Ladies and gentlemen, welcome to Canaan Inc. Fourth Quarter 2025 Earnings Conference Call. At this time, all participants are in a listen-only mode. After the management's prepared remarks, we will have a question and answer session. 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 Chairman and CEO, Nangeng Zhang, and our CFO, James Jin Cheng. Leo Wang, Vice President of Capital Markets and Corporate Development, and Shi 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 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 Nangeng Zhang, CEO of Canaan Inc. Welcome to our earnings call. Together with our CFO, James Jin Cheng, 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 above its previous all-time high, reaching approximately $126,000. It then fell below $100,000 in mid-November and dropped to below $90,000 by December. At the same time, the wave of new hash rate that entered the pipeline during the price surge in the first quarter came online, pushing total network hash rate to a record high. This puts strong pressure on miners' profit margins. Facing this highly volatile market, we managed our sales pace well at the '4. We secured a large order from a major customer in North America, and efficiently mobilized resources to support production and smooth delivery. At the same time, we steadily expanded our self-mining operations and explored diversified mining partnerships, starting several pilot projects. As a result, our total revenue for the quarter reached $196 million, up 30.4% quarter over quarter and 121.1% year over year. This was our highest quarterly revenue in the past three years and exceeded the midpoint of our guidance range of $175 million to $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 the market's growing recognition of our product performance and delivery capability. To ensure high-quality delivery, our supply chain, production, and operation teams worked closely together to ensure smooth and high-quality…

Thank you, Nangeng, and good day, everyone. This is James Jin Cheng, CFO of Canaan Inc. I'm pleased to share our Q4 financial performance with you. To begin my part, I would like to echo Nangeng'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 hash rate also reached historical highs, significantly impacting the profitability of miners. 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 operation, we continued our deployment and seized opportunities for new pilot agreements. Overall, we delivered solid quarterly results in Q4 despite the market dynamics. Let's take a closer 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 13.4% 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 peaked 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 series 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…

Thank you. We will now begin the question and answer session. As a courtesy to other investors and analysts, please limit yourself to one question and one follow-up. If you have any additional questions after the Q&A session, the Investor Relations team will be available after the call. For the benefit of all participants on today's call, to ask a question, you will need to press 11 on your telephone and wait for your name to be announced. To withdraw your question, please press 1. 1 again. We will take our first question. The first question comes from Ben Summers from BTIG. Please go ahead.

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

I think you're asking about our generation rigs. Right?

Yeah.

Right now, we are sending the A16 SIP machines to our customers. And I think they're doing the testing phase. We are moving ahead with mass production preparations right now. And I think the mass production will start after the Lunar New Year holiday. And we expect it to begin volume ramp-up by the end of the first quarter. Currently, we don't have any issues or anything to broadcast. On the chip side, the chips are already in mass production. And on the production side, our main focus now is refining the product at a system level. And, also, you know, besides the air-cooled version, we will have the liquid-cooled and immersion-cooled models. So now we are working on these different models so we can better match different customer deployment needs and site conditions. I hope this answers your question. Thank you.

No. 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, the A15, and kind of how you think this can potentially, you know, help keep margins strong moving forward?

I will take this question, Ben. I think currently we observed market price has some influence from the Bitcoin price. So it seems like the industrial machines' profitability seems to be under pressure. But it looks like the home series continued without serious competition in the market. So we can still maintain good profitability. But in Q4, on 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 the home series play a more important role in our category to generate profit. I don't know if I answered your questions, Ben.

No. That was super helpful, and thank you guys for the update.

Thank you, Ben. Thank you.

Thank you. We will take our next question. The question comes from Nick Giles from B. Riley Securities. Please go ahead.

Good morning. This is William Chan speaking on behalf of Nick Giles. Thank you for taking my question, and congratulations on the quarter. It was good to see your heat recovery proof of 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?

Thank you. Hello. Morning. I fully understand the market's interest in the new energy and ESG-related computing products. I think the core value for these objectives is transforming wasted and concentrated energy into measurable, tradable computing power, and even for cash flow. 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, you know, we have been working on this for more than one year. So the scale of each individual project typically ranges from a few megawatts to several tens of megawatts. The true total addressable market largely comes from the number of these points. But we want to caution against overly optimistic calculations like, okay, we have a few megawatts for each site, and there's thousands of points there, and so there's a multi-gigawatt business. We believe that is too optimistic because the business model is still relatively fragmented, and the pace of progress must be steady. So over the past year, we have started systematically screening potential sites for several POC projects, and we have already implemented some of them, collecting initial operational data. Not only the Canada one, there are many others. But moving forward, our focus will be on three key areas. First is the data and methodology standardization. The second is productization and the model. We aim to change the POCs into replaceable modular solutions. And third is replication and expansion. This is a service. I think once we have reached the surface, then we'll continue to expand similar projects faster. I hope I answered your question. Thank you.

