud business—how you’re seeing prepayments evolve, what percentage of capex you think you can achieve today, and then if you could outline your current customer mix and your desire to shift that mix more toward AI labs with higher rates or more toward hyperscale-like parties. Any color on that would be great.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
I think the prepayment will be well above 50% in the near foreseeable future, and I think the pressure for those demand to pay more upfront payment actually is higher because the capital market is closing the door for all the new labs—sorry, for all the new clouds. So basically, the social engineers—no—any new clouds, I mean our competitors, will be able to do any new capex without very good support from the clients themselves. Which means that we don’t have any kind of meaningful competitions that are willing to accept no prepayment and do the project.
I think that new AI-related startups are very underserved in the market, while frontiers, I think, are a little bit overserved, and the hyperscalers’ terms are not very good profitability, I mean. So we think that we will be quite focusing on the new labs and AI startups.
Nick Giles (Analyst at B. Riley Securities)
Thank you so much, Jihan. That’s really helpful color. For my follow-up question, I wanted to shift to Rockdale, and I was curious if you could just remind us of the status of that interconnection—whether any approvals by ERCOT would be required for the transition, or potential transition, to AI. And then you made some recent land acquisitions. Do you have any more land around the site that you need to acquire in order to fully leverage the site for HPC? Thank you.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
We’re still looking into the situation, but I think the land we acquired is already enough for us to carry out our AI transition on the Rockdale side. And I think we want to apply for more electricity from the grid, which actually we were hinted, or even kind of promised, for quite a long time, but we are still waiting. I think in the next year maybe we can get some more electricity and we can start to build out AI data centers.
Nick Giles (Analyst at B. Riley Securities)
Okay, good to hear. I’ll turn it over for now, but appreciate the update.
OPERATOR
Thank you. Our next question comes from Nick Krema with Barclays. You may proceed.
Nick Krema (Analyst at Barclays)
Good morning and thanks for taking the question. As you think about converting the existing U.S. mining sites to AI and HPC over time, how should we think about the regulatory and permitting backdrop following the midterm elections? And does having those sites already energized put you in a relatively better position? Thank you.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
We do feel a little bit uncertain after the midterm election, but my guess is that the data center constructions will need to have better arrangements with the local community on the tax and community contributions. I think eventually lots of data centers will still get permission from the local community to start. That’s my estimation of the future situation, which means a little bit higher cost for the center to operate in the community, but I think that’s fair.
And of course, those data centers that are already running, I think they do enjoy this kind of short-term uncertainty in the next one or two quarters, I guess, because everybody will be slowed down, but those already being built or finished have an edge over others.
Nick Krema (Analyst at Barclays)
Understood. Thanks for the color there. And then for my follow-up on Tidal, can you just give us an update on the project financing for Phase 1 as you approach year-end delivery and then more broadly on Phase 2? How should we think about the key milestones and execution priorities as you move towards delivery, I think in Q1, Q2 next year? Thank you.
Michael G Potter (Chief Financial Officer)
I think in terms of the financing part, we’re advancing on our plans to do the project financing. I think everybody has seen that our customer Volta—that’s been reported in the news—was out and was successful in securing their GPU financing, and that removed the last major item that was needed to activate the LC, with the remaining part just completing the construction and getting the GPUs working, which is normal for a site like that. Obviously everything is subject to market conditions, and the market is relatively tough for any sort of AI financing. But we believe a combination of very short time left for the construction—so the construction is quite advanced—the fact that the power was on and working for bitcoin mining before we converted the site to AI, so we don’t need to hope that utility will provide the power because it’s there already, and the strength of the LC backstop that we have will help us there.
In terms of milestones, I mean, we’re very close to the December 31st milestone for this year for completing Phase 1, and we believe that we’re on track. And we also believe that we’re on track for Phase 2 as well. Obviously that was tested by the market somewhat when Volta was out doing the GPU financing, and there was a fair amount of external due diligence that was done then. So if you’re going to take any sort of comfort in our progress, I guess you can take a nod to that.
Nick Krema (Analyst at Barclays)
Got it. Thank you very much.
OPERATOR
Thank you. Our next question comes from Mike Colonnese with H.C. Wainwright. You may proceed.
