Canadian Solar Inc. (CSIQ) Q3 2025 Earnings Call November 13, 2025 8:00 AM EST
Company Participants
Wina Huang - Head of Investor Relations
Shawn Qu - Chairman, President & CEO
Yan Zhuang - President of CSI Solar & Director
Ismael Arias - Chief Executive Officer of Recurrent Energy
Xinbo Zhu - Senior VP & CFO
Conference Call Participants
Colin Rusch - Oppenheimer & Co. Inc., Research Division
Philip Shen - ROTH Capital Partners, LLC, Research Division
Brian Lee - Goldman Sachs Group, Inc., Research Division
Alan Lau - Jefferies LLC, Research Division
Presentation
Operator
Ladies and gentlemen, thank you for standing by. Welcome to Canadian Solar's Third Quarter 2025 Earnings Conference Call. My name is Chuck, and I will be your operator for today. [Operator Instructions] As a reminder, this conference is being recorded for replay purposes.
I would now like to turn the conference over to Wina Huang, Head of Investor Relations at Canadian Solar.
Wina Huang
Head of Investor Relations
Thank you, operator, and welcome, everyone, to Canadian Solar's third quarter 2025 conference call. Please note that today's conference call is accompanying slides, which are available on Canadian Solar's Investor Relations website within the Events and Presentations section.
Joining us today are Dr. Shawn Qu, Chairman and CEO; Yan Zhuang, President of Canadian Solar subsidiary, CSI Solar; Ismael Guerrero, Corporate VP and President of Canadian Solar subsidiary, Recurrent Energy; and Xinbo Zhu, Senior VP and CFO. All company executives will participate in the Q&A session after management's formal remarks.
On this call, Shawn will go over some key messages for the quarter. Yan and Ismael will review business highlights for CSI Solar and Recurrent Energy, respectively, and Xinbo will go through the financial results. Shawn will conclude the prepared remarks with the business outlook, after which we will have time for questions.
Before we begin, I would like to remind listeners that management's prepared remarks today as well as their answers to questions will contain forward-looking statements that are subject to risks and uncertainties. The company claims protection under the safe harbor for forward-looking statements that is contained in the Private Securities Litigation Reform Act of 1995. Actual results may differ from management's current expectations.
Any projections of the company's future performance represent management's estimates as of today. Canadian Solar assumes no obligation to update these projections in the future unless otherwise required by applicable law. A more detailed discussion of risks and uncertainties can be found in the company's annual report on Form 20-F filed with the Securities and Exchange Commission.
Management's prepared remarks will be presented within the requirements of SEC Regulation G regarding generally accepted accounting principles or GAAP. Some financial information presented during the call will be provided on both a GAAP and non-GAAP basis. By disclosing certain non-GAAP information, management intends to provide investors with additional information to enable further analysis of the company's performance and underlying trends.
Management uses non-GAAP measures to better assess operating performance and to establish operational goals. Non-GAAP information should not be viewed by investors as a substitute for data prepared in accordance with GAAP.
And now I would like to turn the call over to Canadian Solar's Chairman and CEO, Dr. Shawn Qu. Shawn, please go ahead.
Shawn Qu
Chairman, President & CEO
Thank you, Wina, and thank you for all for joining our third quarter earnings call.
Please turn to Slide 3. In the third quarter, we delivered 5.1 gigawatts of solar modules, in line with our guidance range. In our energy storage business, we achieved a record quarterly shipment of 2.7 gigawatt hours. Total revenue reached USD 1.5 billion, landing at the high end of expectations. Gross margin was 17.2%, exceeding guidance, primarily due to strong contribution from energy storage shipments.
We also achieved a higher share of module deliveries to the profitable North American market. Our solar module factory in Mesquite, Texas, which has now successfully ramped up, contributed meaningfully to both shipment volume and margin. Absent non-recurring expenses for the previous quarter, operating expenses normalized, and we reported net income attributable to shareholders of $9 million or a net loss of $0.07 per diluted share due to the impact of paid in-kind of a preferred shareholder of Recurrent.
Now please turn to Slide 4. Solar industry is at an inflection point. Anti-involution policies in China are gradually taking effect and market conditions have stabilized following the most challenging phase of the solar downturn. A complex macro environment presents both challenges and opportunities.