Yep. That's very helpful. Thank you, and continue best of luck.

Thank you.

We will take our next question. Your next question comes from the line of Mark Palmer from Benchmark. Please go ahead.

Yes. Good morning. As you think about Canaan's manufacturing footprint, from a long-term standpoint, what would that look like, and where would the company's US manufacturing place, you know, to adjust for the tariff environment, fit into that?

Yeah. I think over the last year, the external environment has been very dynamic. Tariffs move back and forth. Compliance requirements become tighter in many markets, and it's very volatile. In this context, computation is not only about the price and the efficiency. It's also about the compliance, capability, display, and how fast you can adjust. So now the supply chain issue is combined with compliance. But we treat compliance as a baseline. So we keep high standards across sales, delivery, and regional operations. Despite 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 asset exposure in certain regions. And so when the policy or the market trends quickly, their adjustment cost is higher. We do not have, you know, firstly, our self-mining is 100% outside China. And also, we build multiple region production and assembly setups across Mainland China, South Asia, Malaysia, and even in California, North America. So this really gives us resilience and continuity as regions become less predictable. So, you know, I think for your question, North America is our most important market. And last year, we built thousands of machines from our US manufacturing facilities. So this year, we will carefully review the whole supply chain and make it safer for US customers and expand our US manufacturing. Yeah. Thank you.

That's helpful color. Thank you. Thank you.

We will take our next question. Your question comes from Kevin Cassidy from Rosenblatt Securities.

Hi. Yeah. Thanks for taking my questions. I wonder at what point price do your customers break even for Bitcoin mining? I think it had been $90,000. Has that changed?

Thank you, Kevin. I think we have two lines for this breakeven point. One, including the price of the machines, we say it's the all-in payback level. I think it's almost like a $100,000 to $110,000 range for Bitcoin price to stay there. And, you know, the hash price should be like $55 per PH per day. Something between that, that's the 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 is already a sunk cost. So in this kind of scenario, it's quite lower compared to the previous all-in payback level. For various miners, of course, it's different. If we use our competitor's machine as a comparison, if the electricity cost is like 6¢ and we use our competitor's S21+, we see the shutdown price is like $50,000. And if it's S19 XP, it's $66,000. And then look back to ourselves or our mining operation. Our average cost is like 4.3¢ globally. So our shutdown price for the A15 Pro version is like $37,000. When Bitcoin price hits like $37,000, we have to shut down the machine. But, of course, in our mining sites, we do have some older generation machines. So it varies from, you know, 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 12.8 joules power efficiency, and the shutdown price is about $30,000, you know, for Bitcoin price. I think this calculation is based on the latest hash price. It always changes because of the network, the total network hash rate changes, and also the Bitcoin price changes. So I think in the current stage, we have already observed in December, early January, and early February, some of the operations in the network have 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 the short term have not been released to the manufacturers. But in the longer term, when the electricity has already been prepared for mining, they will come back asking for better machines, for the latest generation machines. That's something that happened in the past cycles, no matter bear market or bull market. Kevin, I think this answers your question.

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

Yeah. I'll do this one. Since last year, with our assessment of market cycles, we have maintained a fixed price interest strategy with low stock levels. And we secured, but we still secured critical foundry capacity and supply chain resources in anticipation of market recovery this year. Currently, global foundry capacity is indeed very tight, particularly for advanced nodes, which are seeing surging demand for AI-related sectors. But we secured our position earlier and maintained it because we have long-term partnerships. We're utilizing rolling forecast prepayments and collaborative ramp-up mechanisms. So our access to wafers and key components remains stronger than the industry average. What I can say is, yeah, industry average. So regarding the cost, we cannot disclose specific product unit costs for A16 faces upwards pressure in wafers, packaging, and certain system components compared to A15. I think it's for sure. You know, even the metal is rising in price. So our plan is to offset these costs through yield improvements, testing optimizations, and designing more efficient systems. So, overall, what I can say is we expect the unit cost increase for A16 to remain within a manageable range. Ultimately, we measure competitiveness by our customers' life cycle economics, including power efficiency, returns stability, and delivery certainty. So I think if the market recovers, our cost and our performance can give you an opportunity to have good ASP at that time. Yeah. Thank you.

Great. Okay. Thank you.

Thank you. We will take our next question. Your next question comes from the line of Kevin Dede from H. C. Wainwright. Please go ahead.

