Canaan Inc - ADR Stock

Ladies and gentlemen, thank you for standing by and welcome to Canaan, Inc.'s Third Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note that this event is being recorded.

Now I'd like to hand the conference over to your speaker today, Gwyn Lauber, Investor Relations for the company. Please go ahead, Gwyn.

Thank you, operator. Hello, everyone, and welcome to our earnings conference call. Joining us today are our Chairman and CEO, Nangeng Zhang; and our CFO, Jin James Cheng. Leo Wang, Vice President of Capital Markets and Corporate Development; and Xi Zhang, Senior IR Manager, will also be available during the question-and-answer session. Our CEO will start the call by providing an overview of the company and performance highlights for the quarter. Our CFO will then provide details on the company's operating and financial results for the period before we open up the call for your questions.

Before we begin, I would like to refer you to our safe harbor statement in our earnings press release. Today's call will include forward-looking statements. These statements include, but are not limited to, our outlook for the company and statements that estimate or project future operating results and the performance of the company. These statements speak only as of today and the company assumes no obligation to revise any forward-looking statements that may be made in today's press release, call or webcast except as required by law.

These statements do not guarantee future performance and are subject to risks, uncertainties and assumptions. Please refer to the press release and the risk factors and documents we file with the Securities and Exchange Commission, including our most recent annual report on Form 20-F for information on risks, uncertainties and assumptions that may cause actual results to differ materially from those set forth in such statements.

In addition, during today's call, we will discuss both GAAP financial measures and certain non-GAAP financial measures, which we believe are useful as supplemental measures of the company's performance. These non-GAAP measures should be considered in addition to and not as a substitute for or in isolation from GAAP results. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP results, in our earnings press release, which is posted on the company's website.

With that, I will now turn the call over to our Chairman and CEO, Nangeng Zhang. NG , please go ahead.

Thank you, Gwyn. Hello, everyone. This is NG. Thank you for joining in today's earnings call. Over the past few quarters, we have discussed Canaan's strategic transformation on several occasions. Today, I would like to begin with the changes taking place in our industry, then explain how we are adapting our operations and share the latest progress in our transformation. First, the bitcoin mining hardware industry is undergoing profound structural changes. Demand remains weak, industry inventories are elevated and hardware selling prices remain under pressure.

At the same time, rapidly growing AI demand is absorbing significant advanced node chip manufacturing capacity, driving wafer fabrication costs sharply higher and tightening supply. As a result, even higher performing next-generation mining machines may be difficult to produce at a reasonable cost and generate adequate commercial returns. We expect these factors to constrain the production and deployment of new mining machines over the next few years, making it difficult for bitcoin network hash rate to sustain its fast pace of growth.

For operators with low-cost power and existing equipment, this could improve long-term mining economics. In response, Canaan is adjusting its approach to R&D, manufacturing and capital investment. Over the past 2 quarters, several of our next-generation mining ASIC development projects have been completed or are substantially complete. We have received some engineering samples with test results meeting our design expectations and additional samples continue to arrive. We believe these technological advances are sufficient to meet market needs over the next several years.

This enables us to significantly reduce ongoing R&D spending on new products while retaining the ability to begin mass production when commercial conditions are appropriate. We will continue to pursue economically attractive opportunities in mining hardware sales and bitcoin mining, but we will no longer rely on sustained large-scale investments and successive product generations to drive the company's growth. Canaan's long-term focus is shifting towards energy and AI HPC computing infrastructure. This transformation rests on a clear business foundation.

Through years of bitcoin mining operations in North America, we have developed expertise in power procurement, site development, equipment deployment and operations management while building long-term relationships with utilities, local communities and partners. Since 2025, we have deliberately applied these capabilities to developing power resources on a large scale. As demand for AI infrastructure grows rapidly, land and infrastructure with reliable access to power are becoming critical to the industry's development.

Today, I'm pleased to share an important milestone. Our 1 gigawatt large-load project in Duval County, Texas, recently received a conditional Studied Load - Firm classification under ERCOT's Batch Zero process. This achievement reflects our team's sustained development efforts and investment and marks a substantive step forward in the grid interconnection process. We are continuing to advance the remaining grid interconnection procedures. On the commercial development side, we are engaging in deep discussions with a U.S. institution with an investment-grade credit profile regarding a potential strategic partnership.

We believe partners with strong capital resources, credit support and the ability to deliver large infrastructure projects can help Canaan participate more efficiently in major energy infrastructure developments. This is particularly important for a company of our size. By drawing on our expertise, we can play a meaningful role in the value chain of large projects while controlling our capital commitment and risk and earning returns with our contribution. We aim to build this collaborative development model into a sustainable business over the long term.

The significance of this progress extends well beyond a single project. AI data center infrastructure represents a substantial market opportunity; but it also requires significant capital, specialized expertise and long development and construction time lines. Bitcoin mining as another form of high-density computing application offers greater flexibility. It can help the grid balance peak and off-peak power demand and provide additional value to local communities through heat recovery and other uses. Over the long term, Canaan believes power resources can serve different purpose.

For projects suited to AI HPC, we will prioritize the corresponding development and partnership opportunities. For lower-cost power resources that are better suited to bitcoin mining, we are also open to prioritize in bitcoin mining. We will assess the economics and development conditions of each project and the flexibility to select the business model that creates the greatest value. In addition, Canaan has historically adhered to our philosophy of developing jointly with partners across the ecosystem. We believe this approach remains equally applicable to today's AI HPC development project and can help the company go further and more steadily.

At the same time, we are advancing another 1 gigawatt power resources project in Zapata County, Texas. This project is in early stage development with preparations for grid studies underway. The milestone in Duval and our continued process in Zapata demonstrate that we are gradually building both a pipeline of power resources projects and ongoing development capability in North America. We aim to translate the industry relationships and execution experience we have accumulated over the years into an infrastructure business that creates sustained value.

Finally, let me turn to capital discipline. We are taking firm action to control operating expenses. further reduce administrative costs and R&D spending and strictly manage new production commitments and capital investment. For large infrastructure projects, we will prioritize structures that draw on our expertise, bringing in partner capital and allocated risk appropriately. Our goal is to use less of our own capital to participate in higher value projects while gradually improving shareholder value and capital efficiency.

I want to emphasize that Canaan's transformation goes beyond the shift from bitcoin mining to AI HPC data centers. More fundamentally, we are shifting our focus from capital expansion to generating returns on capital and ultimately enhancing shareholder value. This requires us to apply strict capital discipline to our treasury policy, R&D spending and capital investment planning; allocating capital only to opportunities that can clearly create shareholder value at a manageable level of risk. We must be equally disciplined about taking on debt or diluting shareholders. We will pursue growing only when it compounds shareholder value and strengthens our appeal to institutional investors.

The operating environment remains challenging, but Canaan's transformation is moving from strategy to execution. Building on years of technological development, we have completed a major phase of R&D and begun to make tangible progress in energy infrastructure. Our business priorities and path forward are becoming clearer and we are -- I am confident in this direction. We will remain focused on project execution, capital efficiency and commercial returns, demonstrating the value of this transformation through sustained operating results.

Thank you, everyone. Now I will turn the call over to James to review our quarter 3 financial results.

Thank you, NG, and hello, everyone. This is James speaking in our Singapore headquarters. Just now, NG described the 2 shifts underway at Canaan. First is from bitcoin mining toward energy and AI HPC infrastructure and more fundamentally, from capital expansion toward capital returns and shareholder value. My remarks today will show how both are reflected in our numbers this quarter. First, the capital we have committed to our energy projects, then our treasury and shareholder actions and finally, our third quarter results and outlook.

The clearest example this quarter is our Duval County project. As of September 30, we held $50 million in cash collateral for a letter of credit in relation to power projects. This is a concrete commitment of capital behind the milestone NG just described. The funds remain on our balance sheet as restricted cash. Cash and cash equivalents were $22.8 million as of September 30. The decline from the end of June mainly reflected the funds we moved into the collateral account. Including the $50 million collateral, our total cash position was $72.8 million, up 10.3% from the end of June.

Beyond the collateral, we had approximately $3.3 million in project prepayments as of September 30 related to our infrastructure plans. In October, we also completed an approximately $1.1 million power-related land purchase for our 1 gigawatt power resource project in Zapata County, Texas. Securing suitable land is one of the core development capabilities. This piece of land gives us a base for application for power interconnection and reflects our commitment to building a lasting presence in the United States.

For large infrastructure projects, we intend to bring in capital from partners and minimize development uncertainty so that we can participate in high-value projects with less of our own capital. The same discipline guides how we manage our treasury. NG also emphasized discipline in our treasury policy. During the third quarter, we monetized our Ethereum holdings and a portion of our bitcoin holdings generating approximately $40 million in cash. The average selling prices were approximately $79,000 per bitcoin and $2,400 per Ethereum. Of these proceeds, we used $5.4 million to repurchase 13.6 million ADSs.

We regularly review our treasury strategy weighing the potential returns from holding digital assets, investing in our businesses and repurchasing our shares. We believe our shares are undervalued relative to our long-term value. So we prepared to allocate a portion of our capital to buybacks. We believe our current share price does not fully reflect our long-term value and share repurchases are an important capital allocation priority for us. Our Board has approved a new share repurchase authorization of up to $100 million over the next 12 months, under which we are launching an initial program of up to $20 million.

As our available funds grow, we intend to pursue repurchases more actively. The pace and scale of repurchases will depend on our financial position, market conditions and applicable regulations. We are also planning an ADS ratio change. The change is designed to preserve the overall economic interest represented by investors' ADS holdings. By increasing the per ADS trading price, we believe it could make our ADSs more accessible to institutional investors whose investment policies limit holdings of lower priced securities.

As our business increasingly centers on energy and computing infrastructure where progress is measured over quarters rather than months, we have discontinued our stand-alone monthly mining report. Key mining metrics will be included in our quarterly results. We will continue to announce material developments separately when timely disclosure is required under applicable rules. During the third quarter, we appointed KPMG in Singapore as our auditor succeeding KPMG in Beijing. I would like to thank the KPMG Beijing team for their professionalism and support and welcome the KPMG Singapore team. The change aligns our audit arrangements more closely with our increasingly international operations.

Now let me turn to our third quarter results. For the third quarter, we reported revenue of $14.5 million within our guidance range of $11 million to $15 million. Net income was $8.0 million driven primarily by $51.9 million in fair value gains on digital assets and related financial derivatives reflecting the recovery in bitcoin prices during the quarter. Adjusted EBITDA was $22.4 million and also benefited from these gains. If bitcoin prices continue to trend upward, the appreciation of our accumulated digital assets could further contribute to our profitability. This aligns to our digital asset treasury strategy.

Our mining gross margin improved significantly to close to 30% in the third quarter, up from about 20% in the second quarter. We generated 133 bitcoins during the quarter. Even after monetizing a portion of our holdings, we held approximately 1,875 bitcoins at September 30. By maintaining a substantial bitcoin treasury through the downturn, we were able to participate in the subsequent price recovery and the increase in the value of these holdings made an important contribution to our results this quarter. Total operating expenses were $34.2 million, down 15.5% from the prior quarter.

As NG mentioned, several of our next-generation mining ASIC projects have been completed or are nearly complete and we believe these results can support market demand over the coming years. As a result, we no longer need to heavily rely on continuous large-scale R&D investments to drive growth, which allows us to continue managing our operating expenses prudently. Our inventory declined to $119 million on September 30, marking its lowest level in the past 5 quarters. Given the demand environment and rising wafer costs, we are carefully managing new production commitments and aligning purchases with actual customer demand.

For Project ABC, we recognized $1.6 million in equity investment income during the quarter. This gain was noncash and included acquisition-related accounting adjustments. In cash terms, we received $1.9 million from ABC comprising $1.2 million in distributions and $0.7 million in sales proceeds. For us, these cash receipts are the meaningful measure. Our focus remains on ABC's operating performance and its ability to generate real cash returns over time.

Beyond our own mining operations, we are also expanding the applications of our mining equipment to help customers get the most value from every megawatt used. In Northern Europe, an 8-megawatt district heating project repurposes waste heat from liquid cooled computing equipment to provide heating for approximately 2,800 households. We are also exploring greenhouse heating in Canada, home heating through Avalon Home and using associated gas from oil fields that would otherwise be flared.

In summary, this quarter is a profitable quarter with committed capital to our energy strategy, return of capital to shareholders and tight control over costs, expense and inventory. As NG said, our transformation is moving from planning to execution. On the financial side, we will continue measuring the progress by capital efficiency and returns. We will remain disciplined in our use of capital and pursue growth only where it compounds value per share.

Turning to our outlook. We expect total revenue for the fourth quarter of 2026 to be between $18 million and $22 million. This outlook reflects our current assessment of the market and the operating conditions. Actual results may vary with changes in bitcoin prices, industry demand and other market conditions.

Thank you. Operator, we are now ready to take questions.

[Operator Instructions] Your question comes from Kevin Dede from JonesTrading.

This is Sky Moore calling in for Kevin Dede. My first one is with the next ASIC team largely complete, what specific market conditions would lead you to start mass production? Where should R&D spending settle and how do you plan to work down the remaining inventory?

I think over the past 2 quarters, we have completed or substantially completed several next-generation mining ASIC development. And during this quarter, we have received some of the engineering samples and we have tested them and the test results meet our design expectations and additional samples are arriving. So I think this technology portfolio can support the market demand over the next several years. Our next step is to determine when commercial conditions is justified bringing these technologies into volume production.

Advanced node wafer costs have risen significantly while mining equipment demand remains very weak. So we will, therefore, consider customer demand, production cost and expected returns when setting the production schedule. So I think this technology portfolio allow us to substantially reduce ongoing mining ASIC R&D spending while retaining the ability to launch new products where appropriate. So that is consistent with our broader transition, continue to pursue economically attractive mining equipment opportunities while directing more resources towards energy and AI HPC infrastructure.

Great. My next question would be with most of the bitcoin pledged against loans and unrestricted cash floating around about $23 million, how do you guys plan to fund that $20 million buyback alongside the Duval buildout and what would lead you towards the full $100 million?

I think we believe current share price does not fully reflect the company's long-term value. So buybacks are a very important capital allocation priority for the company. After providing normal operations and also committed project funding needs, we would like to become more active in doing repurchasing shares. I think we have already taken action in the third quarter. We generated approximately $14 million in cash from monetizing Ethereum and a portion of our bitcoin holdings and used already $5.4 million of those proceeds to repurchase shares.

So recently our Board approved a new program of up to $100 million over the next 12 months. So we started from a $20 million program and I think our direction is very clear. As long as we continue to see the attractive value in the shares, buybacks will be one of the priority uses of additional available funds. I think the timing and the size of repurchase will reflect our funding position, market conditions and applicable regulatory requirements. So we will take pace and we will consider the market situation and step-by-step do this stock repurchase program. Hope that answers your question.

Your next question comes from Ben Sommers from U.S. Bancorp BTIG.

So I wanted to ask on the Duval County site. Just wanted to ask a bit more on the discussions you're having with a potential partner there. Would this be a potential partnership where you could give up some equity in the site? Would the partner kind of handle the development process? Just how do you think about what you want from that strategic partner?

Yes. We are considered by confidentiality agreements and are not in position to disclose further details at this time. But our objective is to generate positive cash flow from these projects while would support the share repurchase and provide funding for subsequent such power resources development, land acquisition and related activities.

Super helpful. And then switching over, talking about the heating applications of the ASICs. Can you just talk a bit about how that is being incorporated into the development pipeline here as we also manage the inventory and try to optimize the balance sheet?

Let me see. Yes, we have a corporate in North Europe. Currently, we already deployed several megawatts of our equipment generating over 80 degrees Celsius for about 2,800 families for the heating and now it already generates USD 1.1 million in revenue. It is about 8 megawatts hydro cooled mining equipment project we already delivered and in operation. It uses recovered mining heat to serve about 2,800 households. These applications create value from both mining and heating and remain an area we are developing. Still we have ongoing contracts in negotiation and I think the scale of the heating program will scale up.

Your next question comes from Kevin Cassidy from Rosenblatt Securities.

The midterm elections are top of mind for investors. And I just want to understand how Canaan is set up for the Clarity Act and it doesn't look like that would pass and whether you have -- there's a plan B on digital assets.

I think overall we expect the impact to be limited. The Clarity Act mainly concerns regulatory framework for digital assets market and our existing operations do not depend on its passage. And also, our primary focus is shifting towards energy and AI HPC infrastructure where its main drivers are power resources, project economics and customer demand. We will continue to monitor policy developments, but they do not change our current strategic direction.

Okay. And maybe as long as you're mentioning energy and you have these facilities in Texas, are you expecting any issues with the change in the government in Texas, the governor?

I think currently there's so many news comes out in this week. So now I think we are looking closely to any kind of changes in the policy especially in Texas. But personally, I think the energy infrastructure is quite useful and I think it's good for America. So our focus is on the early-stage energy infrastructure development. We are not building -- currently we are not focusing on like building big AI HPC data centers also. So we expect to earn returns that reflect our contribution while keeping the capital requirements and risks manageable.

Your next question comes from Nick Giles from B. Riley Securities.

Good to see some continued progress at these Texas sites. I just was curious if you could outline some of the next steps before the classification becomes final at Duval and how we should think about the ultimate timing of energization there and how much power you think would be available initially?

Yes. I think we are advancing 2 1-gigawatt power projects in Texas. The Duval County project has received a conditional Studied Load - Firm classification in the ERCOT's Batch Zero process and we are progressing the subsequent grid procedures. And the Zapata County is in early stage development and the preparation for grid studies. That means it will be a few months later than Duval. So for timing, we encourage investors to follow ERCOT announcements for the next stage of grid studies and capacity allocation as well. We are currently looking towards 2027 when the specific schedule subject to ERCOT's subsequent update.

Okay. Makes sense. Maybe just next with bitcoin up about 30% since the end of the second quarter, what kind of visibility do you have on demand as we look to 2027? And then what's driving the step-up in revenue in the fourth quarter?

To be honest, we do not have yet seen a meaningful recovery in mining equipment demand. Higher bitcoin prices have not translated sufficiently into new equipment orders. Customers remain very focused on payback periods and the current mining returns and the equipment price are not yet attractive enough to support large-scale new purchases. So we remain cautious about conventional mining equipment demand in the fourth quarter and we'll schedule production and deliveries against actual orders. Our current Q4 total revenue guidance is USD 18 million to USD 22 million. That includes mining equipment sales and other business revenue. So it should not be interpreted as evidence that demand for new machines has recovered.