This year, during the anniversary celebration of Canadian Solar's 24th birthday, I reflected on how we have grown with technology innovation, business model evolution and global diversification. Today's shifting geopolitical landscape allows us to once again differentiate ourselves through a resilient combination of strategy and execution. Most notably, we are making strong progress in our U.S. manufacturing investments.
Phase 1 of our solar cell factory in Indiana is expected to begin production in the first quarter of 2026, while Phase 1 of our lithium battery energy storage factory in Kentucky is on track to start production 2026 year-end. These factories will strengthen our U.S. supply chain, support domestic energy security and reinforce our long-term commitment to the American market.
At the same time, we are planning adjustments to our U.S. business to comply with the One Big Beautiful Bill Act. We are progressing smoothly and remain confident we will be able to successfully position ourselves to continue servicing our U.S. customers.
The rise of AI-driven data center is fueling unprecedented global electricity demand. As I have emphasized in my public speeches over the past 2 years, the most flexible and cost-effective solution for powering data centers is solar-plus-storage. In contrast, traditional energy sources such as natural gas and nuclear power require long construction cycles and have limited scalability.
We are now working closely with multiple data center customers to develop deeply integrated solutions. This requires advanced system engineering where our technical expertise provides a strong competitive advantage.
I'm also pleased to share the significant progress we have made in our emerging business segments. Residential energy storage is on track to become profitable in 2025. We have seen strong growth for our residential energy storage product in Japan, Italy and the U.S., and we are expanding into new markets like Germany and Australia. This marks a major milestone for our energy storage strategy and demonstrates how we are successfully broadening our revenue base beyond utility scale applications.
Recurrent, our solar and energy storage project developer and operator will continue to balance the growth of our operational project fleet to generate recurring cash flow and selective sales of project asset ownership to manage near-term cash flow. Given the current market conditions, I have asked our team to keep the balance a little bit more towards sales of project assets in order to accelerate cash recycling and reduce debt.
With that, I will now turn the call over to Yan, who will provide more details on our CSI Solar business. Yan, please go ahead.
Yan Zhuang
President of CSI Solar & Director
Thank you, Shawn. Please turn to Slide 5. In the third quarter of 2025, module shipments totaled 5.1 gigawatts, in line with expectations. Earlier deliveries to 2 energy storage projects shifted volumes from the fourth quarter into the third. This led to our largest quarter to date with 2.7 gigawatt hours of storage shipments.
Revenue was $1.4 billion and gross margin decreased by 730 basis points to 15%. The sequential decline was driven by margin change in both the solar and storage businesses. In solar, incremental upstream price increases and the underutilization raised unit costs. While module pricing in most global markets remained low in storage, second half margins reflect contracts signed at more normalized levels and the volatile tariff environment drove incremental cost increases.
Without last quarter's impairments and benefiting from internal cost controls, operating expenses decreased sequentially from 15.3% of revenue to 12.3% and we delivered $39 million of operating income.
Please turn to Slide 6 for an update on our e-STORAGE business. In the third quarter, we recognized revenue on 2.7 gigawatt hours of storage solutions. Our deliveries reached countries across North America, Europe, the Asia Pacific and Latin America. As of October 31, our contracted backlog, including long-term service agreements, increased to $3.1 billion, supported by newly signed projects in North America and Europe. We continue to build momentum in our established markets while entering new ones.
In Canada, we signed supply and 20-year long-term service agreements with Aypa Power for the Elora and Hedley projects. Together, they totaled more than 2.1 gigawatt hours and among the largest energy storage facilities under development in Ontario.
Also in Ontario, we contracted to deliver a fully integrated energy storage solution and turnkey EPC services for the 1.6 gigawatt hour Skyview 2 energy storage projects. This marks our largest SolBank delivery to date. Once completed, Skyview 2 will be one of the largest battery storage facilities in nation. As a proud Canadian company, we're honored to help drive our country's clean energy transition.
Across the Atlantic, we just signed a BESS supply agreement and 20-year long-term service agreement in Germany with Kyon Energy, a leading storage developer. As demand expands across both our existing and newly entered markets, we expect to continue scaling our backlog and diversifying its global footprint.