Mike Colonnese (Analyst at H.C. Wainwright)
Hi, good morning, guys. Thank you for taking my questions and congrats on all the progress across the platform. First one for me: I’m just curious, how should we expect the AI cloud revenues to ramp throughout 2027 following this accelerated pace of contract execution you’ve seen in the business as you work to convert this backlog? Assuming five-year contracts against $2.9 billion of TCV, by our math that would equate to about $580 million of ARR. So really the question is: when should we expect Bitdeer to achieve this full run-rate, and how do you expect those revenues to ramp again as we progress through 2027?
Michael G Potter (Chief Financial Officer)
So we’re going to provide 2027 guidance with our Q4 2026 earnings release. All of our revenue turns on when the different phases turn on. We’ve given rough guidance of when we expect energization or delivery. Usually a few months after that, the GPUs can be up and running. But we’ll provide more specific guidance in our Q4 earnings release.
Mike Colonnese (Analyst at H.C. Wainwright)
Got it. Thank you, Michael. And to clarify—so it’s capital expenditure guidance as well as AI cloud revenue guidance in the Q4?
Michael G Potter (Chief Financial Officer)
We’re going to give 2027 guidance in Q4.
Mike Colonnese (Analyst at H.C. Wainwright)
Great. And as a follow-up, if you could just provide a little bit more color on prospective tenant discussions and how those could be progressing at Rockdale following the recent acquisition of additional acreage at the site—the dynamics between you deciding whether or not to convert existing bitcoin mining data center capacity there versus building AI data centers from scratch as you look to utilize capacity at that campus.
Michael G Potter (Chief Financial Officer)
Jihan, do you want to answer the question about the customer plans and discussions in the—
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Yeah, for Rockdale. Well, Rockdale, we’ve looked into really massively winding up the mining operation and then waiting for clients. Our plan is to use the acquired land and try to get more power from the grid, and then we build new data centers on the new land. So actually our pressure to get the customers is not that big. We can build from small and expand over the years and sample various demand. We are quite flexible on this. So right now we are focusing on getting new power from the grid, which I think we can certainly expect to have good news in the near future. And at the same time we will design and do lots of the geo work to prepare for the site. I think these kinds of resources are quite good for us. So I’m not now very concerned about getting clients for Rockdale.
Mike Colonnese (Analyst at H.C. Wainwright)
Great. Thank you for all the color.
OPERATOR
Thank you. Our next question comes from John Todaro with Needham. You may proceed.
John Todaro (Analyst at Needham)
Hey guys, thanks for taking my question. Congrats on the progress this quarter. I just want to go back on the capex and financing part. You listed out some of the mix on bitcoin sales and AI cloud prepayments. Does this suggest deliberately that there shouldn’t be expected equity financing in Q4 ’26 or into Q1 ’27?
Michael G Potter (Chief Financial Officer)
I mean, we’re not going to tie our hands absolutely, because you don’t know what’s going to happen in the world. But the ATM we wound down sort of at the tail end of Q2—that extended a little bit into Q3—and we haven’t been accessing the market for equity recently at all. As we said in our prepared remarks, we do plan on linking contracted cash flows to do more normal project financing, and our GPU-as-a-service, a combination of the large customer prepayments and shorter-term financing.
So we’re not trying to stretch to the limit over five years. We believe that will be doable in the market today and provide ample capital for us to execute on our GPU opportunities. Equity financing we’re trying to limit to earlier-stage investments that typically would be funded by equity. And right now we’re concentrating on executing on our revenue opportunities that are contracted for, that are directly in front of us.
John Todaro (Analyst at Needham)
Understood, that’s appreciated. And then on the Wethersfield, Ohio site, I believe it says you own 40 acres there and that site could go to 300 megawatts. Do you need to buy more acreage there, or would that be enough to support a 300-megawatt either PSP or cloud data center design?
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
That piece of land right now is in the middle of local community discussion. I think we still need more time to get permission from them if we can. And we’ve already seen some other local communities approving data centers after a very good community contribution plan. I think we want to do that in the near future. But on the land-wise, I don’t think that we really need to buy land. We can do a data center with the existing.
John Todaro (Analyst at Needham)
Understood. Thank you. Appreciate it.
OPERATOR
Thank you. Our next question comes from Greg Lewis with BTIG. You may proceed.
Greg Lewis (Analyst at BTIG)
Yeah, hi. Thank you and good afternoon, good morning, good evening. Thanks for my question. I had two questions. The first is around the power footprint. I mean obviously it’s global, it’s big. I guess two questions in there. Are there thoughts about—and hey, we have a lot of pipeline and the existing footprint—but are there thoughts about continued expansion? And then also, is there also thoughts about subtraction? Right? I mean, there’s opportunity—the value of land, the value of power has gone up.