Your question comes from Ian Generous from Compass Point Research & Trading.

This is Ian Generous calling in for Michael Donovan. So firstly, just looking on inventory. How Canaan is positioned today, including finished miners, chips and wafers on hand and then any outstanding purchase commitments?

Let me see. I think by the end of third quarter, the carrying value of our inventory was approximately USD 8.2 million, down significantly from USD 25.3 million in second quarter. The details of our purchase commitments are commercially confidential so we cannot disclose the specific amount. And some of the wafers are secured earlier with price relatively low and we are deferring delivery with our suppliers' support. I think further write-down remains a risk, but the launch of next-generation product is not the only consideration.

The value of existing inventory also depends on whether new machines can be produced at scale to competitive cost, actual market selling prices and customer demand. With wafer costs now substantially higher, even a more energy-efficient next-generation product may not be able to reach the market and scale at sufficiently low price. We will continue to assess inventory value and record any necessary write-downs or provisions as appropriate. At the same time, we will align production with orders and control new purchase commitments and inventory risk.

That's helpful. And then looking at the demand for mining rigs across North America and the rest of the world, our customers, do you see are they pausing orders and how are they financing their purchases?

Okay. I will say bit more about this. I think weak demand for mining equipment is a global issue not only North America driven primarily by inadequate investment returns. I think we can do a simple calculation can illustrate this challenge. At current mining revenue of approximately USD 0.023 per terahash per day and in average the electricity cost is about $0.015 per terahash per day. So the gross margin is only about $0.017 remaining after electricity. So assuming revenue and the electricity costs stay constant and before other operating and financing costs, the equipment price would have be about $6 per terahash.

Based on our understanding of industry costs, that is well below the cost of any existing industry mining equipment inventories across the industry. So producing new machines at this much, much higher wafer prices would widen the gap between production cost and the prices customers can justify paying. AI is also competing for power and sites, making low-cost mining locations harder to secure in some regions and putting upward pressure on electricity prices. Even customers who can obtain financing may therefore struggling to put the deployment opportunities that offer reasonable returns. So you can see the trend in network hash rate.

The total network hash rate is a constant with this assessment. Hash rate has declined overall from its 2025 peak with a period of recovery, but it has not returned to that height. Also based on our market observations, mining equipment used for electronic heating in Mainland China has been switched back on as winter approaches so contributing a seasonal increase in hash rate. This primarily reflects existing equipment retaining into operation rather than new purchases. So taken together; net hash rate, hash price and equipment selling price; indicate to us that the market for new machines are still contracting. So we are not optimistic about a near-term recovery in demand.

I'd like to turn the call back over to the company for any closing remarks.

Thank you, operator, and thanks, everyone, for joining us today. If you have any additional questions, feel free to reach out to us. And we look forward to speaking with you throughout the quarter. Thanks very much.

This concludes today's conference call. Thank you for participating. You may now disconnect.

Ladies and gentlemen, thank you for standing by. Welcome to Canaan Inc.'s Second Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note that this event is being recorded.

I will now hand the conference over to your speaker today, Gwyn Lauber, Investor Relations for the company. Please go ahead, Gwyn.

Thank you, Operator. Hello, everyone, and welcome to our earnings conference call. Joining us today are our Chairman and CEO, Nangeng Zhang, and our CFO, James Jin Cheng. Liang Wang, Vice President of Capital Markets and Corporate Development, and Xi Zhang, Senior IR Manager, will also be available during the question and answer session. Our CEO will start the call by providing an overview of the company and performance highlights for the quarter. Our CFO will then provide details on the company's operating and financial results for the period before we open up the call for your questions.

Before we begin, I would like to refer you to our Safe Harbor Statement in our earnings press release. Today's call will include forward-looking statements. These statements include, but are not limited to, our outlook for the company and statements that estimate or project future operating results and the performance of the company. These statements speak only as of today, and the company assumes no obligation to revise any forward-looking statements that may be made in today's press release, call, or webcast, except as required by law. These statements do not guarantee future performance and are subject to risks, uncertainties, and assumptions.

Please refer to the press release and the risk factors and documents we file with the Securities and Exchange Commission, including our most recent annual report on Form 20-F for information on risks, uncertainties, and assumptions that may cause actual results to differ materially from those set forth in such statements.

In addition, during today's call, we will discuss both GAAP financial measures and certain non-GAAP financial measures, which we believe are useful as supplemental measures of the company's performance. These non-GAAP measures should be considered in addition to, and not as a substitute for, or in isolation from, GAAP results. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP results, in our earnings press release, which is posted on the company's website.

With that, I will now turn the call over to our Chairman and CEO, Nangeng Zhang.

NG, please go ahead.

Thank you, Gwyn. Hello everyone, this is NG, CEO of Canaan. Thank you for joining our earnings conference call today. James, our CFO, and I are here at our Singapore headquarters to share our financial results and recent business updates for the second quarter of 2026.

Q2 2026 remained a difficult environment for the Bitcoin mining industry. In the first half of the quarter, Bitcoin prices recovered from approximately USD 62,000 to USD 82,000, before declining sharply and reaching a period low of about USD 58,000 at the end of the quarter. During the quarter, hash price fluctuated between USD 0.028 and USD 0.039 per terahash per second per day and remained at low levels.

At the same time, in the U.S., the capital markets and the traditional mining companies continue to shift more attention toward AI and HPC. Localized armed conflicts in certain countries and regions, and the tightening mining related policies in China and elsewhere are factors that affected miners' investment willingness and capacity. From the second half of the quarter onwards, miner sales weakened noticeably and elevated industry inventory levels further intensified price competition.

During the quarter, the company generated total revenues of approximately USD 32 million, below our previous guidance range of USD 35 million to USD 45 million. We sold 2.5 exahash per second of computing power, and our mining business produced 243 Bitcoins. When we issued guidance in May, our demand outlook was primarily based on the market conditions in the first half of the quarter. We did not fully expect the later decline in the demand and pricing. In the second half of the quarter, Bitcoin prices fall quickly. Miners become more cautious about equipment purchases, and high industry inventory levels increase price competition. These were the main reasons revenues came in below expectations.

In response to the revenue decline, we further tightened our spending, strengthened cash flow and liquidity management, and continued organizational optimization. At quarter end, the company held 1,915 Bitcoins and 3,952 ETH, bringing our digital assets treasury to another record high.

I'll start with our mining machine business. As mining economics weakened, many miners delayed equipment purchases, and the average selling prices remain under pressure. During the quarter, we generated approximately USD 14 million in product revenue. In response to softened demand, we adjusted pricing more flexibly. At the same time, we again emphasize production based on actual sales. We control new production according to real orders and put more focus on inventory management, cash flow, and order quality. For existing machines, we will evaluate whether to sell them to customers or deploy them in our own mining operations, based on cash collection, deployment conditions, and additional investment required.

Now let me turn to our mining operations. In Q2, we adjusted our deployed hash rate in a timely manner based on power prices, load management, and operating conditions at different sites. We allocated resources to projects with better economics. Mining revenue was approximately USD 18 million, accounting for more than 50% of the company's total revenue in Q2. And continue to cover direct operating costs such as power and hosting. As of the end of June, our installed hash rate in non-JV projects was approximately 10.05 exahash per second and the average all-in power cost in June was about USD 0.043 per kilowatt hour. Overall power and hosting costs remained relatively competitive.

Project ABC remains one of our key priorities. For Project ABC, we focused not only on current period profit, but more importantly on the cash it can generate and the long-term returns that can be created after optimization. This is consistent with the transformation the company is pursuing, whether it is Project ABC or the longer term power resources, we are developing. Our goal is to gradually build mining and energy infrastructure assets that have a cost advantage and can generate sustainable cash flow.

In Q2, together with our partner, we continue the mining machine upgrade at Project ABC. The project generated positive cash flow and maintained efficient operations. At the end of July, installed hash rate at Project ABC reached 4.85 exahash per second, up 10% from the end of March. Through this project, while increasing hash rate and the cash generation, we also gained experience in operations, power dispatch, fleet upgrades, and the management of low-cost power resources.

Beyond traditional mining use cases, we continue to advance energy utilization partnerships such as compute to heat. Our earlier Nordic project has already validated the feasibility of using hydro-cooled equipment for district heating. Our high temperature hydro-cooled equipment can supply the hot water needed for heating, which is particularly suitable for winter heating demand. These projects are still relatively small in scale, but using the heat generated from computing for comfort heating is a useful exploration of compute-to-heat closed-loop.

Regarding R&D and the products, we continued advancing the A16 series in the second quarter with a focus on cost-effective air-cooled models and the high temperature hydro-cooled models. We pay close attention to products' full life cycle economics, including purchase cost, power consumption, stability, maintenance, and deployment efficiency. On the consumer side, the second quarter was mainly devoted to R&D on new Avalon Home products which we will prepare for mass production in the third quarter. These home series products are designed for home heating use cases and we hope to capitalize on the winter heating season in the northern hemisphere and deliver a solid sales performance.

Long-term power resources remain another key focus of the company's transformation. Over the past several quarters, we have been advancing long-term, stable, cost-advanced, and expandable power resources projects in North America. On our fourth quarter 2025 earnings call, we mentioned our confidence in our ability to secure substantial load by year end 2026, potentially reaching the gigawatt scale. Based on the progress we have made so far, our confidence in securing gigawatt scale load by the end of 2026 remains intact. We continue to work with all the stakeholders and hope to provide an update when we are in a position to do so.

Finally, let me discuss capital allocation. Since the beginning of the second quarter to date, the company has not utilized the ATM program or raise capital. Under the existing $30 million share repurchase authorization, we recently mentioned a portion of our digital assets and used the proceeds to repurchase the company's ADS. We believe that the current share price meaningfully undervalues the company while retaining the capital needed for operations and project environment -- investment.

We chose to use a portion of our digital assets for buybacks. We deployed around USD 2 million in the first quarter and around USD 5.4 million in August to repurchase our ADS shares. In 2026, the company deployed USD 7.4 million and repurchased approximately 16.4 million ADSs so far. James will provide more details on the execution of digital assets monetization and the buyback. James and I also continue to purchase the company's ADS in the open market this quarter. We remain confident in the company's long-term development and the transformation underway, and we hope these purchases further align management interest with those of our shareholders.

Digital assets remain an important part of the company's asset allocation. We will continue to evaluate the risks and the rewards of holding digital assets, investing in mining and the power infrastructure, and the repurchasing the company's shares. While meeting our operating and liquidity needs, we will choose the uses of capital that we believe can create the best long-term value per share.

Finally, during this quarter, the company completed its transfer to Nasdaq Capital Market and have been granted an additional 180-day grace period to regain compliance with the minimum bid price requirement with a deadline of January 11, 2027. We will continue to monitor the trading price of our ADS and take necessary actions to regain compliance and maintain the company's listing status.

Despite the significant impact of industry volatility on our second quarter financial results, changes in the macro environment have actually strengthened our conviction in Bitcoin as a decentralized financial asset. In response to market changes, we managed the inventory, expenses, and cash flow more strictly, while continuing to organize our mining fleet, advanced fleet upgrades at Project ABC, explore compute to heat applications, and prepare Avalon Home products for mass production. Work on power resources also continued. In addition, we monetized a portion of our digital assets to fund share repurchases. We believe we have come through the most difficult period and were able to sell these digital assets at relatively favorable prices.

Looking ahead to the third quarter, although Bitcoin price recovered somewhat at the end of August, miner procurement remains cautious, and the industry inventories still needs to be digested. Some competitors have adopt a more aggressive pricing strategy to speed up cash collection. And we expect miner sales and average selling prices to remain under pressure in the third quarter. Therefore, we remain cautious about the near-term mining machine market.

Based on the current market and operating conditions, we expect total revenues for the third quarter of 2026 to be between USD 11 million and USD 15 million. This outlook reflects management's current judgment, and actual results may differ due to changes in macroeconomic conditions, policies, Bitcoin prices, and industry demand. That concludes my remarks. Thank you.

I will now turn the call over our CFO, James.

Thank you, NG. Hello everyone. This is James speaking to you from our Singapore headquarters. NG just walked you through the market environment and our business progress during the second quarter. To summarize, in the second half of the quarter, Bitcoin prices and hash price weakened again, miners became more cautious with equipment purchases, and the elevated industry inventory added further pressure on pricing. This pressure directly impacted our financial results. Computing power sold, average selling price and revenue all declined, while lower market prices also affected the value of our inventory and fixed assets.

In this environment, we are not waiting for the market to turn. We are focusing on what we can control. From a financial perspective, we are focused on 3 things. First, managing cash and maintaining sufficient liquidity. Second, actively allocating capital including repurchasing our shares when we believe they are significantly undervalued. And third, managing and optimizing our strategic assets so they can generate stronger cash returns over time. Let me go through each of these areas.

First, the cash. As market conditions weakened, we tightened our expense and inventory management and placed even greater emphasis on cash flow. At the end of the second quarter, we had $66 million in cash, up about $23 million from $43 million at the end of the first quarter. So despite generating less revenue during the quarter, our cash position improved.

During the quarter, we collected $54 million in cash from product sales. We also received $15 million from value-added tax refunds, cash distributions from equity investments, and Bitcoin-backed financing. On the cash outflow side, we used $35 million for operating expenses and working capital, and another $11 million for wafer purchases. We also continued to exercise discipline on expenses. Total operating expenses were $40 million in the second quarter, including $9.2 million of the impairment charges on property, plant and equipment, and $2.7 million of credit losses.

Excluding these items, operating expenses were $28.2 million, down 9% sequentially, and 14% year-over-year. This is consistent with the build-to-order approach that NG discussed earlier, control new commitments, accelerate collections, manage inventory and working capital, and protect liquidity. In a down cycle, we believe financial resilience itself is a competitive advantage.

Second, capital allocation. At the end of the second quarter, we held 1,915 Bitcoin and 3,952 Ethereum. Based on their carrying value as of June 30, our digital assets holdings were worth $112 million. Digital assets remain an important part of our asset base, but we do not believe they should simply be held passively. We continuously compare different uses of capital, holding digital assets, investing in mining and power infrastructure, and repurchasing our own shares. When our shares trade significantly below what we believe to be their long-term intrinsic value, we view share repurchases as an attractive use of capital.

In late August, we were authorized by our board and sold all our Ethereum holdings and 54 Bitcoins, generating approximately $13.9 million in cash. We used a portion of these proceeds to repurchase our shares. In the first half of 2026, we repurchased 2.8 million ADSs for about $2 million. In late August, we repurchased an additional 13.6 million ADSs for $5.4 million. Year to date, we have repurchased 16.4 million ADSs for a total consideration of $7.4 million.

Third, strategic assets, especially mining sites in the United States. NG discussed the operational recovery and Bitcoin miner upgrades at Project ABC. Let me add a few points from a financial perspective. During the second quarter, we've been upgrading the mining fleet at Project ABC to improve efficiency, received $5.2 million in cash from Project ABC, including sales collections and cash distributions. As of August 31, cumulative cash received had reached $8.4 million. We would like to thank our partner, WindHQ, for the top-tier management and the results of Project ABC. Their long-term commitment and shared vision give us confidence in the future of ABC.

As part of this process, the retirement of older mining machines and the related accounting treatment resulted in a one-time loss. As a result, we recognized approximately $4 million of equity investment losses for Project ABC in the second quarter. There is an important distinction here between short-term accounting results and the long-term cash generating ability of the assets. We are not trying to preserve the book value of older equipment. We are trying to improve asset quality and generate more computing power and better economics from the same energy resources. So for project ABC or other strategic assets, we look beyond the current period earnings. We are focused on the cash the asset can generate and the returns it can produce after optimization. This is also consistent with the broader transformation NG discussed earlier.

Let me briefly add a few operating and financial metrics. Total revenue for the second quarter was approximately $32 million. Product revenue was approximately $14 million. We sold 2.5 exahash per second of computing power at an average selling price of $5.5 per terahash per second. Mining revenue was approximately $18 million with 243 Bitcoin mined during the quarter. Mining accounted for 55% of total revenue. Excluding depreciation, the gross margin for the mining business was 20%. These numbers also showed that during a weak market for mining machines, our mining operations have become an important contributor to revenue. Over time, we want to build a business that combines technology and mining machines, Bitcoin mining, power resources, and infrastructure.

Next, let me briefly discuss our income statement. Adjusted EBITDA was a loss of $74.9 million in the second quarter, broadly in line with the previous quarter. Our results included several significant non-cash accounting adjustments, primarily inventory write-downs, impairment of property, plant and equipment, and fair value losses on digital assets. These items reflect changes in the mining machine market and digital assets prices during the quarter, but they did not result in an equivalent amount of cash outflow.

So when we're looking at the quarter, I think it is useful to separate 3 things. First, the real operating impact of weaker demand and lower pricing. Second, the accounting impact from inventory fixed assets, digital assets, and investments, a significant portion of which was non-cash. And third, what we are particularly focused on today, cash flow, liquidity, capital allocation, efficiency, and the ability of our assets to generate cash returns.

Finally, turning to the third quarter, as NG mentioned, although Bitcoin prices have recovered recently, miners remain cautious, industry inventory still needs to be absorbed, and mining machine pricing remains under pressure. As a result, we remain cautious about the near-term market environment. Based on our current market and operating conditions, we expect the total revenue for the third quarter of 2026 to be between $11 million and $15 million. This outlook reflects our current assessment of market and operating conditions. Actual results may vary depending on changes in the macroeconomic environment, policy developments, Bitcoin prices, and industry demand.

Let me close with 3 words that summarize our financial priorities today: cash, value and assets. First, manage cash, maintain financial discipline and liquidity so that we can navigate through the cycle. Second, manage value, continue to compare different capital allocation opportunities. When we believe our shares are significantly undervalued, we are prepared to repurchase them to create value for long-term shareholders. Third, manage assets, whether it is our digital assets or Project ABC or the mining and energy infrastructures we are developing, our goal is not simply to own assets. Our goal is to make those assets more efficient, generate cash, and create long-term returns.

We will not build our strategy around predicting short-term movements in the price of Bitcoin. What we can do is maintain financial discipline during the difficult markets, improve capital allocation, and continuously upgrade the quality of our assets. So when the next industry cycle comes, we want Canaan to have stronger balance sheets, higher quality assets, stronger operating capabilities, and greater strategic flexibility.

Thank you. We will now open the call for questions.