In addition to our established utility-scale storage solutions, we continue to expand our offerings and strengthen our capability in both C&I and residential storage. Notably, the residential storage segment is gaining momentum, have turned profitable this year.
Building on the strong growth we have already achieved in Japan, Italy and U.K., we will be launching our new 3-phase solution to drive further expansion in markets such as Germany. We also plan to enter Australia in the first half of next year. In the U.S., we have successfully introduced the second generation of our residential energy storage solution, which better caters to the needs of the market and is demonstrating strong initial performance.
In the C&I storage segment, where we see promising market growth potential, we continue to refine and diversify our portfolio to better serve emerging opportunities. Though smaller in scale, these segments have proven to be profitable and we expect them to contribute more meaningfully next year.
With that, I will hand the call over to Ismael, who will provide an update on Recurrent Energy, Canadian Solar's global project development business. Ismael, please go ahead.
Ismael Arias
Chief Executive Officer of Recurrent Energy
Thank you, Yan. Please turn to Slide 7. In the third quarter, we generated $102 million in revenue. We monetized over 500 megawatts of projects, including 2 high-margin sales, the battery storage project in Italy and a hybrid project in Australia. Gross margin was 46.1%, a sequential increase of 137 basis points, primarily driven by the contribution of more profitable project sales.
During the quarter, we closed $825 million in construction financing and tax equity for the 600-megawatt hours Desert Bloom Storage and 150-megawatt Papago Solar projects, both parts of our multi-project partnership with Arizona Public Service. These assets are under construction and are expected to begin operations in the first half of 2026.
In the U.S., in addition to what we have in construction, we have already safe harbored 1.5-gigawatt peak of solar and 2.5 gigawatt hours of battery storage projects. By the summer of next year, we expect to have safe harbored at least 3-gigawatt peak of solar and 7 gigawatt hours of battery storage projects, giving us significant visibility over our execution pipeline for the next 4 years.
Until our IPP business scales further, near-term profitability will continue to depend primarily on global project sales. As maintaining financial discipline remains our top priority, we will balance the growth of our operating portfolio and project assets with selective project ownership sales to prudently manage cash flow and debt levels. Looking ahead to 2026, we expect to increase the level of project ownership sales to enhance cash recycling and reduce leverage.
Now for an update on our pipeline, please turn to Slide 8. As of September 30, we have interconnection rights for approximately 8 gigawatts of solar and 15 gigawatt hours of storage globally, excluding operating projects. Our total development pipeline now includes 25 gigawatts of solar and 81 gigawatt hours of storage capacity. The reduction in solar pipeline reflects a natural rebalancing. Some projects progress into more advanced stages, while others were removed.
At our current scale, our focus is increasingly on executing our high-quality pipeline rather than expanding it. For example, in the U.K., we recently received government approval for our Tillbridge Solar and battery storage projects in Lincolnshire, U.K. This project is planned to be an 800-megawatt PV plus 1,000 megawatt hour BESS project, making it the largest co-located project in the U.K. to date. We are proud that Tillbridge will connect to the grid through a substation that was previously used by a decommissioned coal plant, continuing to support the U.K.'s decarbonization goals while providing reliable and sustainable energy to the communities it will serve.
Over time, energy storage continues to emerge as a key growth driver. Not only are battery energy storage systems becoming increasingly cost effective, but they are also profoundly reshaping energy markets from grid stabilization and peak shaving to enabling renewables to integrate at scale. Notably, data centers are now placing ever greater demands on power infrastructure, requiring round-the-clock reliability and often clean energy integration. In response, the opportunity set for longer duration, higher specification BESS is expanding rapidly.
We have started to deep our toes into the data centers business through regional JVs with data center experts, mainly in Spain and the U.S. We see significant synergies with our core expertise as land acquisitions, interconnection processes, permitting and community engagement are 4 of our core competencies that are crucial to the successful and timely deployment of data centers.
Furthermore, powering data centers with clean and reliable electrons is one of the key bottlenecks to data center development, where we have significant expertise to bring to the table. In Spain, we already have 112 megawatts of projects with interconnections and land secured in Barcelona, Bilbao and Madrid, plus an additional 40 megawatts with interconnections in Madrid waiting to secure land.