Are there sites now that maybe the company would think about kind of high-grading as they look at positioning the portfolio over the next kind of one, two, three years?
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Michael, could you answer the question?
Michael G Potter (Chief Financial Officer)
Sure. We’re not looking at selling any of our powered land. As you noted, the value for that’s gone up, and it’s because of scarcity, and we don’t think that turning powered land into some short-term gain when we have very long-term plans to keep growing the company [is right]. We do have the advantage of a very robust and strong ability to mine bitcoin with leading ASIC development and the actual miners and in great position. So if we need to wait a little bit until we can actually do an AI data center or something different, we can still generate pretty good cash flow from the sites we have.
So I wouldn’t think that we’re going to try and turn any of the land we currently have into a short-term gain. Again, nothing is absolute. I guess if the pile of cash was big enough and Jihan decided it was the right thing to do, we would consider it. But right now we’re pretty comfortable with our position.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Well, we do feel everything will be slowed down in Norway, but I think it’s too early to say that data centers will be impossible to build in a way, because AI is like the next industrial revolution; it is very fundamental elements. I don’t think Norway will give up fully, but we do see that everything will be slowing down in permission, in approval.
OPERATOR
Thank you. Our next question comes from Mike Grondel with Northland. You may proceed.
Mike Grondel (Analyst at Northland)
Hey, thank you, guys. Michael, you mentioned something. $350 million left at Tydal. Could you just repeat what you said about that? And then do you have a rough range for the financing needed at Tydal? If you could help us with that, that would be great.
Michael G Potter (Chief Financial Officer)
We said we’d be spending about $500 million to complete the construction of Tydal. And that’s just a moment-in-time measurement of what’s left to be spent in terms of between phase one and phase two. Obviously, the activity is relatively heightened. So phase one is very, very close to completion now, with us being a couple weeks into November and the December 31st deadline. So that’s just an indication of how much is left to do in terms of the spending, not the actual construction activity, which is quite advanced.
Financing-wise, if we’re spending $500 million, we’d want to get at least $500 million. And we’ve indicated that we’re going to be looking to get more financing to pull out some of the equity that we’ve put into the site in the past. The total amount that would be financed is always going to be dependent on market conditions and when we do go out, what the reception is to our plans. But we do have a plan to take out more than we’ve put in for recent capex.
Mike Grondel (Analyst at Northland)
Got it. And any update on Fox Creek or Clarington?
Michael G Potter (Chief Financial Officer)
No update. Sorry. Sorry. Go ahead, Jihan.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Yeah, Michael?
Michael G Potter (Chief Financial Officer)
Yeah. No update on Clarington. It’s the same status it’s been for a while. We continue to pursue, you know, relief against the lawsuit that was brought against us that we don’t really think has a good basis. And we’re currently planning on using it for bitcoin mining. We believe we can get good value out of that site for bitcoin mining if we don’t do an AI data center there. So our current plan is bitcoin mining at Clarington. Fox Creek: we continue to build the power plant and it’s expected to be in production mid to end of next year. We’ve done some behind-the-scenes work to secure gas supplies and to make the electricity generation more robust. So whereas the data center hall was being built for joint use between AI and bitcoin mining, the power would have been a little bit potentially intermittent for AI, but we’ve kind of removed that in the background. So we’re working to get Fox Creek online.
And that’ll be a 100-megawatt site mid to third quarter of next year.
Mike Grondel (Analyst at Northland)
Thank you.
OPERATOR
Thank you. Our next question comes from Brian Kinzlinger with AGP. You may proceed.
Brian Kinzlinger (Analyst at AGP)
Great. Thanks so much for taking my questions. I’m curious, once at scale, what is the target gross margin, EBITDA margin range for the AI cloud business?
Michael G Potter (Chief Financial Officer)
So we’ll talk about the financial targets and the model in a lot more detail when we report the Q4 results. So right now, reaching scale will be happening next year and it’ll be part of our guidance for next year.
Brian Kinzlinger (Analyst at AGP)
Okay, and then can you speak at a high level to co-miner and how you’re thinking about production and capital allocation going forward?
Michael G Potter (Chief Financial Officer)
Jihan or Haris, do you want to talk about our co-miner?
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Yeah. We will continue to invest into the expansion of mining operations. I think next year we will slow down a little bit on the pace of investing in new capital, but we will continue to expand our mining operations.
Brian Kinzlinger (Analyst at AGP)
Okay, thanks.