[Operator Instructions] The first questions will come from the line of Logan Hennen from Northland.

First one from us. Can you provide some additional color to help us understand how we should be thinking about this 1 gigawatt pipeline? For instance, should we expect a dual deployment strategy balancing Bitcoin mine and HPC? And if your team does pursue HPC, should we expect Canaan to go the co-location or the GPU cloud rental route? Any color here would be great.

Thank you. I will take this one. Yes, we are currently advancing several power resources project internally. However, securing and developing power resources involve many steps, and it's become more and more complex. The market attention and the competition are both very high now. Even for the projects that are moving the fastest, currently there are still, key matters that need to be completed. So as a management team of a public company, we need to be very, very careful about when we disclose project details. If we disclose too early, it may affect project execution. It may also cause the market to view an ongoing process as a confirmed outcome.

So for the benefit for the company and our shareholders, we do not think this is appropriate at this stage to disclose the exact number and the location of the sites, the power capacity for each project or a specific stage of each project. What we can say is our view from the Q4 2025 earnings call earlier this year, based on the progress we have made so far, our confidence in securing gigawatt scale power resources the end of this year remains unchanged. We will continue to working with the relevant parties and provide updates to the market when we are able to share more specific information.

Yes. And about the co-location or the GPU cloud rental. I think it's still too early to decide whether we will focus on co-location or GPU cloud rental. There's many steps between -- even between we secure the power resources and providing computing services. So for a project of meaningful scale, we normally need different parties to work together. Many different parties, including capital providers, engineering contractors, and other resources partners, which parts Canaan will participate in and what type of partnership we will use will depend on the specific project.

Our approach is to use the capabilities and resources we already have, participate where we can add value and work with partners that bring complementary strengths. We do not assume that we need to manage the entire chain from the power to computing services by ourselves and we will not make investments beyond our capabilities or capital capacity to just to build end-to-end model. So at this stage, our main focus is still on advancing long-term cost-advanced power resources. Yes, as the projects become clear, we will evaluate the most suitable business model and level for participation. So for now, we do not think it's appropriate to make a firm choice between these 2 routes. Thank you.

We appreciate the color there. Can you kind of formally remind us, how is Canaan strategically positioned versus peers to secure and develop power for HPC? And it seems like you guys will likely go more the development partner route, but should we expect any upcoming hires to build a data center development team internally?

Yes, I think our advantage first comes from the practical experience we have built in Bitcoin mining over the past several years. Mining requires us to work with many types of power providers and to evaluate sites across different regions. The U.S. is a very large market and our local team has spent years visiting mining sites and advancing projects. Through this work, we have developed a better understanding of how power infrastructure works, the power conditions in different regions, and how to work with our local partners. We have also built relevant resources and relationships.

So since late 2024, our strategy has gradually shifted from mainly adopting partnerships to increasing the amount of assets we own directly. We want to have more control over long-term power resources and the site operations. This direction has based on the needs of our own business and started before the market's broader focus on AI and HPC. So as the market attention has moved more towards AI and HPC, we see strong continuity with our existing direction of funding long-term stable cost-advanced and scalable power resources. Our past experience, resources, and local relationships can continue to support us and give us a good starting point that we expand into this area.

Of course, HPC has higher infrastructure requirements, so each project still needs to be evaluated and developed based on its actual use cases. So change in the markets also affect competition for power resources and the project economics. We will also adjust our evaluation and the execution approach as needed. We'll continue to focus on long-term costs and the investment returns. Thank you.

The next questions will come from the line of Kevin Cassidy from Rosenblatt Securities.

With your internal mining fleet improving energy efficiency, can you tell us a percentage in efficiency increase you should see by next year?

I think you're asking about the mining machines efficiency, right?

Yes.

I think for the A15, older machines, we are at like -- I think it's 15.8 to like 17.8 joules per terahash. And the next level is our A16. We have like a 12.8 joules per terahash to 16.8. It's already our cost effective models. So there are a huge advantage for the for these kind of machines. Yes, for our mining fleet, we have plans to deploy at least a part of the 60 machines to our mining fleets next year. So currently our all-in power cost was about $0.043 per kilowatt hour. And still, we have a positive cash contribution.

For the next year, we -- if you have reviewed the total network hash rate, I think it's declining for a few quarters already. So it's quite easy to calculate the income for mining projects. And we assume we will upgrade a significant percent of our machines from the old one to the more advanced and also cost-effective models next year. Yeah, I hope I answered your question. Thank you.

Yes, that helps. Yes. And then just on the power pipeline, are you still pursuing, looking for stranded power around the world? I guess if you could talk more about that development?

I think, Kevin, we are actively -- we're always actively exploring and evaluating stranded power, trying to seek for opportunities. But of course, from different perspectives, it takes quite a long time to evaluate the opportunities. And it's not easy to find the partners like Cipher Mining, Cipher Digital or WindHQ. Those partners in Project ABC, they supported us. They do good deals with us in a transparent way, in a fair way. It's not easy every time we can find this kind of good partners.

So what we do is we mainly look at the long-term power cost, site stability, expansion potential, and grid access, and the local policies and the regulatory conditions. We will also try to evaluate the best use for each site, including mining, and where it is suitable, potential HPC applications, together with the additional equipment and the investment required. So currently, we stay open-minded while maintaining our capital discipline. No matter through acquisition, through JV or joint development, we will only move forward when the risk and the return are attractive and the project can improve the quality of our long-term assets and operating flexibility.

So I should say, currently, we haven't yet deal another one, just like Project ABC, but we do have some projects under discussion. We continue the strategy of expanding in North America, especially U.S. I think this strategy remains same. We will do more, but we do it cautiously, slowly with all kinds of evaluations done, but not immediately jump to many deals. And I think the capital allocation is also in a kind of very cautious way. Kevin, I don't know if I answered your question.

No, that's very clear.

The next questions will come from the line of Ben Sommers from BTIG.

I appreciate the commentary on the current state of the product market. So with Bitcoin prices up around 20% over the past month, just curious if you've seen any positive implications for the global demand for mining machines?

Yes, global demand, to be frank, frankly speaking, mining rig demand outside the U.S. is also very, very weak at the moment. Weaker mining economics together with the policy developments and the geopolitical conditions in some regions have affected customers' willingness and ability to invest. We have not yet seen a meaningful improvement in overall demand.

One specific area of progress we can share is ESG related applications, particularly compute to heat. Our Nordic heating projects already have 2 megawatts of equipment in operation. And customer ordered another 6 megawatts in March this year. These projects combined with mining with real heating demand, so the same energy input produce both computing power and the usable heat. We will continue to develop this kind of applications, but the market is still at an early age and small scale. So this does not yet indicate that broader recovery in demand outside the U.S. Thank you.

Super helpful. And then just wanted to touch a little bit on the Avalon Home Series. It seems like you guys are making some positive developments here, but can you just talk a little bit more about what you're seeing for those machines? And I think you guys mentioned some product developments there that could help potential winter seasonality. So if you could just talk a little bit more about that.

Yes, Avalon Home is a consumer product line that we have committed to developing. In Q2, it generated approximately $1 million in revenue. So it is still relatively small. I think the long term value lies on serving household users by combining computing and heating, which can broaden our customer base and revenue resources over time. After the heating season ended, sales of our home products came down. And at first I was a little disappointed, but then we realized that this may actually show that consumers are really using these products to heating -- other heating devices. After all, it's not very easy to sell heaters in the summer. So it reminds us that we need to understand the use cases and the seasonality of this kind of business as a true consumer product business.

So we began preparing new products for this year, heating season in Q2 and in Q3. Currently we are working on preparations for mass production, to launch new products and updates during the Christmas shopping season. So most of our key product line will receive updates and we may also introduce additional products. So at the same time, we will focus on sales channels, after sales service, and the user community, continue to improve our noise levels, easy for use and many different kind of stuff related to the consumer product.

So yes, we hope Avalon Home will grow its revenue and the contribution to the company and become a consumer business with lasting value. So I think it's still too early to give a specific revenue mix target. Its shares of total revenue will also depend on heating seasonality and the changing -- changes in our other business. But for now, the priority, to get the product experience and the business, fundamentals right, so growth is supported by real household demand. I hope I answered your question. Thank you.

[Operator Instructions] Our next question comes from the line on Nick Giles of B. Riley Securities.

This is Bill Chen on for Nick Giles. First I want to congrats on the maintaining positive cash contribution from mining in this pressured pricing environment. I guess on the share repurchase program, with approximately $7.4 million deployed against the, I think the total of $30 million authorization. What's your expectation for the pace of repurchases between now and the program's expiration in mid-December?

And I guess any color on how you're balancing the capital allocation strategy against liquidity preservation would be appreciated.

I think you asked a very forward-looking question about the future stock repurchase. Currently, I think it's difficult to answer because in September we will see a lot of activities in the U.S., especially on September 15, the Senate will start to vote for CLARITY Act. We don't know if it's passed, then what will happen to our industry that could make the Bitcoin price jump very high. And also our share price could fluctuate together with the Bitcoin price. So it's depending on what kind of a share price we were traded in the market and it depends on how we consider it's better to do some allocation to do stock repurchase. So it's difficult to predict the pace.

But to be very honest, as CEO and I myself, we have already purchased to get more shares of the company. We do have strong confidence that in the second half our share price should come back again with some good trajectory. And even we are currently under the risk restriction of Nasdaq compliance requirement. So we're better to regain compliance very soon. So that's why we consider stock repurchase is something we need to put a lot of efforts trying to make sure we do, and we allocate funds to do. And that's why we discussed with our board, and we got the authorization from them, and say management can start to sell the digital assets to generate cash and to repurchase.

I think our board understands the shareholder value quite well. They support us to do this and we are together with our shareholders, and we are the shareholders. So we will do everything we can to do but not within September, the particular month, but through the whole second half year or even future. So I think that's our commitment. That's my answer. Thank you, Bill.

I guess maybe one more, if I could, specifically on the Project ABC, we'd love to see all the progress. Given now it has reached 4.85 exahash, what's the next milestone or target for the project? If there's roughly -- a rough timeline you can provide.

I think we have been actively exploring and evaluating this type of opportunity. But the -- if the power cost -- we need the power cost stability, scalability and the grid access are suitable. We can use mining as a flexible load to push the resources to work relatively quickly. So if the site conditions allow, we can also -- we have the flexibility for other computing load in the future. So yes, our approach is stay open-minded and while maintaining capital discipline, whether through acquisition, JV, or joint development, we will only move forward when the risk and returns are attractive and the project can improve the quality of our long-term assets and operating flexibility.

We have -- we do have a very good machine even maybe still at the wafer level, but we can manufacture machines in relatively, short time to -- and deploy them into new sites if we find somewhere it can fulfill our requirements.

So -- but currently, I think it's still at some stage of a bear market, it's not a bull market, right? So investment is very, very -- we will have very high cautious in investment and consume our cash flow to do this kind of project. So currently, we are still evaluating the resources and wait for the best timing to get the scale up of our mining fleets include upgrade the machines. Thank you. I hope I answered your question.

We will now take the last questions. Our next question comes from Michael Donovan of Compass Point.

This is Ian Generous dialing in for Michael Donovan. And firstly, on potential JV partners, what type of companies are you engaging with today? And are prospective AI customers already involved in creating these sites? Additionally, what requirements are they emphasizing? And what would you expect to secure -- would you expect to secure customer commitment before making a significant development investment?

Yes, we are open to using JVs or development partners, from securing power resources to building infrastructure, and finally providing computing services. Different stages require different capabilities and capital. So for each project, we will choose a structure that allow all parties to contribute their strengths. Canaan has all always valued working with partners. Our cluster of joint mining projects, as well as Project ABC with WindHQ, have given us useful experience. At Project ABC, we combine our mining machine and the technology capabilities with our partners' site operations and power management experience. Together we are upgrading our fleet and improve the project's cash returns. This experience can also support future partnerships.

I think since late 2024, we have gradually increased the amount of assets we own directly, but this does not mean that we need to do everything by ourselves. So we want to have more control over our core assets and the long-term resources. While still working with partners that can bring complementary competitive expertise and the resources. So for future projects, we will not limit ourselves to advance to a JV, current development or any other structure is accessible. So we will focus on how investment and the responsibilities are shared.

So whether the risk Canaan takes, it matches by the return we can earn. Our final decision will always be based on the long-term shareholder's interest and value per share. So -- but currently, we do not have any specific partnership agreements that we can disclose. Thank you.

That's great color. And then lastly, if I may, more strategically, how should we think about the allocation of capital and management resources between pursuing the AI-powered land opportunity and then the core Bitcoin mining business?

I think for -- currently, we have already invested in the Bitcoin mining machines' core parts like the wafers. We already have them in our inventory or our partners' inventory. So when we have the clear opportunities to deploy the mining machines to have a good cash returns in the future, then we will do it.

For the data center, I just mentioned, we are very open to do any kind of partnerships with partners -- with third parties, with any kind of companies can contribute their benefits to the projects. We are not, to be clear, we are not going to do the end-to-end projects by ourselves to -- because I think to develop a meaningful scale of AI HPC data center needs a lot of capitals, which may be out of our capital capabilities. So yes, I think overall, our first priority is to -- is the interest of our shareholders, what kind of benefits we can get for every share is the most important part. Thank you.

I would like to add some color in this. As NG mentioned, we do have the inventory of mining machines now, so when we do some new mining cooperations, this can be injected to the project as well as a kind of CapEx investment and without spending any cash. So for the new HPC data center or this kind of power infrastructure thing, sometimes we have to use some cash as a deposit to secure some opportunities. That's something we do. We spend some of the money, we allocate some of the money as a deposit to secure some of the opportunities. That's what we are doing now.

I think from long term perspective, we will continue to balance between the different business and we will utilize our technology in mining machine side and mining -- do mining operations. And in the other side I think step by step we start from initial capital in a kind of a smaller scale but in future with pipelines, with different projects on air, and we will see step by step we accumulate more and more assets. And with this assets, we can use this assets as collateral to leverage to get more funds to sponsor the business in the data center part. I think that's something in future step by step we will reach there.

But first of all, we should try to secure gigawatt level power and with this on hand and then we can start to talk about the next step. As NG mentioned, we do have the long-term ambition, but in short term we will not just immediately try to do everything in the ecosystem. We will do step by step and gradually grow ourselves. Thank you.

I would now like to turn the call back over to the company for any closing remarks.

Thank you everyone for joining us today. If you have any further questions, please feel free to reach out to us directly or through the contact information on our website. Thanks.

That concludes today's conference. Thank you for your participation. You may now disconnect your lines.

Ladies and gentlemen, thank you for standing by, and welcome to Canaan's Inc.'s First Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note that this event is being recorded. Now I'd like to hand the conference over to your speaker today, Gwyn Lauber, Investor Relations for the company. Please go ahead, Gwyn.

Thank you, operator. Hello, everyone, and welcome to our earnings conference call. Joining us today are our Chairman and CEO, Nangeng Zhang; and our CFO, James Jin Cheng, Leo Wang, Vice President of Capital Markets and Corporate Development; and Xi Zhang, Senior IR Manager, will also be available during the question-and-answer session. Our CEO will start the call by providing an overview of the company and performance highlights for the quarter. Our CFO will then provide details on the company's operating and financial results for the period before we open up the call for your questions.

Before we begin, I would like to refer you to our safe harbor statement in our earnings press release. Today's call will include forward-looking statements. These statements include, but are not limited to, our outlook for the company and statements that estimate or project future operating results and the performance of the company. These statements speak only as of today, and the company assumes no obligation to revise any forward-looking statements that may be made in today's press release, call or webcast, except as required by law.

These statements do not guarantee future performance and are subject to risks, uncertainties and assumptions. Please refer to the press release and the risk factors and documents we file with the Securities and Exchange Commission, including our most recent annual report on Form 20-F for information on risks, uncertainties and assumptions that may cause actual results to differ materially from those set forth in such statements.

In addition, during today's call, we will discuss both GAAP financial measures and certain non-GAAP financial measures, which we believe are useful as supplemental measures of the company's performance. These non-GAAP measures should be considered in addition to and not as a substitute for or in isolation from GAAP results. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP results, in our earnings press release, which is posted on the company's website. With that, I will now turn the call over to our Chairman and CEO, Nangeng Zhang. NG, please go ahead.

Thank you, Gwyn. Hello, everyone. This is NG, CEO of Canaan. Thank you for joining our earnings conference call today. James, our CFO, and I am here at our Singapore headquarters to share our financial results and recent business updates for the first quarter of 2026. Q1 2026 was a very challenging quarter. Bitcoin prices dropped sharply from the height at the beginning of the year. and half price fell to very low levels. As a result, finance around the world became much more cautious with their investments.

After entering in the second quarter, the market saw some recovery, but the recovery has still been limited. At the same time, uncertainties related to the Middle East situation, energy prices, global equity and the policies continue to keep the industry contract environment. For us, our company going through a transition period, this kind of environment created a lot of pressure.

But today, I want to focus less on the gates we faced and more on what we did during the difficult times. I believe investors want to see whether we have strong execution, deplant operations and ability to navigate through market cycles.

In the first quarter, we completed several concrete tasks. First, we completed the final stage of production, delivery and revenue recognition for our large order from a leading North American customer while entering the market downturn with a relatively light inventory position.

Second, we continued expanding our mining business, which still generates positive cash contribution even under extremely low hard price conditions, while further increasing our digital asset treasury. Third, we completed the acquisition of ABC projects through a share exchange transaction, obtaining a 49% equity interest in 3 energies and operating mine site with low power cost invest taxes.

Fourth, we continue to advance in the R&D of 16 series and our next-generation products to prepare for the next mining equipment update cycle. Fifth, we continue shifting the company's strategic focus from a pure mining machine business towards energy computing infrastructure. Taken together, these actions show that during a difficult market environment, we did not simply wait for the market to recover instead, we actively strengthened our survivability, improve our asset quality and expanding our long-term strategic options.

In the quarter, we generated total revenues of USD 62.7 million, in line with our previous guidance range. As of the end of the quarter, we had 1,808 bitcoing and 30,952 Ethereum and our digital asset treasury reached another record high. In mining machine, industry demand was clearly under pressure in the first quarter. We sold 4.1 exahash per second of computing power is an average selling price of about 10.5 per terahash, generating USD 42.9 million in revenue.