Finally, our Operations and Management or O&M business also continues to grow healthily. This quarter, we earned 2 internationally recognized certifications from TUV Rheinland, ISO 9001:2015 and ISO 45001:2018. These certifications affirm that our power services meet globally recognized standards for quality and workplace safety. Today, Recurrent has over 14 gigawatts of solar and storage projects under O&M contracts across 11 countries.
Now I will hand the call to Xinbo to review our financial results. Xinbo, please go ahead.
Xinbo Zhu
Senior VP & CFO
Thank you, Ismael. Please turn to Slide 9. In the third quarter, we delivered 5.1 gigawatts of solar modules and 2.7 gigawatt hours of energy storage systems. With contributions from accelerated storage shipments, total revenue reached $1.5 billion. Gross margin was 17.2%. The sequential decline primarily reflected the absence of one-time benefits recorded in the second quarter and the normalizing margins in both solar and storage manufacturing businesses.
Operating expenses decreased sequentially to $222 million, reflecting lower shipping costs from reduced module volumes and ongoing internal cost reductions. Net interest expense declined to $29 million, driven by higher interest income. We recorded a net foreign exchange loss of $17 million, primarily driven by the appreciation of Chinese yuan.
Net income attributable to shareholders was $9 million or a net loss of $0.07 per diluted share. This result included a positive $35 million HLBV impact equivalent to $0.51 per share from tax equity arrangements tied to certain U.S. projects. The $0.20 per diluted share preferred dividend impact brought the total diluted loss per share to shareholders to $0.07.
Please turn to Slide 10 for cash flow and the balance sheet. Net cash used in operating activities was $112 million compared with an inflow of $189 million in the second quarter. The difference was primarily driven by change in working capital, notably a decrease in inventories during the prior quarter.
Total assets grew to $15.2 billion, with project assets rising to $1.9 billion. Solar power and battery energy storage systems remained steady at $2 billion as we paced the construction activity to manage leverage at the group level.
Capital expenditures totaled $265 million, primarily related to U.S. manufacturing investments and the existing capacity expansions. This implies a larger CapEx outlay in the fourth quarter and we expect to end the year slightly below our full year guidance of $1.2 billion.
Looking ahead to 2026, we continue to refine CapEx plans amid an uncertain policy environment, but currently expect spending to remain at levels similar to this year. Most investments will continue to target the U.S. market. Total debt increased incrementally to $6.4 billion, mainly due to new borrowings tied to project development assets. We closed the quarter with a cash position of $2.2 billion.
Now let me turn the call back to Shawn, who will conclude with our guidance and business outlook. Shawn, please go ahead.
Shawn Qu
Chairman, President & CEO
Thank you, Xinbo. Please turn to Slide 11. For the fourth quarter of 2025, we expect module shipments to be in the range of 4.6 to 4.8 gigawatts as we continue to maintain disciplined volume management. For our energy storage business, we expect shipments between 2.1 to 2.3 gigawatt hours, which includes approximately 600 megawatt hours delivered to our own projects. This guidance reflects the shift of certain volumes from the fourth quarter into the third.
With Recurrent delivering its largest quarter of product sales this year, we project fourth quarter revenue to range between $1.3 billion to $1.5 billion. We expect gross margin to be between 14% to 16%. For the full year of 2026, we project total module shipments of 25 to 30 gigawatts, including approximately 1 gigawatt to our own projects.
Energy storage shipments are expected to range between 14 to 17 gigawatt hours. We will continue to focus on profitable solar market and drive growth in our storage business. While we will continue to develop solar and energy storage projects, financial prudence remains our top priority. Accordingly, Recurrent Energy will increase project ownership sales in 2026 to recycle more capital and manage the overall debt level.
With that, I would now like to open the floor for questions. Operator?
Question-and-Answer Session
Operator
[Operator Instructions] And our first question for today will come from Colin Rusch with Oppenheimer.
Colin Rusch
Oppenheimer & Co. Inc., Research Division
Congratulations on all the progress. On the project sales, can you talk a little bit about the strategy of timing and leverage that you guys are going to deploy in these sales? You obviously have a great land position, nice interconnection queues and certainly potentially can leverage some of those positions into data center deals.