OPERATOR
Thank you. Our next question comes from Steven Glagola with TD Cowen. You may proceed.
Steven Glagola (Analyst at TD Cowen)
Hey, thanks for the question. Haris and Michael, at Rockdale, can you clarify what infrastructure and related rights were transferred to Bitdeer Technologies as part of the $100 million transaction for the 200 acres? And specifically, did the purchase include SLR’s interconnection rights and water rights and the SLR-owned substation?
Michael G Potter (Chief Financial Officer)
Thank you. So it didn’t include the water rights, but we do have an arrangement to get water to the site as part of the deal. But the water rights themselves were not transferred over. In terms of power, we have control over the interconnection now and the ability to transfer power from our existing site to the new site.
Steven Glagola (Analyst at TD Cowen)
Okay, great. And did you guys purchase the substation as well?
Michael G Potter (Chief Financial Officer)
We have an agreement to share the substation because SLR does use a small part of the substation that we mainly use. So it’s more formal and with us more in control versus them being in control as our landlord.
Steven Glagola (Analyst at TD Cowen)
Thank you.
OPERATOR
Thank you. And as a reminder to ask a question, please press star-1-1 on your telephone. Our next question comes from Kevin Dede with JonesTrading. You may proceed.
Kevin Dede (Analyst at JonesTrading)
Good morning. Thanks for taking my call. Appreciate it. Kind of curious about how you see Malaysia developing. I know there’s been a big focus on the AI cloud business, and I’m wondering if you see Bitdeer in an advantageous position competing for, I think, what you said is underserved AI startups in that region versus other competitors, perhaps, and maybe offer a little comparison with how you see that business evolving in the U.S.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
I think it’s quite good for us to be able to expand in Malaysia, especially in this kind of midterm uncertainty. U.S., we will be able to expand our AI cloud business on time without worrying about our own data center construction pace. And right now most of the U.S. companies, Asian-based tech companies, are all looking to Malaysia capacity. Our U.S. sales team get lots of inbound demand inquiries about our capacity in Malaysia because right now it’s really hard to rent any data centers in the United States within next year’s RFS.
So I think it’s really great that we can lock lots of capacity in Malaysia and continue to lock more and grow the AI cloud there, and while at the same time our U.S. data centers have been built and developed on the U.S. speed. So we don’t need to really worry about whether our AI cloud can grow or not. We can just execute and fulfill the demand from our customers.
Kevin Dede (Analyst at JonesTrading)
Thanks, Jihan. I’m also curious about wafer allocation from TSMC and your plan to use it. I know last quarter you mentioned Scrypt mining, and I’m wondering how you’re looking at Zcash and the wafer availability itself and associated pricing. Thank you.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
I think our wafer availability is a little bit tight, much tighter than before, because of the strong demand from AI. So I think we will slow a little bit on the capex crypto mining business next year, and we have no plan for doing Zcash. And Scrypt mining right now is really good for us because, in terms of revenue dollar per megawatt, it is higher than bitcoin mining, and we will have an edge over other competitors in terms of the mining machines in this algorithm.
So we will continue to do more Scrypt mining.
Kevin Dede (Analyst at JonesTrading)
Well, okay, I appreciate you. The feedback I’ve gotten is that Zcash is incredibly profitable, and there’s a very huge shortage of machines given that there only seems to be one manufacturer who you know well.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Yeah. Yes. Yeah. But the competition for Zcash mining, I think, will also be very fierce. Lots of potential new mining developers are already in the market to tape out and to order lots of wafers from Samsung and TSMC. So I think to strategically give up Zcash is good for us.
Kevin Dede (Analyst at JonesTrading)
Perfect. Thanks, Jihan. Really appreciate the color. Congratulations on a great quarter.
Jihan Wu (Founder, Chairman, and Chief Executive Officer)
Thank you.
OPERATOR
Thank you. I would now like to turn the call back over to Michael Potter for any closing remarks.
Michael G Potter (Chief Financial Officer)
Thank you, everybody, for joining our call today, and I look forward to talking to you in the future.
OPERATOR
Thank you. This concludes the conference. Thank you for your participation. You may now disconnect.
Disclaimer: This transcript is provided for informational purposes only. While we strive for accuracy, there may be errors or omissions in this automated transcription. For official company statements and financial information, please refer to the company’s SEC filings and official press releases. Corporate participants’ and analysts’ statements reflect their views as of the date of this call and are subject to change without notice.