Many customers delayed fortress due to low cash price and high market uncertainty and the market pricing also came under pressure. In this environment, we did not pursue short-term sale growth through aggressive inventory buildup or lower quality orders. Instead, we raised higher priority on inventory control, cash flow management and all quality. This also reflects the operating discipline.

We have emphasized over the past several quarters. In Q4 of 2025, we captured the market window and secure the large North American order with all the deliveries completed. In the first quarter of this year, we completed the final stage of execution through this -- the successful completion of this project. We further strengthened our brand reputation and customer base in the North American market. mining machine business may not be the hottest story in the capital market today, but it remains the formation of cane.

As long as the is network continues to operate and low-cost power resources continue to exist around the world. Manas will continue to need machines that are more efficient, more reliable and easy to deploy. Our job is to run the mining machine business with stronger discipline and stay closer to the real needs of our customers. In the fourth quarter, we continued advancing customized products and system-level solutions.

Recently, we expanded our collaboration with Tier by providing customized high-density tax for models for its next-generation emerging mining and unfit systems. This type of partnership shows that leading customers are shifting from pursuing single sand miners to see interpretive systems that are modular metal, gradable and adaptable to different operating scenarios.

For Ken, this is exactly where our long-term strength in ASIC demand system engineering, supply chain management and the global progress delivery can create value. In addition, as we announced earlier today, we sold approximately 8 megawatts of hydro crude equipment to Nordic citing service provider to produce high-grade hot water for the strip heating systems.

Projects like this show that mining machines are gradually expanding beyond pure mining use cases into broader energy utilization scenarios. The combination of computing power, heat recovery and the local energy infrastructure is also an area we will continue to explore going forward. In the consumer and SMB market, the main focus of having a home series in the first half year, has been channel expansion, customer reach and service system development.

Since the beginning of this year, Aram products have entered platforms, including Best by Canada's online channel and Amazon. The consumer market is very different from the industrial mining machine market. Customers are not only about half rate, but also about noise level, stability product design, easy for installation and after-sales stories. We are currently working on the product upgrades for the -- for several or home models and hope to lock them in the second half of this year.

We hope that better products, stronger sales channels and the year-end shopping season, together can help this business line contribute to higher core revenue. Now we move to our mining business. The mining environment in the first quarter was very challenging. In January, during the winter storm across North America.

We materially power down and hotel operations in the certain regions to prioritize electricity supply to local residents and the power grid. We want to be trusted and responsible partner with labor computing load for the grid rather than adding additional pressure during a period of great stress. More importantly, even under a low has price environment, our mining business continued to show strong competitiveness.

During the quarter, we generated 257 bitcoins in total and recognized $19.12 million in mining revenue. From a cash operating perspective, this business continued to contribute positive liquidity inflow to the company. By the end of the quarter, our global installed part reached 15%, up 66% year-over-year and 11% quarter-over-quarter.

Our operating base continue to expand. While our power and hosting costs remained relatively competitive. In April, our non-JV installed hash rate remained around 11 hash per second with an average all-in power cost of about $0.044 per pear. At the same time, ABC JV program also add 4.82 ex hash per second installed cap rate, and 120 megawatts of installed power capacity.

I believe these numbers show 1 important thing. The mining business still has value even during the low point of the cycle. It helped us to accumulate BDC and help us better understand the real operational needs and the pain points of minus. More importantly, it helped us to build real power consumption and operational capabilities. We continue to advancing energy and computing infrastructure future.

The most important development this quarter was the ABC project in late February, we acquired EBITDA interest in par and Simonton products in U.S., Texas from Cyber through a share exchange transaction together with 6,840 Avalon A15 pro mining machines. The biggest advantage of the ABC products is highly competitive power cost, which is below USD 0.03 per kilowatt hour.

Because of this cost advantage, the products maintained a strong profitability and a high uptime, even during a period of Bitcoin price volatility, which significantly improved over time. We have also been working closely with our partner, with HQ to steadily upgrade the mining fleet at a site.

At the end of April, the projects installed hash rate increased from about 4.4 exhash per second to 4.82 exhash per second. In addition, the JV program has potential for future power load expansion, and we are currently evaluating related opportunities. Overall, the AT products operate and their hybrid mining -- power model, combining wind power and a electricity is total installed capacity of 120 megawatts and power cost below us and U.S. do per kilowatt hour.

The projects currently have an in-store hash rate for approximately 4.82 per exhash second. We have maintained a strong long-term relationship with Cyber over the past years. The commission of the ABC Project transaction also reflects our ability to take over high-quality assets released during the doing CIOs business transaction. based on our longstanding cooperation, we believe high-quality power resources and the infrastructure accountabilities will become increasingly important competitive advanced advantages in the industry over the long term.

The completion of ABC projects only future strengthened our footprint in North America energy and infrastructure, but also response represent is an important step in advancing our long-term Energy+ computing infrastructure strategy. Following the transaction, Cyber also become -- became an important shareholder of Canaan, lining the foundation for deeper cooperation between the 2 parties in the future.

This product has very important meaning for us. First, these are low-cost power assets that are already energized already operating and already generating computing power in today's North American market assets with real operations are much more valuable than pipeline opportunities on paper. Second, program is concrete result of our energy strategy. In future, strengthens our access to low-cost power resources, mining operation experience, and the local partnership networks in the United States.

Third, it also provides us with stronger infrastructure commodities and greater strategic flexibilities after we continue to explore future AI and ATC opportunities. Regarding our energy pipeline, we have indeed made some meaningful and encouraging progress. However, as a responsible public company, we do not believe these developments have yet reached the closure milestones required for us to provide more specific details publicly.

So at this stage, I cannot share too much additional information. but I can reaffirm our site view high-quality power resources will become one of the most important barriers in future computing infrastructure. Our goal is to secure power infrastructure that is controllable, developable and available in regions that are compliant close to major customer markets large-in-scale capable for long-term grid connection and expandable over time.

The United States remains one of our most important markets. We hope that in the future, once project conditions become more mature and discoder requires are met. We will be able to provide the market with more concrete and substantial updates.

Now let me talk about our R&D and products. In the fourth quarter of last year, we officially launched the Avalon A16 XP. It delivers up to 300 terahash per second per machine with energy efficiency as well as 4.8 terahash per tons. During the first quarter, some customers received simple units and bigger testing.

Based on the feedback we have received so far, the A16 series has performed well in castrate stability, energy efficiency, noise control and deployment capability. We have also seen growing attention from the mining community and the third-party reviewers towards the A16 Series, which has been very encouraging for our team.

A16 Series will become the core of our future industrial mining machine product line. It is not only a performance upgrade but also represents our overall mobilities in system airing some more design formware, reliability and the cost control. Advanced semiconductor process are becoming increasingly expected and simply pursuing the voice oer Terahash does not always deliver the best returns on investment for customers.

We pay more attention to the products for life cycle economics for the -- for customers, including machine pricing, power costs. operational stability, maintenance costs, delivery certainty and risk deal value. because we have secured a part of our key product capacity early and have maintained the long-term cooperation with our foundry and supply chain partners. We are still able to move forward with A16 series mass production and the future product introductions in a more stable and cost-controlled way.

Even under the current environment where AI-related demand is competing for advanced semiconductor capacity. For our mining machine delivery, we leverage manufacturing capacity across Malaysia, the United States and Mainland China. This allow us to remain compliant while responding more flexibility to changes in the global trade environment and tariff policies. In -- particularly during the delivery of our large North American order, our manufacturing quality control and logistics teams worked closely together and successfully handled the pressure from concentrated shipments and tight delivery schedules demonstrating the resilience and execution capabilities of our supply chain team.

In addition, assembly capacity for our Avalon home series has also been expanded to our Malaysia facility. Beyond the current A16 series, the R&D of our next-generation products has also entered the final stage. And some projects have recently completed tape-out for technical validation. After we complete product testing and the real operation -- operating conditions, we will close more detailed technical specifications to the market.

We are confident in the performance improvements of our next-generation products. And we will continue to follow our principle. Customers are not just buying subscriptions for systems that can operate stably by over the long term and generate stable and reliable returns. Today, I also want to talk more systematically about our AI and HPC strategy. As AI computing demand continues to grow rapidly, power resources sale centers and the computing infrastructure are becoming increasingly important.

The market is also paying close attention to mining companies moving into AI and HPC. We understand this interest many companies are talking about AI and HPC, but I hope investors will see can support -- we will be steadier and more practical. For AI HPC, our long-term strategy has 2 major pillars. The first pillar is energy.

The compilation of the ABC product shows that we have already real progress in energy and infrastructure. These are not conceptual pipeline projects, but assets that are already energized, already operating and already generating computing power and cash flow. At the same time, we are also advancing large scale and a more controllable power resource development.

Our goal is to gradually build the power infrastructure abilities that are financeable, developable and available by the company in compliant regions that are close to key markets and have long-term expansion potential. Energy infrastructure products, orally take a long time. The process from permitting land acquisition and a great connection to construction and operational or request time.

Therefore, we will not make over aggressive promises based on the short-term market segment. But once these projects are completed step by step, we believe that will become one of the most important long-term mats for our future computing infrastructure strategy. The second pillar is our computing systems. Over the past decade, Canaan has been deeply involved in ASIC design, mining machine development, large-scale delivery and real-world mining operations.

We are familiar with turning high-density computing equipment into products that are standardized, modernized mass productible remotely manageable and easy to deploy at scale. We believe broader AI and HPC infrastructure in the future will increasingly require these same capabilities. Our thinking is how to gradually make AI computing systems, which may become the largest source of new computing demand in the future, more like mining machines with a scalable divestment, standardized operations and clear economic models.

This growth process will not have overnight, but we believe the direction is becoming increasingly clear. I don't believe BTC mining and AI HPC are completely separate businesses. For Kena, blockchain computing is a proven workflow today that already generates cash contribution and help us validate power assets and operational capabilities. AI and HPC represents future computing demand with larger scale and higher into structure standards.

The cost transformation is already fully underway internally. Our power infrastructure planning is being designed for long-term and higher-density computing demand. while our chip and system abilities are also gradually expanding towards broader computing platforms. But at this stage, we performed to spend less time talking about concepts and more time building real asset products and engineering come abilities.

What we want to do is gradually expect our existing strength in mining, energy, chip and system engineering into broader and the broad chain computing infrastructure. We believe the right approach is to first build a strong condition in power sources and operations and then gradually integrate new types of computing system when the timing is right. In this way, the company can continue benefiting from the cash contribution and the flexible low value of BTC mining while also creating long-term opportunities in AI HPC and in variable and settlement enabled digital economic network in the future.

This past fits well with the foundation we have built over the years. We believe we already know where the industry is heading. In the future, Scott resources will gradually shift from GPUs themselves to compliant low-cost power, this patch for loads to mass specific texture-based AI computing systems and long-term organizational mobilities. The hardest part is finding the right path from where we are today to the future. what now, including the ABC products, direct power pipeline development, chip demand, system engineering and organizational efficiency improves. This essentially building the foundation for that part.

Finally, I want to talk about our organization and cost structure. Since the fourth quarter of last year, we have continued optimizing our organization in Q1 of 2020. The result of these efforts already started to appear in our operation -- operating expenses. Going forward, we will continue to focus on our resources on core products, key projects and areas that can build long-term competitive advantages.

At the same time, we are also introducing AI tools more deeply across the company, including R&D collaboration, coding and testing, supply chain planning, financial analysis customer support and operational management. My view on AI is very practical. AI is not only a market that we may serve in the future, but also a tool that helps us to improve our on organizational efficiency today.

We want to achieve more go deeper and deliver higher-quality work with a leaner organization. Going forward, we will continue management expenses with stronger discipline while improving business responsiveness and futuring the efficiency. We believe these are critical probabilities for the company to successfully navigate industry cycles.

In March this year, James and I also purchased companies ADS in the open market using our personal funds. The amount itself is not a key point. What matters is that management stands on the same size as all shareholders. to this market environment is in deeply challenging, but we remain confident in the covenants long-term direction and our ability to execute.

Looking ahead to the second quarter, we remain cautious. Although Bitcoin price and cash price have recovered somewhat from the lowest to the first quarter -- in the first quarter. miners globally are still taking a conservative approach to the investment. In addition, energy prices and the geographic -- and the geopolitical uncertainties may continue to affect customer decisions. Therefore, we expect total revenues for the second quarter of 2026 to be between USD 35 million and USD 45 million. This outlook is based on the card market and operating conditions, and actual results may differ due to changes in the market conditions, policies, compliances and customer demand.

In the short term, China is still going through a difficult transition period. We do not award this reality, but I also want to make it clear that the company is not standing still. We are reducing inventory, controlling costs, advancing new products, expanding sales channels, strengthening money operations, securing low-cost power resources advancing our U.S. power infrastructure pipeline and exploring long-term opportunities in AI and the PC computing systems.

The industry cycle will continue to fluctuate and market segment will continue to change. But we can control, but what we can control are our situation, this plan cost structure products asset quality and long-term direction. As long as we continue improving in these areas, we believe Ken will become stronger in the next cycle. My -- that conclusion is in my remarks. Thank you again for your continuous support. I will now turn the call over to our CFO, James, to discuss our financial results in more detail. Go ahead, James.

Thank you, NG, and good day, everyone. This is James speaking in our sincere quarters. As NG highlighted, the first quarter of 2026 was defined by significant volatility global liquidity was tightening and the Middle East geopolitical conflicts were escalated during the quarter, together with energy prices and regulation bonds. .

Bitcoin entered the year trading near $95,000 level in the middle of January before experiencing a quick decline and bottoming at approximately $66,000 in early March. This fluctuation of bitcoin price directly impacted industry-wide mining economics and harsh price, forcing a cautious wait-and-see posture across the institutional sector. Despite these headwinds, our operational performance demonstrates our resilience of going through industry cycles.

We successfully delivered the total revenue within our guided range we increased the revenue from our North American sales, and we strengthened our mining operations by securing a 49% membership interest in 3 high-quality mining projects. At the same time, we also derisked our inventory position through the accrued write-down, continuing to optimize our operational efficiency by continuous expense control.

Collectively, these actions allow us to remain lean and agile, positioning us to navigate ongoing market volatility and prepare to capture future high-margin opportunities as the cycle eventually turns.

Moving on to our financial performance. We delivered total revenue of $63 million in the first quarter, which was within our guided range. Our product revenue contributed $43 million to the top line. This represents a sequential decline because of the market environment change from Q4 to Q1. North American customers contributed over 80% of total product sales, which increased from 75% in the last quarter.

During the quarter, we sold 4.1 exahash per second of computing power at an average price of $10.50 per terahash per second. Within product revenue, our Avalon home service generated $2.7 million as we continue to invest in channel development for this segment.

Our mining business generated $19 million in revenue. While this figure reflects the lower bitcoin prices during the quarter, the business continues to serve as a consistent engine for our asset accumulation. By maintaining our mining activities throughout the market cycle, we are effectively strengthening our digital assets churn and building long-term value for our shareholders.

Turning to our mining operations. We concluded the first quarter with a total installed hash rate of 11 exahash per second up 11% from Q4 last year, and this indicates a year-on-year growth of 66%. The growth is mainly driven by our development in North America. In Q1 we have expected our installed mining hash rate in North America 7.7x of Q1 '25. North America's occupation increased from 11.5% to 53.6% in our quarterly global cash rate. This is fully aligned with our set strategy of continuously investing in mining operations in North America.

This has not even included another 4.4 exahash install hash rate in the JV Corporation acquired from Cyber Digital in February as we own 49% of the interest. As part of our old strategy, we ended the quarter with 1,808 Bitcoin and 3,952 Ethereum on our balance sheet. With the production of 257 Bitcoin points in this quarter, the total market value of our Bitcoin holdings stood at $121 million as of March 31, 2026. This growing reserve serves as a key pillar of our balance sheet strength.

With the recent price recovery towards $77,000 level, the market value of Bitcoin holdings has increased to nearly $140 million. I would like to address our gross loss of $23 million this quarter. which was entirely driven by a $25 million noncash inventory write-down entering product costs. Excluding this impact, our adjusted gross profit was approximately $1 million, representing a breakeven adjusted gross margin. This accounting treatment was due to continuous pricing pressure and aligned our inventory cost structure with the market environment.

Moving to our financial efficiency. Total operating expenses for the first quarter were $31 million an 11% reduction from last quarter and an 18% reduction from $38 million in the same period last year. This improvement reflects our efforts to streamline our organization across all functions and our set discipline to control expenses.

Specifically, research and development expenses were $15 million, down 19% year-over-year. Selling expenses were lower to $1 million to 59% year-over-year, and the general and administrative expenses were reduced to $15 million, down 11% year-over-year. These expense reductions are the direct result of our ongoing commitment to eliminating nonessential spending and focusing our resources on core strategic priorities by all methods, we have built a leaner and more cycle-resilient organization.

Now I would like to provide more details on the noncash items that impacted our bottom line results. This quarter, we recorded a $41 million fair value loss on our digital asset holdings. This reflects the significant bitcoin price fluctuation, which declined from approximately $87,000 by the year end of 2025 to $67,000 by the end of the first quarter of 2026. I want to emphasize that this is a market-to-market accounting adjustment and does not represent realized cash loss as we continue to hold these assets on our balance sheet.

Consistent with industry practice, these fair value changes are included in our adjusted EBITDA calculation. Consequently, our adjusted EBITDA loss for the quarter was $76 million, reflecting the combined impact of the operational environment and the period and revaluation of our digital assets.

Regarding our liquidity, we ended the first quarter with a cash balance of $43 million. On the cash outflow side, we allocated $57 million during the quarter for manufacturing and operations to support our global supply chain, $6 million in wafer procurement payments to secure future production capacity and $2 million for share repurchases. This strategic increase were partially offset by in total cash inflows, which mainly consists of sales collection, ADR relate and value-added tax refund.

The sequential increase in our cash balance from $81 million last quarter was primarily driven by collection timing and our planned capital outlays. This position has already been changed as we have collected $42 million in cash receivables from minor sales in April. This post quarter cash recovery demonstrates that our liquidity remains healthy and provides a solid foundation to navigate near-term market conditions while remaining prepared to capture future opportunities.

I would also like to provide more details on the project ABC acquisition that closed in late February. This transaction was structured as a share for asset exchange, where we issued approximately 54 million ADS with a total fair value of $25 million. This consideration was allocated between 2 key assets, $14 million as equity investment for 49% of stake in the JV comprising Alberta and Chief Mountain, and $11 million for the 6,840 A15 Pro mining units now recognized as part of our PPE, property, plant and equipment.