Hi, Nangeng and James. Thanks for having me on the call. A couple of things for you. One is just I know you spoke to strategic priorities, but I just like to understand the one-gigawatt facility objective and what that pipeline looks like. Maybe you could add some color there, please.

Yeah. Okay. Hello. Good morning. Yeah. We will share more details when the time comes. But currently, we are quite confident in the gigawatt-level power opportunities, which is based on our results from work at this stage. We have already worked on this for maybe close to one year, I think. Yeah. Second, yes, we believe our goal is to co-locate AI HPC and Bitcoin mining much better. So we are talking about high-quality power resources.

Okay. Thanks, Nangeng. James, you mentioned, I think, a cash outlay of $100 million for wafers and $80 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, you 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.

Yeah. 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 is usually lower than the small orders for the retail miners. I think that's the case. And, also, we prioritized the industrial miners in Q4 instead of the home series. Even the margin side, the home series is better. But we have to make sure our strategic partner, the client, feels satisfied to get our delivery on time. So we tried our best in Q4. And, you know, still a few batches were delayed to early January. But we completed most of the deliveries in Q4. I think that's very helpful to the client. But not helping our financials. It looks like the profitability part is not as good as we expected for Q4. Just like I mentioned, the $38 million is the total expense for Q4. We did some work, which is slightly lower than Q3, 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. It's ongoing. We still, but with a smaller volume of payment, continue that trajectory in Q1 to our wafer partner. I think that's some color I added to this question, Kevin.

Thank you very much, James and Nangeng. Appreciate being on the call.

Thank you, Kevin. Thank you.

Thank you. Once again, if you wish to ask a question, we will take the next question. The question comes from Kevin Dede from H. C. Wainwright. Please go ahead.

Hi again, gentlemen. I figured I'd hop on again given the opportunity to do that. Nangeng, can you talk a little bit about your product development? I understand the 12 joules per terahash target for the A16, but you also mentioned chip development that could push you down to maybe five or six joules 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 chip might be in the market?

Yeah. I think it's a very open question. I think for the A16 series, currently, we achieved 12.8 joules for the ASIC XP with manageable cost rise. So our target is to, when the market recovers, you know, currently because of the deep dive for the Bitcoin price, so the whole market is some kind of freezing for a few weeks maybe. And after the market recovers, the competition comes back, we can have very competitive cost and performance to our competitors and give benefits to our customers. And for the next generation, yes, we have already moved to the next generation development for the chips. But, you know, because we are already at sub-nano process nodes, I think the benefits from the process itself are very tightened now. And, also, the cost for the manufacture of the chips is rising a lot. We are trying different methods to further improve the energy efficiency. But it looks like after, I mean, after the 12 joules stage, when we come to the sub-10 joules product, it's very, very hard to say that the manufacturing cost is still manageable if we are using today's standards. We already observed that our competitors' products have to be priced very high because we know the rough cost for the system. You cannot sell at a loss forever. Right? So, currently, in many different internal meetings, we are continuing to discuss if our target is the best power efficiency, then we may have to 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 our customers. Yeah. So also, we, you know, we also started the big-scale infrastructure in the US. So I think on the operational side, we are not only the equipment provider, we also got involved in operations. Any pains our customers have previously, we will react to ourselves. So we are thinking about this more and more carefully. Yeah. So, currently, I think in conclusion, I think the development for the new system, we will not accelerate, but also it will not delay. It will just go at a very natural progress. We are sure we will have new products this year. And, yeah, and also, you know, we hope at that time, you know, the AI HPC will not relocate 100% of the semiconductor capacity. Yeah. And also, we don't have, like, the DRAM and HBM kind of memories. So sometimes we are in a very good position. Still, we can use the rapid or, you know, sometimes the peak capacity is released from the foundries. We can use this kind of capacity to get one-time deals to fill our inventory. So this is good. We are waiting for this kind of opportunities. Yeah. So, basically, I think the industry is not coming to the end. It's still going at a normal speed. Yeah. This is my personal view. Yeah. Thank you.

Okay. Okay, Nangeng. Yeah. Thank you for the detail there. Does Canaan have a 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.

Yeah. I think our priority is R&D and delivery to our customers first. And because of the current market situation, so we moved the priority to allocate energy resources instead of just putting more rigs on the shelf. So, yeah, the infrastructure will give us the ability to scale it up when the right window opens. Yeah. So, currently, I don't have a fixed number for this year. We have internal goals for electricity infrastructure. Yeah. So, yeah. And after that, if the window opens, we will rapidly ramp up the hash rate. I think controllable energy resources and facilities will give us more opportunities to try different business models and to provide different kinds of products to our customers. Yeah. I hope, you know, the hash rate sales can come through to our mainstream maybe next year. Yeah.

Thank you. 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.