By utilizing an entirely share-based structure, we secured 100 megawatts of high-quality North American power infrastructure with electricity costs below $0.03 per kilowatt hour without cash outlay. This approach allowed us to preserve our liquidity while onboarding cyber as a strategic shareholder. We view this project as a highly capital efficient deployment of our equity that significantly strengthens our North American footprint and cements our long-term partnership with Cyber.

We remain anchored in long-term strategy that prioritizes structural resilience and asset quality over short-term market fluctuations, while we maintain a cautious and disciplined stance for the upcoming quarters. The fundamental value of our linear cost structure and the risk the balance sheet will become increasingly evident as this industry cycle evolves.

By securing critical infrastructure and optimizing our manufacturing operations, we have built a platform that is prepared to capture the next wave of institutional growth. Moving forward, we will continue to safeguard our liquidity and leverage our technological edge to drive sustainable value.

Given the headwinds and uncertainties in Q2, we are taking a very prudent approach to provide our guidance. We estimate our revenue would be $35 million to $45 million. This concludes our prepared remarks. We will now open the floor for questions. Thank you.

Thank you. We will now begin the question-and-answer session. [Operator Instructions] Your first question comes from the line of Logan Hannan from Northern Capital Markets.

First, can you just help us educate us again on how Cana is strategically positioned to secure and develop power for HPC infrastructure and will you be making any upcoming hires or working with a development partner to make this transition.

Let me start with the resion because that is the most important part is we want to do this. We are building our resources in short term, I think mine is the best immediate load for the power. It is simple for us to deploy a flexible in the long term. these power sources and our partner network can become our entry point into AI HPC infrastructure. This is already underway.

On the last earnings call, I said a was transformation. Only 3 months has passed, we already have future progress. We believe we will continue to show progress. On the our chips, we have -- personally, I have long been looking for a way to turn large scale highly dedicated AI workloads into ASIC friendly workloads closer to the mining computation today.

For the end state, I think that direction is almost certain. For many years, we were searching for the right path. Now the path is that can truly use our Assistant it became -- we came in much clear. So in summary, for our question, the summary is very simple. We want to build around energy, computing infrastructure and specialized ASIC design.

Mining give us the starting load. AI HPC gives us the long-term opportunity. Yes. And about the partners, yes, I think the specific sites we always have their own design. But the development and the corporate with other partners is an important model for us. The value is not only about putting money and AI HPC in the same place. The bigger value is time-based load management when AI HPC is powered or when total power is limited, mining can release load.

When there is exist power, low pricing for AH PC demand is in a low use period, mining can ramp up and in some cases run at a higher performance. So the economic benefit is clear because mining machine is relatively low cost and the clean load. So more importantly, it has social value grades like stable, controllable load, and this model can help power assets the grid and computing customers to work together more efficiently. Yes. I hope I answered your question.

Yes, that was very helpful. Then one more. Is there any additional color you can provide into your pipeline maybe how many sites are in that gigawatt, what stage are these sites in? Are they under exclusivity, development, due diligence? Any color there and the current steps being made would be great.

I really want to say harm, but we sit down with our compliance advisers and agree that it's better to announce details after some important commercial and legal documents are formally signed with the greater and our partners. There is -- certainly, there is still uncertainty as always, with large power projects. But our target, what we're working on is very clear. We want sites that can support both mining and AI HPC, and also have scale, have low power cost and give us enough control to lead the project ourselves.

So today, I will not disclose the site count, capacity by stage status, but I can say the work is moving fast, very quick, and our direction is unchanged. Thank you.

We will take our next question. Your next question comes from the line of Ben Sommers from BTIG.

I appreciate all the color on the ABC acquisition. I was just kind of curious, talking about the power pipeline. If you could talk about maybe if there are potential opportunities out there similar to that one, maybe acquire, whether it's a stake or a full project from a previous miner or someone that was mining bitcoin there and just kind of what you're seeing in the market for potential opportunities similar to that one.

Yes, I think the ABC acquisition has been very good for us. it gives us direct exposure to high-quality, low-cost power, invest assets. The electricity cost is below 0.3 per kilowatt hour. So the product remains resilient even when the required cash price volatile operationally, MVC has been 1 of our strongest side with very high uptime. We also been upgrading our miners with HQ by the end of April. The hash rate had to 4.82 exahash per second Also, the Arbor side also added grid connection, which improves our time to have wind plus grade structure. This product provides our low-cost power and execution matter. We will keep looking for similar assets and another app team opportunities.

And then my next question, just kind of given the current market conditions and the outlook you guys provided, how do you think about the future growth for the Avalon Home Series? And just kind of curious on what you're seeing from the demand profile for those rigs?

Yes. I think for the -- currently, I think we are under some pressure. But I see -- yes. This year, over a home was hit by some policy changes in some important markets. For example, in a strengthened restricts on mining products later last year. and other countries also had policy changes. This made us more aware that compliant and a stable market must be our main better field. So this year, our focus has been channel building and product investment.

In the second half, we have plans to launch several new products and several upgrade in these models upgrades. We are also building channels that match a more complete product line. We hope that can support higher revenue in the second half. Also, the gross margin still go on our product quality. I poised you to look at the community and the KOL reviews on YouTube, I think home mining product line, we believe we are far ahead. And yes, and also about the expansion in home serious Yes, we have a...

The question comes from Mark Palmer from Benchmark, StoneX.

Yes. You mentioned that have seen a pickup in the price of Bitcoin during the second quarter and that, that had caused some recovery in the Bitcoin mining equipment market, but it has been limited. If you could just provide some perspective on this. In the past, when we've seen significant drawdowns in the price of bitcoin, and then a recovery. To what extent does Bitcoin need to recover and then stay at higher levels before you begin to see an increase in demand for your products? .

Yes, I think we -- first, we're talking a little bit about the bitcoin price. Yes, I think two new highs last year have partly related to a weak U.S. dollar last year. It's not a very typical breakout cycle. So this year, from a technical perspective, on has shown some patents of falling to rate higher other than pulling back.

Yes. And I think currently, the -- in Q1, our is ASP is about $10.5 per terahhas currently because the demand supply imbalance and high price decrease. The ASP is really under pressure. But in my experience, if there are some index can -- I can help you to observe the recovery of the machines market. I think it's about the hash price. Currently, I think the hash price is about 30-some dollar per hash per day. So it's quite low.

When the hash price growth to like $40 to $45, then you will observe a significant market recovery for the money machines. And the market will want crazy when the hashhit the $55, And you can check the number on life side in real time. Yes. So I think in the month to month, hash price is climbing slowly, but steadily to close to $40, but it's dropped back in the last few weeks. So I think that the market still needs some more time to have a real recovery.

We will next question. And the question comes from Michael Donovan from Compass Point.

NG and James, can you discuss how much 815 series inventory remains in terms of Xs? How should we think about the time line for ramping A16 production.

Like the end of 2022 or early '23 at that time, our inventory was higher than this cycle. -- just because in Q4, we locked the giant order and we deliver in Q4 and early quarter 1. So actually, our inventory is not high. But for certain older generation machines, we still have some inventory, and we lowered down the price we try to clear that entry within quarter 2. I think that's the plan. It seems like the semiconductor sector is in fierce competition with AI-related applications.

They are occupying more and more wafer capacity, that's why for the second half, we still need to prepare for the wafers for our supply and make sure the demand can be covered. And we don't believe the market will continue to be very quiet, like the quarter 1 and quarter 2. And with all this all this news like clarity be approved by the banking committee, and we will see clarity go to the senator.

And eventually, we will see second half the bitcoin price has the possibility of going up. At that time, the machine demand could recover. So we're better prepared for that. So even currently, our inventory structure is not bad, it's quite light. And the cash flow is good, but still, we would like to prepare for second half.

Yes, I will add some thoughts on this. Our production preparation for A16 is ready. The tests are public, you can check it on YouTube, I mentioned. Also, the product performance is real and strong. So -- yes. And we -- another information is so for -- so even currently, we have low inventory. But whether -- if the market improves, we are in a good position to respond.

Appreciate that. What are you seeing in minor demand outside the U.S., which international markets are showing strongest today.

I think after U.S., we have some customers from Europe like we have corporate with hot water prime for their homes. I think we just announced about 8 megawatts orders from our European customers. And also, we have same likely customers from other countries that I want to mention. So -- but I think today, other regions still have opportunities. But near term, the U.S. is still the main focus because this is where we see that most important they have power mining fleets and AI HPC infrastructure. Yes, they remain the most important part for minerals.

Next question comes from Nick Giles from B. Riley Securities.

Yes. Thanks, operator. James. I was wondering if you could think to I was wondering if you could speak to the Teva relationship and just touch on maybe just a little bit more on the economics of that deal. And how could this expand? I believe that the agreement includes an option for additional volume, but just wanted to get a better sense for the overall revenue opportunity in this partnership.

Yes. I think we already cooperated with the technical line and their R&D departments -- for some really long time. So -- yes. I think the customized development service, like Titaro just mentioned, just need more than a standard machines. So we build do core R&D and build specialized customized modules using the different both to our mass production model. And also, we provide software and hardware system level solutions program and also they take the development by themselves for very significant part. Yes. So by this, I think we are quite close to have some mass production contracts. Yes. So it is what we are here today. I hope we can do some announcements after the last one.

Yes. So the other thing is we are doing open source -- we have already released the code, and we will continue to improve the quality of our open source work. So Tiger is a pioneer customer, but sure in the of the last for the third-party solutions, I think Ken is clearly 1 of the brands manufacturer. We provide open source code for software. And we also can sell chips. So we want to -- we provide the most easy way for our partners to build their own system. And I think it will be more and more friendly in the future.

Thank you so much, NG. I really appreciate the update this morning.

Thank you. As there are no further questions now, we would like to close the call. Thank you once again for joining today. If you have further questions, please feel free to reach the company through the contact information provided on its IR website.

Ladies and gentlemen, thank you for standing by, and welcome to Canaan Inc.'s Fourth Quarter 2025 Earnings Conference Call. [Operator Instructions] Please note that this event is being recorded.

Now I'd like to hand the conference over to your speaker today, Gwyn Lauber, Investor Relations for the company. Please go ahead, Gwyn.

Thank you, operator. Hello, everyone, and welcome to our earnings conference call. Joining us today are our Chairman and CEO, Nangeng Zhang; and our CFO, Jin James Cheng. Leo Wang, Vice President of Capital Markets and Corporate Development; and Xi Zhang, Senior IR Manager, will also be available during the question-and-answer session. Our CEO will start the call by providing an overview of the company and performance highlights for the quarter. Our CFO, will then provide details on the company's operating and financial results for the period before we open up the call for your questions.

Before we begin, I would like to refer you to our safe harbor statement in our earnings press release. Today's call will include forward-looking statements. These statements include, but are not limited to, our outlook for the company and statements that estimate or project future operating results and the performance of the company. These statements speak only as of today, and the company assumes no obligation to revise any forward-looking statements that may be made in today's press release, call or webcast, except as required by law. These statements do not guarantee future performance and are subject to risks, uncertainties and assumptions. Please refer to the press release and the risk factors and documents we file with the Securities and Exchange Commission, including our most recent annual report on Form 20-F for information on risks, uncertainties and assumptions that may cause actual results to differ materially from those set forth in such statements.

In addition, during today's call, we will discuss both GAAP financial measures and certain non-GAAP financial measures, which we believe are useful as supplemental measures of the company's performance. These non-GAAP measures should be considered in addition to and not as a substitute for or in isolation from GAAP results. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP results in our earnings press release, which is posted on the company's website.

With that, I will now turn the call over to our Chairman and CEO, Nangeng Zhang. Please go ahead.

Hello, everyone. This is NG CEO of Canaan. Welcome to our earnings call. Together with our CFO, James, we are calling from our Singapore headquarters to discuss our Q4 2025 business results and latest updates with you.

During the first quarter, it comprises experienced significant volatility. In early October, Bitcoin briefly broke about its previous all-time high, reaching approximately USD 126,000. It then fell below USD 100,000 in mid-November and dropped to below USD 90,000 by the end of December. At the same time, the wave of new hatchery that enters the pipeline during the price surge in the third quarter came online, pushing total network cash rate to a record high. This puts strong pressures on miners' profit margins. Facing this highly volatile market, we managed our sales pace well at the beginning of Q4. We secured a large order from a major customer in North America and efficiently mobilized resources to support production and the smooth delivery. At the same time, we steadily expanded our self mining operations and diversified mining partnerships, signing several private products. As a result, our total revenue for the fourth quarter reached USD 196 million, up 30.4% quarter-over-quarter and 121.1% year-over-year. This was our highest quarterly revenue in the past 3 years and exceeded the midpoint of our guidance range of USD 175 million to USD 205 million.

We achieved a major breakthrough in mining machine sales in this quarter, benefiting from our continued focus on the North American market. We secured a large scale order of more than 50,000 A15 pro models from a leading mining company. This milestone collaboration drove our strong sales performance and underscores markets growing recognition of our product performance and the delivery comply. To ensure high-quality delivery, our supply chain, production and pricing teams worked closely together to ensure smooth and high-quality execution. This resulted in an all-time high of 14.6 exahash per second in computing power sold during the quarter, up 45.7% quarter-over-quarter and 60.9% year-over-year. While average selling price declined slightly due to volume discounts for large scale of orders, the surge in sales volume drove product revenue to USD 165 million, up 39.1% quarter-over-quarter and 124.5% year-over-year. This marks our highest single quarter revenue in the past 13 quarters.

We will continue creating value for customers through product upgrades and customized services to deepen our partnerships with global customers. And although the company focused most of its resources on rig sales in Q4. Our sales mine operations continue to progress steadily and the principle of resources alignment and the efficiency first. In Q4, while we steadily expand and optimize global development at the end of the fourth quarter total installed has rate increased 8.6% quarter-over-quarter to 9.91 exahash per second, of which 7.7 exahash per second was energized. We mined approximately 300 Bitcoins during the quarter, further contributing to our currency reserves. At the end of 2025, our assets holding were 1,750 Bitcoins and 3,951 Ethereum. This holding reflects the combined contribution from our mining activities and ongoing DAT management strategies.

In the current market environment, this crypto portfolio not only provides liquidity, but also for its potential upside if crypto prices recover. We continue to explore innovates mining applications and promote deeper integration between computing and energy. In October, we partnered with local energy infrastructure provider in Canada to convert flare natural gas at [ wheelhead ] into computing power. This project marks our initial step from utilizing standard energy towards broader participation in energy infrastructure. It demonstrates the value of high-performance computing within emerging energy systems and opens up more space and the long-term facility for expansion of our mining business. In R&D and supply chain management, we maintained our focus on product performance, are driving technological upgrades in tandem with capacity organization.

Last October, we officially launched the A16XP, our flagship next-generation air-cooled model, it achieved breakthroughs across multiple performance metrics by delivering over 300 terahash per second per unit with an industry-leading power efficiency of 12.8 per terahash. This showcases our deep technical and R&D expertise in the design of high-performance AC chips. After we continue to advance our product generation updates, upgrades, we work closely with our wafer foundry partners to optimize manufacturing processes. These efforts have led to higher yields and lower costs for our A15 series, allowing us to deliver more computing power from the same amount of wafers. On the supply chain front, our production footprint across Malaysia, the U.S. and Mainland China, allow us to remain compliant with flexibility adapting to increasingly complex global trade environment.

During the mass delivery of the large scale order this quarter, our teams are across manufacturing, quality control and logistics worked in close coordination, successfully withstanding the dual pressure of shipment volume and the tight time lines. By early January 2026, the entire order has been fully delivered, demonstrating that resilience and the execution strength of our supply chain.

In summary, 2025 was a challenging year we navigated account international trade environment while continuing to expand our business across multiple dimensions. For the full year, total revenue was USD 530 million, surging 96.7% year-over-year. We strengthened our prices in North America, partnered with leading customers and increased total computing power stores by 40.7% year-over-year to a record 36.5 exahash per second. In terms of products, we achieved mass production of the upgraded A15 series launched the next-generation A16 series and expanded our home series into a multifunctional product lineup, significantly improving revenue growth and the brand influence. Our mining business reached a key milestone in 2025. With full year revenue exceeding USD 100 million for the first time. Global installed hash rates rose 82% and energized tax rate grew 61% year-on-year. We now operate 9 mining projects globally with total power capacity exceeding 250 megawatts.

We also expanded into innovative energy scenario by exploring wind power, started gas and the computing generate key reuse, driving deeper integration of computing and energy. We completed the deployment of our assembly and production capabilities across Malaysia, the U.S. and Mainland China, building a flexible and resilient global delivering system. We also established our digital assets to management framework, enhancing our crypto reserve capacity and capital allocation flexibility to support our long-term development. As we enter 2026, the external environment remained highly volatile. Shifts in macro liquidity and risk appetite are making digital asset prices and industry demand more cyclical and faced. We do not base our operations on short-term views of price movements is that we focus on navigating cycles through controllable factors, including product competitiveness, delivery and operational abilities, inventory and cash flow discipline, compliance as well as lower cost and more scalable energy and infrastructure capabilities. This post has enabled us to remain resilience and achieve growth in the complex environment of 2025.

More importantly, we do not see Ken's next page as being defined may as an equipment provider or single load computing player. We have a clear long-term vision. Computing and energy infrastructure are becoming increasingly integrated. Bitcoin mining and AI HPC colocation may appear to be 2 different business on the surface, but they are highly complementary and infrastructure level. The share electricity, facilities, power distribution, cooling systems and human and technical resources from -- for operations and maintenance. By leveraging different growth characteristics, we can improve power utilization efficiency and overall project economics. At the same time, these applications can interact with power grid more attractively. They can absorb energy when the power supply spend it and reduce the loan when the growth is constrained, ultimately contributing to more resilience and dispatchable computing infrastructure.

So in 2026, our strategy centers on 2 core pillars with execution as our top priority scaling proven models, streaming line nonrepeatable pilots and laying the groundwork early for long-term commodities. Our first track focused on power and computing infrastructure. We are shifting our strategy from securing power resources from domestic asset-light approach to a more systematic upstream development path to secure reliable and economic power resources and leverage our North American resources base built since 2022, we will prioritize applying for power directly rather than bidding for capacity with the existing third-party projects. We have made significant progress on a robust pipeline to secure directly -- direct power capacity in the U.S. We are confident of securing substantial load by the year end of 2026, potentially reaching the gigawatt scale.

At the same time, we are exploring ways to integrate Bitcoin mining with AI HPC colocation. This approach can improve returns on invested capital while supporting dynamic load management for the power grid and strengthening our existing positive relationships with great operators. The development of power and infrastructure is not as great, but a long journey with steady gates. The process spans marketable stages include site selection, than the great interconnection assessments, negotiations with power partners, contract trucking engineering construction and commissioning. Each step requires careful and delight fusion. Accordingly, our primary objective of 2026 is to establish a pipeline of executable projects and the clear development pathways. We do not intend to pursue one-off large-scale capital outlay. Instead, we will move forward with a framework of capital this plan. We will leverage partnerships and project financing as key tools relying on asset level cash flows and project level financing to sports expansion. This approach limits unnecessary volatility in our overall financial position. This also means we will prioritize secure high-quality power resources that are well suited for AI HPC location.

Second, in the consumer and small- to medium-sized business segments. We will take more sematic approach to building our 2C SMB business in 2026. Last year, we saw strong potential on both the demand side and the gross margin structure for these products. But we also understand that success in the consumer market is hard. Users expect excellent products experience, stability and attention to detail and service. And we must treat this market with a complete respect. That's why in 2026, we will continue to improve our product line and launch new models. At the same time, we will raise our standards and take more cautious about long-term product reputation matters. We will focus on stability, easier to use knowledge control and user experience at our top priorities. Also, we will continue to strengthen our product competitive while we will also focus heavily on building out our channels, a key priority of growing our 2C and SMB business. Our product experience has shown that the consumer market, the core competitiveness comes not only from the product itself, but also from the strengthening of our channels and the service system.

In 2026, we will make systematic investments in this area. This includes partnerships with online platforms, expanding our offline distributor network, improving after-sales services and content operations and building more efficient user engagement and conversion hits. Our message is clear, even in area, while we are still catching up, we are committed to putting in real efforts and resources. And for areas that are key to long-term success, we will go in to make sure the business plan become more stable and cycle-resistant revenue contributor.

Lastly, I will share our view on the operating pace for 2026 and our preparations. From an operational standpoint, we expect 2026 to show a clear stage-by-stage characteristics. Industry demand and pricing may remain [indiscernible] during the first half of this year. Our focus will mainstreaming -- maintaining a strong discipline in cash flow and inventory, strengthening products and delivery mobilities. And the one thing key interactive in power and infrastructure earlier on. At the same time, we are preparing our supply chain and execution teams for potential demand recovering later in the year. if the industry presents a clear structural opportunity, we will be ready to act quickly with strong execution and a healthy cost structure to capture market share and grow efficiency. We believe that this strategy centered on execution and the long-term capability building, we will allow tenant to remain competitive with mining value chain and gradually become a trusted infrastructure participated with the computing power and energy ecosystem. Our goal is to create more sustainable long-term value for our customers, partners and shareholders.

Before concluding, I would like to note that the outlook about contains forward-looking statements. Our actual results may vary due to changes in macroeconomic conditions, industry cycles regulations and market dynamics. We will continue to communicate with transparency and respond to market expansions through clear verifiable institution progress. Given recent global market on uncertainties include ongoing monetary tightening involving geological developments and heightened volatility in the digital asset market, we maintain a relatively cautious view about the market environment in the first quarter of 2026. Our global miners have adopted a weight and see approach in response to the recent decline in Bitcoin prices, the sale of mining rigs are facing considerable changes. While we expect total revenue for the first quarter of 2026 to be in the range of USD 60 million to USD 70 million. This outlook is based on current market conditions and operating assumed patients. However, actual results may differ due to policy uncertainty and marketability.

This concludes my prepared remarks. Thanks, everyone. Now I will hand it to our CFO, James.

Thank you, NG, and good day, everyone. This is James, CFO of Canaan. I'm pleased to share our Q4 financial performance with you.

To begin my part, I would like to echo NG's perspective on the fourth quarter industry environment, it was a very volatile quarter for the Bitcoin price. Bitcoin reached a new high in October, hitting $126,000 before dropping below $100,000 in November and below $90,000 in December. During the quarter, network cash rates also reached historical highs significantly impacting the profitability of minus. Fortunately, our operation was robust in Q4. We successfully secured large-scale orders from key clients in the North American market and globally and our strong supply chain relationships ensured timely production and delivery. In our mining operations, we continued our deployment and see opportunities for new pilot agreements. Overall, we delivered a solid quarterly results in Q4 despite the market dynamics.

Let's take a close look at the details. First, I will highlight our strong top line results in the fourth quarter and for the full year of 2025. In Q4, we delivered $196 million in total revenue, up 30.4% sequentially and 121.1% year-over-year. Our total computing power sold also reached a record 14.6 exahash per second. This growth was primarily driven by the massive delivery of our A15 series. Our revenue increased consistently throughout every quarter in 2025. This trajectory picks at a new quarterly high for Q4. Consequently, our full year revenue reached $530 million, nearly doubling 2024 results. Within product revenue, our Avalon Home Service also delivered exceptional growth in 2025, contributing approximately $25 million in revenue. Notably, our Q4 revenue mainly came from the North American market. Revenue from North American customers reached $125 million, accounting for over 75% of our total product sales. This demonstrates that top-tier institutional mines in North America continue to recognize Canaan as a primary long-term partner.

Regarding our mining operations and the treasury strategy, we continue to scale our infrastructure while maintaining a robust asset base. By the end of Q4, our in-store computing power reached nearly 10 exahash per second, up 7% from Q3. Our digital assets treasury also remains a core pillar of our financial strategy. As of December 31, 2025, we held 1,750 Bitcoins and 3,951 Ethereum. At year-end prices, these holdings were valued at approximately $166 million. While we manage through market fluctuations, this robust reserve provides a solid foundation for our balance sheet and long-term liquidity. Mining revenue in the full year of 2025 was $113.2 million compared to $44 million in the full year of 2024. The increase was mainly due to the increased computing power energized for mining, especially the expansion in the United States.

On the operational front, we achieved a notable gains in efficiency and supported our liquidity through disciplined capital management. Despite our business scaling up, our operating expenses in Q4 were $38 million, decreasing 6% quarter-over-quarter. This improvement reflects our efforts to streamline our organization and focus on core strategic projects. Our strong sales and financing activities have also strengthened our cash position. In Q4, we generated approximately $75 million in cash inflow from sales and received approximately $80 million from a strategic straight equity financing and the brief utilization of our renewed ATM. This healthy liquidity funded our Q4 payments of $100 million to secure our wafer supply and $89 million for production and operations. These investments ensure our flexible manufacturing footprint across Malaysia, the United States and Mainland China.

Consequently, we ended the quarter with a cash balance of $81 million. This aligns with our commitment to strict cash flow discipline, allowing us to navigate market cycles without compromising our strategic road map. Reflecting our strong confidence in the company's financial position and long-term shareholder value, we have already repurchased approximately 2.8 million ADSs for $2 million under our $30 million stock repurchase program announced in December. We intend to continue executing this plan optimistically as market conditions allow, underscoring our firm belief in the company's prospects.

Turning to our margins. We have taken a proactive approach to address market pressures and derisk our balance sheet. In Q4, our gross margin was $14.6 million compared to $16.6 million in Q3. This compression was primarily due to 3 factors: First, we delivered several large-scale institutional orders. These orders are strategically essential for securing our long-term market share in North America. Second, Bitcoin price softened in the latter half of the quarter, this time weakened market demand. These headwinds lowered our average selling price. Last but not least, we prioritize the delivery of industrial machines to strategic customers instead of the Avalon Home series in Q4. Additionally, considering the severe Bitcoin price volatility early 2026, we recorded inventory write-downs of $13.9 million in Q4. These impairments are based on management's latest estimates and reflect our cautious expectations under current conditions.

Below gross profit, the year-end in Bitcoin prices resulted in a $44 million noncash fair value loss. Another $15 million noncash fair value loss was recorded for the conversion of the final batches of preferred shares. There will not be any fair value loss regarding preferred shares conversion for the next quarter. These noncash items led to an adjusted EBITDA loss of $40.5 million. It is important to note that our cash position remains stable, providing us with sufficient liquidity to fund our operations and R&D plans. Furthermore, our ongoing expansion into the consumer and the small and medium-sized business segment is expected to contribute to a more balanced and resilient margin profile over the long term.

Finally, I want to outline our cautious yet resilient outlook. We are monitoring the very volatile Bitcoin price in the first 2 months of 2026. On February 5, the Bitcoin price dropped to $60,000. Low Bitcoin price triggered machine shutdowns and operations closures for higher cost miners. Profitability of existing miners is also under pressure recently including our own mining operations. Given the headwinds and uncertainties, we are taking a very prudent approach to provide our Q1 guidance. We estimate our revenue will be in the range of $60 million to $70 million. In Q1, our priority is to maintain a healthy cash position and derisk our balance sheet. We will allocate our capital carefully between power source investments and wafer supply for computing hash rate, and we were prepared to capture the next market recovery.

This concludes our prepared remarks. Now we are open for questions.

[Operator Instructions] We will take our first question. The first question comes from [ Ben Semmes ] from BTIG.

So it was good to hear about the strong progress in the supply chain efficiency. Kind of curious how the A16 mass production is progressing and kind of any updates around the time line there.

I think you're asking about our new transition rigs, right? Right now, we are sending the A16 machines to our customers. And I think they're during the transacting phase. We are moving ahead with mass production alterations right now. And I think the mass production will be started after the Lunar new year holiday and we expect to begin volume ramp up by the end of the first quarter. Currently, we don't have any issues or anything can broadcast. On the chip side, the chips are already in the mass production. And on the production side, our main focus now is refining the product at a system level. And also, you will know besides the air-cooled version, we have to -- we will have the liquid cool and the emerging cool models. So now looking we're this different models. So we can better match different customer deployment needs and site conditions. I hope this answers your question.

Awesome. That was super helpful. And then just kind of -- I know you touched on it briefly, but just curious for a little bit more color on the difference in the margin profile between the home series and the A15 and kind of how you think this can potentially help keep margins strong moving forward?

I will take this question, Ben. I think currently, we observed the market price has some influence from the big coin price. So seems like the industrial machines, profitability seems to be under pressure, but it looks like the home service continued without serious competition in the market, so we can still maintain the good profitability. But in Q4, the delivery side, we prioritized the industrial orders because it's from the strategically important customers from North America. Looking forward, I think we will continue to see home series play a more important role in our category to generate profit. I don't know if I answer your question then.

The question comes from Nick Giles from B. Riley Securities.

This is [ Wim Chan ] speaking on behalf of Nick. Congratulations on the quarter. It was good to see your heat recovery, your proof concept announcement in Canada in early January. So I wonder, can you speak to the size of the TAM for this opportunity ideally in megawatt terms? And as a follow-up, how scalable is this specific solution? And what are some other ways that you can expand your energy efficiency initiatives going forward?

I think I fully understand the market's interest in the new energy and ESG-related computing products. I think the core value for these executives is transforming its transforming real estate and concentrated energy into matchable, treatable computing power and even for the cash flow.

Yes. However, I'd like to be more candid. These opportunities are highly dependent on specific scenarios such as resource availability, grid connection conditions and even compliance pathways and also the operational capacity. I think the -- we are working on this for more than 1 year. So the scale of each individual projects are typically regions from few megawatts to several tens megawatts. The true total addressable market largely comes from the number of these plants, but we want to conscience against overly optimistic configurations like, okay, we have a few megawatts for each guidance thousands of points there, and so there's [indiscernible] business. We will get that it's 2 mistakes because the business model is still relatively fragmented and the pace of progress [indiscernible].

So over the past year, we started systematically screening potential sites for several POC products. And we already have implemented some of them flex initial operational data. not only the China one, there's many others. Yes. But moving forward, our focus will be on 3 key areas: First is the data and metrology standardization. The second is productization and modularity. Yes, we aim to attend the POCs into replaceable modular solutions. And third is replication and expansion. This is a surface. Yes. I think when we reach the surface, then we will continue to expand similar projects faster. I hope I answered your question.

Your next question comes from the line of [ Mark Palmer ] from Benchmark.

Yes. As you think about Canaan's manufacturing footprint from a long-term standpoint, what would that look like? And where would the company's U.S. manufacturer place to adjust for the tariff environment fit into that?

Yes, I think -- yes. I think over the last year, the external environment has been very dynamic traps, move back and forth. Compliance requirements become tightened in many markets, and it's very volatile. The Bitcoin in this contract, competition is not only about the press and efficiency. It's also about compliance comobility discipline and how fast you can adjust.

So now the supply chain issue is combined with clients. So -- but we treat compliance as a baseline. So we keep high standards across sales, delivery and regional operations. These multiple policy changes last year, we did not take any meaningful surprise loss from policy swings. This is not always the case in this industry. More particularly, some peers have a heavier fixed as exposed in certain regions. And so policy or the market attends quickly their adjustment cost is higher. We do -- we have -- firstly, our sale money is 100% outside China. And also, we built mutiple region production and assembly setup across Mainland China, South Asia, Malaysia and even in California, North America. So this really gives us resilience and continues to become -- region become less predictable.

So no, I think for your question North America is our most important market. And last year, we built thousands of machines from our U.S. manufacturer facilities. So this year, we will carefully review the whole supply chain and make it safer for U.S. customers and expansion of our U.S. -- mid-U.S.A. products.

Your question comes from Kevin Cassidy from Rosenblatt Securities.

I wonder at what point price in the year is the breakeven for your customers sort of Bitcoin mining? I think it had been $90,000, has that changed?

Thank you, Kevin. I think we have 2 lines for this breakeven point. One, including depreciation of the machines, we say it's an all-in payback level. I think it's almost like 100,000 to 110,000 Southern range for Bitcoin price stay there. And the hash price should be like $55 per day. Something between that, that's all in payback level.

Another interesting metric to measure this is the marginal shutdown level because we only consider the energy cost, the variable cost when we start the operation because the CapEx already -- some cost already. So in this kind of scenario, it's quite lower compared to the previous on payback level for various -- for different miners, of course, it's different. If we use our competitors' machine as a comparison, if the electricity cost is like $0.06 and we use our competitors S21+ and we see the shutdown price is like 50,000. And if it's S19XP, it's 66,000. And then look back to our sales for our mining operation. Our average cost is like [ $0.043 ] globally. So our shutdown price for A15 Pro version is like 37,000 when Bitcoin price hitting like 37,000. So we have to shut down the machine.

But of course, in our mining side, we do have some older generation machines. So it varies from like 40,000 to 50,000 to 60,000 in certain cases. So I think that's the part. If we change that to the A16 series, it's a 12.8 [ to ] power efficiency and the shutdown price is about 30,000 for Bitcoin price. So I think this calculation is based on the latest hash price. It will always change because of the network -- the total network hash rate changes and also the bitcoin price changes. So I think in current stage, we have already observed in December, early January and early February, some of the operations in the network has been shut down and the total hash rate moves back to like 900 exahash from previously 1,100 exahash. So we do see a lot of needs in short term has not been released to the manufacturers. But in longer term, when the electricity has already been prepped -- prepared for mining, they will come back for asking for better machines for the latest generation machines. That's something happened in the past cycles, no matter bear market or bull market. Kevin, I think this answers your question?

That was a great answer. Yes, very good. As you get more orders for the leading edge, then what is the foundry availability on the A16 and what's the cost difference for a wafer versus A15?

Yes, I'll answer this one. Since last year with our assignment of market cycles. We have maintained its plant interest strategy with low stock levels. And we secured -- but we still secured a critical foundry capacity and the supply chain resources in anticipation for market recovery this year.

Currently, global foundry capacity is indeed very [indiscernible], particularly for advanced nodes, which are seeing surging demand for AI-related sectors. But we secured our position earlier, so -- and maintain -- because we have long-term partnerships, we're utilizing, rolling forecast prepayments and collaborative ramp-up mechanisms. So our access to wafers and key comments remains stronger than the industry average, what I say is industry average. So -- and regarding the cost, we cannot disclose in specific products of the unit cost for A16 phases upwards pressure in wafer packaging and certain system components compared to A15.

I think it's for sure. Even the macro is ready priced. So our plan is to offset these costs through yield improvements, testing of medicines and design more efficient systems. So overall, what I can say is we expect the unit cost increase for A16 to remain with a manageable rate. Ultimately, we measure competitiveness by our customers' life cycle economics including power efficiency recent stability and delivery certainty. So I think if the market recoveries, our cost and our performance can give you an opportunity to have a good ASP for [indiscernible].

We will take our next question. Your next question comes from the line of Kev Dede from H.C. Wainwright.

A couple of things for you. One is just -- I know you spoke to strategic priorities, but I'd just like to understand the 1 gigawatt facility objective and what that pipeline looks like? Maybe you could add some color there, please.

Yes. Okay. Yes, we will share more details when the time comes. But currently, we are quite confident in the [indiscernible] level power opportunities, which is based on our results from work at this stage. We already work on this for maybe close to 1 year, I think, yes. Second, yes. We believe our goal is to colocate AI HPC and they combined better. So we are -- our hoping is high-quality power resources.

Okay. James, you mentioned, I think, cash outlay of $100 million for wafers and $89 million in operating costs, I think, in the fourth quarter. Can you offer more detail on the wafers you secured? And of the 14 exahash sold, usually give us sort of an average price per terahash. And I was wondering if you could offer some color on that. And whether or not that figure would include the Home series.

Yes. Thank you, Kevin. I think we start from the average selling price. I think the Q4 average selling price is $11.3, slightly lower than Q3, but I have explained a little bit on the margin side just because our average selling price for the institutional miners in a big order, usually it's lower than the small orders for the retail miners. I think that's the case.

And also, we prioritized the industrial minus in Q4 instead of the home service, even the margin side. Home service is better, but we have to make sure our strategical partner, the clients feel satisfied to get our delivery on time. So we tried our best in Q4 and still a few batch delayed to the early January but we completed most of the deliveries in Q4. I think that's very helpful to the clients but not helping our financials, it looks like the profitability part is not is not as good as we expected for Q4. But we streamlined the expenses as well just like I mentioned, the 38 million is the total expense for Q4, which is slightly lower than Q3 because we did some work in streamlining the organization. And I think the wafer supply side, the $100 million secured is most likely the wafer delivered to the customers and also some of the inventories carried to Q1 is ongoing. We still continued that trajectory in Q1, but with a smaller volume of payment to our wafer partner. I think that's something -- some color added to this question, Kevin?

[Operator Instructions] The question comes from Kev Dede from H.C. Wainwright.

NG, can you talk a little bit about your product development? Understand the 12 tools per terahash target for the A16, but you also mentioned chip development that could push you down to maybe 5 or 6 jewels per terahash. Can you talk about that? And whether or not you see a product cycle shortening and when you might think that latest generation ship might be in the market?

Yes, I think it's an open [ price ]. Yes. Let me think about it. I think for the A16 series, currently, we achieved 12.8 doors for the [indiscernible] with manageable cost right. So our target is to -- when the market recover, currently because deep dive for the [ aircon ] price. So the home market is some kind of trading for a few weeks, maybe -- and after the market recovery, we can -- the competition comes back, we can have a very competitive cost and government to our competitors and give our good benefits to our customers. And for the next elution, yes, we will already move to the next-generation development for the chase.

But because we -- you know because it's already at some nano process node, I think the benefits from the process itself is very tight now. And also the cost for the manufacturer of the chips is writing above. We are trying different measures to further improve the energy efficiency. But it looks like after the -- like the [indiscernible] stage when it comes to the [ sub-10 ] products, it's very hard to say that the manufacturer cost is still manageable if we are using today's standard. We already observed that our competitors' product is they have to price very high because we know that roughly cost for the -- for system, you cannot sell at loss forever. So currently, we -- in many different internal meetings, we are what we are, continuing to discussion is if we -- our target is the best power efficiency, then we may be pay for like twice or triple the cost, how can we avoid this kind of situation because the most important part is to let the TCO for the -- for our customers.

So -- also, we also started big-scale infrastructure in the U.S. So I think in the operational side, we are not only the equipment provider. We also got involved the operations any pens our customers have previously will react to our ourselves. So we are thinking about this more and more carefully. Yes. So currently, I think in conclusion, I think the development for the new system we will not accelerate, but also, it will not delayed. But we'll just go at the very natural progress. We will -- I think we will have new -- we're sure, we will have new products this year. And yes, and also the -- we hope at that time, the AI HPC will not relocate -- we will not relocate 100% of the service conductor capacity. And also, we don't have like the DRAM and HBM kind of memories. So sometimes, we are at a very good position. Still, we can use the rapid -- sometimes the peak capacity is released from the foundries, we can use this kind of capacity to get onetime deals to fill our inventory. So this is what we are waiting for this kind of opportunities.

So basically, I think the industry is not coming to the end. It's still going at a normal speed. Yes, this is my personal view.

Okay. Does Canaan have self mining target for 2026? Congratulations for reaching 10 exahash, I know that was a target in '25. I was wondering if you thought about and are willing to communicate where you hope to be at the end of this year?

Yes, I think we -- our priority is R&D and deliver to our customers first. And because of the current -- the market situation, so personally, I move the priority to allocate energy resources instead of just put more rigs on shelf. So yes, the infrastructure will give our the ability to skew it up when the right window opens. Yes. So currently, I think I don't have fixed number for this year, but we have internal goals for electricity infrastructure. So -- and after that, we -- if the window opens, we will rapidly ramp up the hash rate. I think controllable energy resources and the facilities will give us more opportunities to try different business models and do not provide different kind of products to our customer. Yes, I hope like the hash rate sales can come through to our mainstream in the -- maybe in the next year.

As there are no further questions now, I'd like to turn the call back over to the company for any closing remarks.

Thank you for joining the call today, and we look forward to speaking with everyone throughout the quarter. Thanks.

Thank you. That concludes the call today. Thank you, everyone, for attending. You may now disconnect.

Ladies and gentlemen, thank you for standing by. Welcome to Canaan Inc.'s Third Quarter 2025 Earnings Conference Call. [Operator Instructions] Please note that this event is being recorded.

Now I'd like to hand the conference over to your speaker today, Gwyn Lauber, Investor Relations for the company. Please go ahead, Gwyn.

Thank you, operator. Hello, everyone, and welcome to our earnings conference call. Joining us today are our Chairman and CEO, Nangeng Zhang; and our CFO, Jin James Cheng; Leo Wang, Vice President of Capital Markets and Corporate Development; and Xi Zhang, Senior IR Manager, will also be available during the question-and-answer session. Our CEO will start the call by providing an overview of the company and performance highlights for the quarter. Our CFO will then provide details on the company's operating and financial results for the period before we open up the call for your questions.

Before we begin, I would like to refer you to our safe harbor statement in our earnings press release. Today's call will include forward-looking statements. These statements include, but are not limited to, our outlook for the company and statements that estimate or project future operating results and the performance of the company. These statements speak only as of today, and the company assumes no obligation to revise any forward-looking statements that may be made in today's press release, call or webcast except as required by law. These statements do not guarantee future performance and are subject to risks and uncertainties and assumptions. Please refer to the press release and the risk factors and documents we file with the Securities and Exchange Commission, including our most recent annual report on Form 20-F for information on risks, uncertainties and assumptions that may cause actual results to differ materially from those set forth in such statements.

In addition, during today's call, we will discuss both GAAP financial measures and certain non-GAAP financial measures, which we believe are useful as supplemental measures of the company's performance. These non-GAAP measures should be considered in addition to and not as a substitute for or in isolation from GAAP results. You can find additional disclosures regarding these non-GAAP measures, including reconciliations with comparable GAAP results in our earnings press release, which is posted on the company's website.

With that, I will now turn the call over to our Chairman and CEO, Nangeng Zhang. Please go ahead.

Thank you, Gwyn. Hello, everyone. This is NG, CEO of Canaan. Welcome to our earnings call. Together with our CFO, James, we are calling from our Singapore headquarters to discuss our Q3 2025 business results, and let me start this with you.

During the third quarter, the global macro environment remains highly uncertain. In particular, the U.S. recently [indiscernible] tariff policy increased mining costs in North America. However, we also saw the resilience of North America market. Once there was a bit more clarity, demand started to recover clearly during this quarter. Bitcoin prices increased from approximately [ 107,000 ] at the end of June and about [ 114,000 ] at the end of September. This showed a rapid increase in total global hashrate, which rose from [ 846 at hash per second ] at the end of Q2 to [ 1,041 at hash per second ] at the end of Q3. Accompanied by a significant rise in mining difficulty with growing energy competition globally, the mining industry is facing higher operational alternatives.

Despite complex external environment, we delivered results that exceeded expectations. Total revenue for the quarter exceeded USD 150 million, up 50.2% quarter-over-quarter at 104.4% year-over-year and beat our guidance range of USD 125 million to USD 145 million. Gross profit reached USD 15.6 million, much higher than the USD 9.3 million reported in Q2. This improvement in the revenue and gross profit reflects our rapid response to the market, demand and ongoing optimization of global mining operations. Supported by strong sales and revenue momentum, our cash balance at the end of the quarter increased to USD 119 million, representing an 80.9% [ screen show increase ].

In mining machine sales, we delivered a record high of turn [indiscernible] of computing power in Q3, up 55.6% sequentially and 37.7% year-over-year. Our average selling price increased 33.8% year-over-year to USD 11.8 per terahash. Despite a slightly -- slight rise in cost for terahash due to the changes in international trade policies, during this quarter, we achieved a product gross margin of approximately 17%. We continue to serve strong hashrate demand in Asia and also captured the recurring demand in North America.

Notably, during this quarter, we secured large orders from well-known customers in the region, including [ Bitfury, Cyren Spark and Luxer ] in early October, we signed a purchase agreement for over 50,000 A15 pro models with a U.S.-based minor client. This highlights growing recognition of our product performance, quality and service by North American institutional customers.

In the consumer grade mining machine market, our overall home series continue to lead in this emerging space. In additional to revenue marketing and promotional activities, we have also included the Home series in our open source code program. We are actively growing our user and developer community and expanding our brand influence. At the same time, we are exploring new applications of the Home series in smart home scenarios. Currently, we are developing software to make our products compatible with [ Matter ], the mainstream protocol standard for smart home devices. In terms of consumer-grade product sales, we delivered 14,000 units of the Avalon Home Series in Q3, generating over USD 12 million in revenue, a sequential increase of [ 115.3%. ] The Avalon Q model was the top performer this quarter by supporting scale sales through channel partners, the Home Service achieved nearly USD 4 million in gross profit with a solid gross margin of around 33%.

Overall, our total product revenue reached USD 118.6 million with gross profit close to USD 20 million in Q3. The Avalon Home series contributed 10.3% of total product revenue and about 20% of product gross profit. Based on what you are seeing, competition in the consumer money market remains relatively healthy. We plan to maintain solid gross margins with launching new products and expanding channel coverage to drive scale.

Turning to mining operations, despite a notable increase in mining difficulty [ due ] in this quarter, our disciplined execution allow us to steadily advance hashrate development, amortization and overall money efficiency. As a result, we generated another quarterly record of USD 30.55 million in mining revenue while maintaining competitive power costs. In the third quarter, we added approximately [indiscernible] of newly deployed capacity in North America, bringing our total deployed hashrate to 9.3x hash per second by the end of third quarter and approximately 7.8x hashrate per second energized. We mined 267 BTC during this quarter, which further contributed to our crypto asset balance. Our Bitcoin holding reached an all-time high of 1,582 BTC by the end of the quarter, providing solid support for our balance sheet.

We are also actively exploring innovative mining projects. This quarter, we partnered with Soluna to deploy machines at 20 megawatts wind powered mining facility in Texas. In Canada, we worked with a local energy infrastructure partner on a pilot project that convert standard -- stranded natural gas into computing power. We also supply the mining equipment for project design to support local grid stability. These projects mark our first step into the energy infrastructure space, bringing with the utilization of stranded energy. Our long-term [ region ] is to integrate high-density interruptible bitcoin mining [ laws ] with energy-intensive AI and [ HPC ] workloads, building a future where computing and energy infrastructure grow together.

We are entering an era in which AI, software and data centers will profoundly shape daily life. At the same time, we believe that public awareness and the demand for sustainable energy will continue to grow. Throughout Canaan's history, we have held a consistent belief technology should make society more efficient. Today, we are seeing that vision becomes to materialize.

We have unique advantages in this transformation with more than a decade of developing technologies that make chips and systems more energy-efficient. We are now extending these capabilities to both home use and traditional energy sector. Energy operators can use our computing system to balance the grid, improve transmission efficiency and generating new revenue.

On the consumer side, utilizing excess heat from home mining, it's only the beginning. Over time, we envision this concept expanding into broader home computing applications.

For R&D, we continue to innovate and upgrade our products. At the end of October, we officially launched our next-generation [indiscernible] [ A16 ] series, the air-cooled A16XP model delivers 300 [ terahead ] per second of hashrate per unit with an industry-leading power efficiency of [ 4.8 joules per terahash. ] This marks the first time our air-cooled monitor have reached 300 terahash level, clearly showcasing our strong leadership in bitcoin ASIC design. We see improvements to production and supply chain. Our global delivery system is now more flexible and resilient. Today, we have manufacturing [ abilities ] layouts in Mainland China, Malaysia and the U.S., working seamlessly together to support delivery and after-sale service for consumers worldwide.

While enhancing our product and supply chain capabilities, we have also sharpened our focus on core operations. Starting this quarter, we realigned our R&D team and [ around a team ] around the projects that offer clear revenue visibility and strategic value. We have also streamlined head count to support this focus. In addition to organizational and cost optimizations, we are also allocating additional resources to expand our business footprint. We have established a dedicated consumer product team to optimize product quality and accelerate product iteration.

Additional, we are also looking more resources to our hashrate finance and energy infrastructure interactive. We see new power-related opportunities in many regions from home users and small business to power utilities. In Europe and Asia, customers are exploring ASIC-based grid balancing applications. In North America, stranded energy opportunities continue to grow and with similar projects that are emerging globally, including the Middle East.

In our digital asset treasury management, we continue to executing our flexible strategy. At the end of the third quarter, we held [ 1,182 ] Bitcoins and 2,830 [indiscernible]. In early November, during a market [ pullback ], we [ strategically ] acquired an additional 100 Bitcoins as part of crypto asset management strategy, further enhancing our asset allocation and the potential liquidity.

To sum up, Q3 was a highly strategic quarter in Canaan's development [indiscernible] We achieved strong revenue growth and improved gross profit while also optimizing our business structure and organization. At the same time, we made an encouraging progress in several new areas. Looking ahead, we are fully focused on driving Q4 sales, fulfilling large customer orders and converting preorders for our new A16 series. At the same time, we are accelerating the deployment of new [indiscernible] innovate mining products to further expand our money hashrate. We are closely monitoring the impact of U.S. tariff policy, macro liquidity conditions and the potential changes in global mining and energy regulations. Taking all of these factors together, we remain cautiously optimistic for the fourth quarter and expect total revenue to be in the range of USD 175 million to USD 205 million. This outlook is based on current market and operation -- operating conditions and the actual results may vary with policy uncertainties and market volatility.

This concludes my prepared remarks. Thank you, everyone. Now I will hand it over to our CFO, James. Please, sir.

Thank you, NG, and good day, everyone. This is James, CFO of Canaan. I'm very glad to share our Q3 financial results with you. Even today, we are witnessing bitcoin price under big pressure. As NG stated at the start of the call, the macroeconomic environment in Q3 was highly uncertain. Reciprocal tariff policies from the United States added mining costs in North America. Global network hashrate growth continuously outpaced the Bitcoin's price appreciation. This all led to increased mining difficulty and intensified operational challenges across the industry. Despite market volatility, we delivered strong results this quarter. Our revenue exceeded our own expectations. Our gross profit showed consistent growth with the average selling price climbing again and our reserves of cash and digital assets increased significantly in our ending balance sheet of September.

Let me give a quick summary of our financial performance. First, we delivered a total revenue reaching $150.5 million, exceeding our guidance and representing 104% year-over-year increase. This was primarily driven by growth in our product sales of $118.6 million, surpassing the $100 million milestone for the first time in the past 3 years. This growth was achieved while we set a new record of [ 10 exahash of ] quarterly computing power sold and average selling price continued rising to $11.8 per terahash per second, a new high for the past 2 years. After a very quiet Q2, our clients from the United States started actively placing sizable and repeating orders for the A15 series. Sales of North American customers contributed 31% of our total revenue in quarter 3. We are happy to witness the strong demand recovery of the North American market.

Also, our sales of Avalon Home Series generated $12.2 million in revenue during the quarter, representing a 115% quarter-over-quarter increase. This is the first time Avalon Home products have contributed over 10% of total product revenue since the launch just over a year ago. As NG said, we are cultivating the consumer market and establishing our leadership position in the newly defined household mining category.

Second, our mining business also delivered another record result this quarter. Mining revenue reached $30.6 million, an all-time high and a 241% year-over-year increase. We mined the 267 Bitcoins during the quarter, representing 82% year-over-year growth. During the quarter, we deployed over 8,000 mining machines across our projects in the United States and other countries, expanding our total deployed hashrate by 14% from 8.15 [ exahash ] per second at the end of quarter 2 to 9.3 exahash per second at the end of quarter 3. Our in-store computing power in the United States also grew by 20% from 3.66 exahash per second at the end of quarter 2 to 4.4 exahash per second at the end of quarter 3. We also strategically closed our mining operations in Kazakhstan and initiated a small-scale project in Malaysia.

Next, our profitability continued to improve this quarter. Gross profit reached $16.6 million, up 78.6% quarter-over-quarter, marking a significant turnaround from a gross loss of $21.5 million in the same period last year. Product gross margin reached 17% this quarter. Both gross profit and margin continued their growth in quarter 3, expanding the upward trajectory and reinforcing the positive trend. Our Avalon Home series generated nearly $4 million in gross profit with a gross margin of approximately 33%. The Avalon Home series accounted for around 10.3% of product revenue, and it contributed 20% of the products gross profit. The Home series has already become a stable revenue pillar and our recognized gross profit contributor.

Last but not least, our total cryptocurrency [indiscernible] reached approximately 1,582 Bitcoins and 2,803 [indiscernible] with an estimated market value of approximately $189 million at the end of Q3. Our unrealized holder gain was approximately $87 million, reflecting the appreciation in value of the digital assets accumulated from mining and other operations. As of October 31, our total Bitcoin [indiscernible] increased to 1,610, as previously disclosed in our monthly report. In early November, we further strengthened our digital assets portfolio by purchasing another 100 Bitcoin.

Turning to expenses. Our operating expenses totaled approximately $40.5 million. We recorded $1.5 million in onetime expenses relating to the operational efficiency initiatives, including organizational optimization, travel control measures and other related items. In addition, we recorded $1.2 million in impairment related to mining machines deployed in Kazakhstan. By the end of quarter 3, the price of Bitcoin increased to around $113,000 versus around $107,000 at the end of quarter 2. The price of [indiscernible] increased to around [ 4,100 at ] the end of quarter 3 versus around [ 2,500 at ] the end of quarter 2. This price appreciations resulted in an aggregate unrealized fair value gain of $5.7 million on our digital asset holdings.

A noncash change in fair value of preferred shares impacted our Q3 bottom line by $9.5 million. This included a $5.4 million from the Series A-1 preferred shares, which were fully converted during the quarter and another $4.0 million from Series A preferred shares, which were fully converted in early October. To provide a clearer view of our underlying operational performance, we have excluded the impact of this accounting treatment from our non-GAAP measures. With all preferred shares now fully converted, we expect Q4 to include a final impact related to the change in fair value of these instruments. Benefiting from strong top line growth, improved margins and firm cost discipline, we delivered a positive adjusted EBITDA of $2.8 million in quarter 3. Our net loss per ADS narrowed to just USD 0.05 versus $0.27 in the same period last year, demonstrating continued momentum toward profitability.

Turning to our balance sheet and cash flow. We generated a net cash inflow of $53 million in Q3. This was driven by $189 million in sales collections, the highest quarterly level in the past 2 years and supplemented by approximately $10 million in export [ VAT refunds ]. These inflows fully converted the quarter's major cash outflows including $56 million in wafer prepayments and $90 million for production and operations. As a result, our cash balance increased to $119 million at quarter end.

Now moving to our contract liability. The balance of contract advances reached nearly $87 million as of this quarter end, over 85% contributed by North American clients. As of the end of quarter 3, we recorded account receivables of $7 million, all from the customers using Bitcoins as collateral for installment payments. We will continuously evaluate market demand and adopt corresponding credit policies with caution.

Now turning to our recent fund raising. In early November, we closed a strategic investment totaling $72 million with 3 top-tier institutional investors, [ Brevan Howard, Galaxy Digital and Waste Asset Management. ] The proceeds are intended to fund the acquisition and deployment of North American data center sites as well as the expansion of our Bitcoin mining machine production capacity. In late October, we renewed the [ ATM ] program to broaden banking relationships and enhance our financial flexibility for future growth initiatives. Following the renewal, we sold approximately [ 4.8 million ] of ADSs, raising gross proceeds of about $7.8 million as previously reported in the monthly report. We have selected to post further sales under the ATM for the remainder of 2025.

As of the date of the earnings we have cumulatively repurchased approximately 5.1 million ADSs for approximately $3.4 million under our share repurchase program. In the future, we plan to execute on our repurchase plan as market conditions allow us. Moving forward, as our CEO just mentioned, strategically, we will continue our technology-driven efforts with the goal of improving the efficiency of society. These efforts included the development of energy-efficient chips and systems, similar to what we did in the past decade. This includes an extension of our energy operations, which leverages computing technologies.

Also on the consumer side, these efforts include Bitcoin computing and heat reuse. To better utilize our resources, we set up additional internal controls to oversee the operation of our business. These priorities of the strategic importance and will help to provide us with additional revenue visibility. We will increase the expansion of our consumer products and energy operations. But at the same time, streamline existing R&D and administration cost structure.

In cash flow management, we will continue to invest in R&D on new products and wafers in the supply chain. And we will also seek opportunities that will increase our energy operations around the world as well as help our digital assets [indiscernible] to accumulate more digital assets on our balance sheet. All this will happen in a very dynamic environment. We remain cautiously optimistic as we execute on our strategy while focusing on protecting and increasing our shareholder value. We expect revenue for the fourth quarter to be in the range of $175 million to $205 million. This forecast reflects current market conditions. Actual results may vary given policy uncertainties and market volatility.

This concludes our prepared remarks. We are now open for questions.

[Operator Instructions] Your first question comes from the line of Mike Grondahl from Northland.

The 50,000 machine order on the [ A15 PROs ], can you talk a little bit about delivery timing there and gross margin on those sales?

Good morning. Yes. This order for more than [indiscernible] it's one of our most important deals this year. So under the contract, we expect to complete all deliveries by the end of 2025. So far, we have shipped a part of the orders and progressing in the remaining production and logistics [indiscernible] Yes, given the size and the tight time line of this order and the fact that Q4 is generally a peak period for supply chain logistics constraints, our production and operations team are working at full strength to ensure on-time delivery while maintaining product quality.

And also at the same time, we are expecting deliveries for other customers in parallel to avoid any impact on our long-term -- other long-term partners. This is a key test of our delivery management capabilities [indiscernible] Yes. So for the gross margin, yes, we have -- I think we have positive gross margin, yes. But maybe I cannot [indiscernible] numbers. Yes, [ we have gross margins, yes ].

Got it. And then just maybe a follow-up. Your home mining sales have done really well lately. What are the margins on that business line versus the industrial mining equipment?

I think for our home Home Mining series, in quarter 3, we get 33% of gross margin. And by the end of this year, I think we should maintain 30% -- about 30% gross margin. It is significantly higher than industrial [ manners ]. Yes, [ I think that would be it ]. Yes, yes, right. So yes.

So for -- I think for the home [ miners road map ], we plan to launch several new products over the next 12 months and further [ about to see the 2C product ] portfolio, always consumer products need a refresh every year. So we need to refresh almost all existing models in the coming year. So -- but still, for 2026, our most important KPI for the home series still go mainstream and break out of the [indiscernible] [ niche ] please give us some more time.

Next question come from the line are Nick Giles from B. Riley.

Thank you, operator, and good morning or good evening, everyone. This is Henry [indiscernible] On for Nick Giles. So for my first question, when is the earliest you guys could ship your new A16 models and at what scale? And what are your expectations on price and margin, respectively?

Yes. The A16 series was officially launched at the end of October. And now we are at the first batch [indiscernible] production and yes, and we finished the testing stage. According to our plan, we will start shipping [ samples ] to selected customers by the end of this month for their testing and evaluation. Yes, this is constant with our [ euro large ] strategy and we expect to begin our volume shipments in the first quarter of 2026. And yes, and we will adjust production and delivery pace dynamically based on the presale and customer feedback.

For pricing, we will adhere to market-driven principles, taking into supply/demand [ account ] and the competitive demand dynamics and the customer mix. So [indiscernible] a new flagship product, AS16 delivers major performance. The air-cooled A16XP can offer over 300 terahash at [indiscernible] per hash, so which is really industry-leading. So I think it will provide higher [indiscernible] per unit. And also, we can share these benefits with our customers, yes.

So [indiscernible] margins based on the current wafer material and manufacturing costs, the per terahash cost for A16 is under control and this in our expectations and also the yield is acceptable. So I think the product pricing power will help us to offset some cost pressures. Sure, the A16 cost per terahash is higher than A15. So yes, so let's see. Thank you.

Yes. And then for my follow-up, I wanted to get your guys' thoughts on the fact that several public Bitcoin miners have been very vocal about winding down their mining operations in the medium term and then at the same time, supply of ASIC appears to be increasing. So what do you guys think of the market impact will be? And then how is Canaan responding to this trend?

Yes, for -- I think for this question, yes, we will observe that some listed miners maybe they are facing balance sheet pressure, share price performance issues and a desire to pivot towards [ AI, HPC ] have publicly started their intention to reduce Bitcoin mining over the medium term, yes.

But for -- in my perspective, firstly, I think the slowdown, the -- I don't think the global hashrate will slow down the near term and also the AI HPC deployment is still need some time. But our investigation to the energy market in the U.S., the [ AR HPC ] applications needs high-quality energy electricity and which is the high quality always means higher cost. So I think fundamentally, at the next 12 years, the mining -- the power [indiscernible] for mining is not the competition with the energy used for AI/HPC is not the same electricity. So I know our customer, including ourselves, is thinking about how can build mining -- AI-ready mining facilities for the future. But in this stage, deploying more ASIC Bitcoin miners is still the best way to allocate energy today and generate revenues in -- from this stage not waiting for another 1 or 2, 3 years.

So I think still -- the things, it's hard to foresee for long term. So we focus on -- yes, so [ because there's the answer for ] 3 or 5 years later. So we -- now we are focused on cooperate with our partners to fulfill their requirements. We also -- we are also trying to find more energy resources in U.S. and building our own mining sites today. And maybe -- and we should have a potential possibilities to transfer to the AI infrastructure in the future. This is what we are personal observed in the past maybe 6 months.

Our next question comes from the line of Kevin Cassidy from Rosenblatt Securities.

Congratulations on the strong results. And your guidance for $190 million for the fourth quarter is impressive. Is this -- orders also scheduled out into the first quarter? I guess what kind of visibility are your customers giving you?

Thank you, Kevin. I think quarter 4 is a big quarter in terms of seasonality, and we provided the guidance in a very optimistic way. And also, we have already collected some of the orders. We try our best to deliver in quarter 4. Looks to me, quarter 1 traditionally is the low season because there is a New Year and the Chinese New Year together in the western part of the world and the eastern part of the world, both having all kinds of holidays and the global logistics supply chain is not in a normal shape. So I don't personally seeing another peak time for quarter 1. I think the revenue could be going down a little bit. But we will try our best to deliver quarter 4 first and then we predict quarter 1 later when we have a clear understanding about the demand.

Also recently, the Bitcoin price is not in a good shape. So it's under turbulence. And some of the customers, especially the smaller ones, they tend to be more cautious and hesitate to make up their decisions immediately. That will also have a kind of impact on quarter 1 orders. So we will try to make a flexible supply. Anyway, currently, I think the demand is still higher than supply. We're just focusing on quarter 4 delivery first. And then let's see how it goes in quarter 1. Maybe we can balance between the demand the sales and also the sales mining side. If we do have some inventory, we can allocate to self mining in the United States. That will also be long-term strategic goal for us. Yes, I think that's my 2 cents, Kevin.

Thank you, James. That's very good detail. Then maybe you did note that there's a rebound in demand in the U.S. Is the U.S. market, which is less sensitive to the price of Bitcoin?

Sorry, come again?

Okay. Yes. Just you had mentioned that with the price of Bitcoin being down in just very recent times last few days, and you mentioned that would be sensitive to the demand for mining machines. And I was just wondering if the rebound we've seen in the U.S., I think you said it was 31% of revenue in the third quarter, whether that continues even with, I guess, is it less sensitive in North America to prices of Bitcoin versus the rest of the world?

Yes, Kevin, look, to me, in my observation, North America is now the leading area for global mining industry. The whole total hashrate in North America is some percentage between 35% to 40% globally. And there are around 20 listed companies in North America doing mining. They are kind of institutional players. They are more professional building of the site, the electricity facilities and eventually becoming mining sites.

So they have their schedule. It's not easy for them to stop their own schedule even when Bitcoin price has some short-term turbulence. For them, they look at long-term goals. That's why they are not very price sensitive in very short-term time. But we observed the tariff did have a kind of impact on their cost structure, that increased their mining cost. That means some of the miners, especially the smaller ones, even the city United States with consistent policy advantage, they're going to steal withdraw from Bitcoin mining to other activities. They may want to change their [ minus ] purchase plan in quarter 4. So I should say U.S. customers are most important customers for us. And we observed the worries in the short term, but we also respect their long-term strategic goals, and we try our best to support their strategic goals to get realized. So that's something we do together with them.

Yes. And also, I think for looking at this year, especially the market [ initially ] expected is the demand will [ flow ] rapidly into North America. However, changes in tariff policy led to a significant contraction in North America demand from late Q1 to Q2. And at that time, [indiscernible], but [ Asia ] demand ramps up quickly and partly offset the weakness in North America. And in Q3 and Q4, North America customers showed very strong resilience.

Together, we adapted to the new trade environment and the demand there has recovered quickly since [indiscernible] In fact, for the potential already delivered in Q4, North America has begun -- has again accounted for more than 50%. So [indiscernible] price mentality is constant. Sharp moves over a few days or weeks do cause some customers, especially small customers to pause and receives. But over multi years time frame, I think the impact is underlying demand trend is limited. And I highly disagree with like running business by [indiscernible] numbers. So this is my personal opinion. Thank you.

Our next question comes from the line of Michael Donovan from Compass Point.

James, how much inventory do you have left for the A15 series? And then for Q2 -- guidance, what mix do you expect between A15 orders and preorders for A16?

Thank you for the question, Michael. I think our inventory in the end of quarter 3, it's like $200 million, including some of the raw materials like wafers, like other components. And it mainly reflects the strong demand in quarter 4. And you have already known, we got the big order around 50,000 units to United States. So we have to prepare the inventory.

Other than that, if we digested the inventory in quarter 4, I don't think our inventory level will be that high. In quarter 1, we will see a lower inventory level for A15. And that's because we are expecting the uncertainties of the market demand in quarter 1. And for A16, I think it's mainly quarter 3 to be the mass delivery. I think the early delivery could be late quarter 2. But in the transition, we will continue to produce A15 and make it better and better. I think that's the plan. Did I answer your question, Michael?

Yes. You did, James. I appreciate that. And then I guess for my next question, can you expand a bit more on the pilot projects that you have, the 2.5 megawatts in Alberta, Canada and 4.5 megawatts in Japan. What are the growth opportunities in those 2 countries?

Yes, I think for -- we are running several similar pilot projects globally. Actually [indiscernible] Japan, Canada, U.S. and as well as some small projects, it's ongoing in Europe and Asia -- other Asian countries. Since these are pilots, our primary goal is to validate the technical approach and business model rather than maximum like early-stage financial returns. Yes, it's thanks to the use of [ trans gas ] and energy, the power cost for these pilot projects is relatively low and the project level gross margins are decent, but like most money operations, a meaningful economic [ benefits ] ultimately requires scale, yes.

But based on the current results, we believe these pilots all have the potential for scale up. This is very important to remember that power and gas infrastructure are very, very traditional long-cycle industries, building trust and proving out a new model takes time [ on the patients ]. Our strategy is to run the pilots in a stable way, cement the partnership and then we [indiscernible] the larger megawatt levels at the right time.

For example, the Canada [ strand gas ] product. There's a very highly possibilities to scale up to 20 megawatts in the middle of next year. And also, we can do more like, I just mentioned, the AI-ready mining site, [ many farms ] in U.S. with our partner [indiscernible] . So yes, so I think still there's -- please give us some time.

Our last question comes from the line of Kevin Dede of HCW.

Gentlemen, thanks very much for having me on the call. I appreciate it. NG, I'm wondering about your self-mining objectives. Can you refresh us on where you plan to take self mining? In particular, Ethiopia, which remains the largest contributor of your exahash. We're hearing that power tariff rates have increased there, and we're wondering how you might rethink hash deployment.

Okay. Thank you. And I think for our strategy, now in short term, there is some pullback in [indiscernible] And many people are asking the question about our strategy of [indiscernible] Yes, I think in the near term, over the -- maybe over -- next few months, our attention will be on delivering large [indiscernible] orders and which does slow the pace at which we add our self-mining hashrate. Please remember there's still other customers. We cannot [indiscernible] our long-term partners at this time.

So we -- because of the lack of machines at the same time, we are actively developing more [indiscernible] of this including potential greenfield sites. These projects have longer construction cycles, but relatively controllable cash outlays and the [ over better ] long-term value and operation flexibility. The [ gas to compute pilot ] in Canada with area energy and the [indiscernible] data center projects in Texas with Soluna [indiscernible] examples.

For what we see in the market, I think this is indeed more attractively priced mining assets now. The pullback for Bitcoin price gave us benefits to get more energy resources, especially in U.S. yes. So I think [indiscernible] projects with solid resources but short-term funding pressure. So this offers us better entry points. We are continuously screening [indiscernible] return on the risk control and we aim to expand our self mining footprint in a more prudent value attractive way.

So yes, so [ in fact ], I think we are still keeping the [indiscernible] in U.S. And we are moving to the more fundamental size like the energy infrastructure [ in the ] long term, yes. And I think the big order and Bitcoin [ pullback ] [indiscernible] time [ to direct our direction ] to find a more better way to expansion in U.S.

James, I was wondering if you could offer a little more color on the [ 56 million ] wafer purchase and the [ 90 million in ] processing. Would that include pretty much everything that you need for the A15 and 16XP at least as you see orders initially? And how much of it do you think translates to the Avalon Home series?

Well, Kevin, I don't think [ 56 million ] is all the wafer supply we can get for quarter 1. It's actually some payments happen to be in the face of payments just in quarter 3. So the $56 million is some prepayment and also some close payments for the previous contracts. And I think in quarter 4, we will pay more. It's just kind of pacing difference.

And for the Home series, I think currently, it's wintertime. We observed the demand from North America is actually getting stronger compared to quarter 3 and quarter 2. So it seems like we will allocate more chips to Home Series. But of course, we don't want to generate a lot of inventory. We will still produce according to the orders. But to be very honest, currently, we have already noticed the Home series will occupy a higher percentage in quarter 4. And while the total revenue is so big, so we are expecting the sales for quarter 4 of Home series.

And actually, a lot of buyers, a lot of consumers, they posted in their social media talking about Avalon Q. They like it because it's quiet and it can generate Bitcoins and using the same kind of energy in the past, [indiscernible] can do. Actually, we can feel the passion from the consumers asking more for the supply side. That's why when we do allocate the chips, I think internally, we have some discussions and sometimes even very fierce competition between the consumer sector and the industrial sector. But of course, NG will try his best to balance different product lines and different categories, try to satisfy most of the customers and consumers.

And also, currently, the market environment is indeed very complex and changing very fast and especially for the semiconductor sector. You know the -- like the [ DRAM ] price is maybe doubled in the past few months. It's only -- it's indicated that how the types of the global capacity for the semiconductor industry. So currently, I think we have demand because the demand of [ advanced cheap ] growth, especially for the AI related locations and many other stuff, the foundry capacity today is very tightened and also the price is trending up significantly.

I think this could impact both our manufacturing costs and mining CapEx. And this -- but this is influenced the whole industry, not only us. That said, well, we cannot share the exact figures, but we have -- already have secured meaningful wafer allocation for next year at favorable pricing and payment terms. Thanks to our strong relationships with our key suppliers, the volume is built on the cautious number, but this will -- I think this will definitely give us a good cost position heading into 2026. Yes, this is my [ consent ].

James, you didn't touch on the $90 million processing. Can you just give us a feel for that and what the implications are for future cash use?

$90 million. I think...

It's too much.

Yes, I think it's -- you mean, that [ $72 million fund ] raise from the strategic investors and also [ $7.8 million ] from ATM program. I think putting this together is like $80 million.

Okay. No, I thought that when you were discussing cash used in the third quarter, you mentioned $90 million. I apologize. I probably have the number wrong, but...

Yes, you mean the operational and the supply chain together, the expenses?

Right.

I think that -- yes, that outflow is for some payments of the supply chain like components like all kinds of production and logistic to shift the components from here and there, a lot of things, including some of the expenses related to that. I think that's a major part of the supply chain expenses.

And also, I think there is the R&D, G&A and also sales and marketing fees inside this, I think the run rate is still like $28 million to $29 million. Even the P&L shows it's like $40 million, but that includes a lot of noncash items like share-related salaries, but the rest goes to like $28 million to $29 million to -- for the normal operation. And also, we have expenses related to the operation like travel, like marketing. Especially for the consumer product, we started to have some marketing [ try ] in quarter 3 but not much expenses. But that is something we try to do in the transition from a pure machine company to a kind of operational company with energy and also with the 2 consumer products, we will also increase our marketing expenses in the future. I don't know if I answered your question.

Yes, yes. Just 1 more little nuance. I'm just wondering if those payments include prepayments for supply chain, securing supply chain component through the December quarter and into the March quarter?

That's a wonderful question, Kevin. Usually, we only do prepayment for wafer. Most of the components, we usually get the components first, and then we pay them a little bit later in different kind of terms. For example, like 15 days, 30 days, something like that. It's not -- usually not advanced payment.

Well, congratulations on that 50,000 unit order. Congratulations on the sharp pop in revenue and gross margin.

Thank you, Kevin.

Thank you for the question. As there are no further questions, now I would like to turn the call back over to the company for any closing remarks.

Thank you once again for joining us today. If you have further questions, please feel free to reach out to us, and we look forward to speaking with you throughout the quarter. Thanks.

Thank you. That does conclude today's conference call. Thank you for everyone for attending. You may now disconnect.