And also you potentially can monetize these things earlier in the process and generate a little bit better returns in select areas. So I just want to get a better sense of where you're coming out in terms of timing and kind of relationships that are going to come out of some of these sales.
Shawn Qu
Chairman, President & CEO
Yes. Colin, we are still working on the 2026 AOP. So I don't have the quarter-by-quarter Recurrent project sales number in my hand right now. We target to get that down by February. So when we talk in March, in the March earnings call, we'll give you more detail.
However, we have enough COD operational project to sell. So we don't have to sell project early. When I say project early -- sell project early, I assume you mean sometimes sell at NTP or even before NTP. But before NTP, you don't get the value, right? You leave too much money on the table. So if we can sell after COD, we can not only get the value of the project development, but also the project financing because Canadian Solar, especially Recurrent, is also an expert in the tax equity financing deal in the U.S. market and there's a value there.
So we do have enough project. We have a budget like roughly how many projects we will -- how many megawatts of project ownership we will sell each year. But for next year, I guess, we have enough COD project to sell. And that's for U.S., but we also sell projects in other markets, for example, in LatAm and also in Australia.
Now Ismael, you have anything to add?
Ismael Arias
Chief Executive Officer of Recurrent Energy
Just say, it's great, Shawn. Thank you, Colin. Nice to talk to you. Colin, we have a very strong pipeline and very mature. So we are seeing good opportunities to sell with good margins and we are likely going to take them. That's the overall underlying reasons.
Colin Rusch
Oppenheimer & Co. Inc., Research Division
Okay. Perfect, guys. And then thinking about the battery manufacturing and the supply chain, can you talk a little bit about the maturity of your relationships with suppliers to deliver input materials into the U.S. Obviously, 70% of the supply chain is in China and the vast majority is still in Asia. And so shifting things into North America is a pretty substantial effort. I just want to get a sense of how easily that's coming along for you guys? And any sort of risk that we should be thinking about as you start to ramp up that capacity?
Shawn Qu
Chairman, President & CEO
Actually, there are a lot of supply chain also supplier outside China these days. So we have good selection, good choice for both solar and for energy storage. So we will -- as you know, the OBBBA have some requirement of the material -- non-material assistance level for both storage and for solar. And there's also the domestic content booster, right, 10% booster for both the energy storage and solar. And we have calculated that. So just by those percentage requirement, we think we will be able to meet those requirements in 2026, no problem.
I think 2027, the number will go up 5% also. I think it's roughly 5% each year and we should be able to manage that stack. So we'll be able to meet the OBBBA requirement. And also if we do -- if we make both cell and module in U.S., we will be able to meet the domestic content rule as well. As I said, we will start production of our own solar cell in U.S. by March. So throughout Q2, we will ramp up. So by second half of next year, we should have reasonable volumes already. And those volume will go -- will come with the domestic content, the 10% domestic content boost.
And for solar -- and for energy storage, our plan is to start the battery cell and pack manufacturing in U.S. at the same time. So I said in the -- in my speech that we expect to start production in December. So in 2027, we will have -- we'll be able to provide the energy storage project, which also mee the domestic boost requirement -- domestic content requirement to let our customers to enjoy the 10% ITC boost.
Operator
The next question will come from Philip Shen with ROTH Capital Partners.
Philip Shen
ROTH Capital Partners, LLC, Research Division
First one is on margins. I think your A-share subsidy reported a 7% gross margin in Q3, but you guys reported a 17% gross margin today. So I was wondering if you could help us bridge that gap.
Shawn Qu
Chairman, President & CEO
I don't think we reported 7, do we?
Yan Zhuang
President of CSI Solar & Director
17.
Shawn Qu
Chairman, President & CEO
17% is for CSIQ, together CSIQ, right? And CSI Solar have a different mix.
Yan Zhuang
President of CSI Solar & Director
Yes. Ismael just commented that the project sales in Q3 were with 46% gross margin.
Philip Shen
ROTH Capital Partners, LLC, Research Division
Okay. So it's the project business that really supported and offset the manufacturing 7% gross margin. Is that right?
Shawn Qu
Chairman, President & CEO
Actually, solar may be low, but solar plus the energy storage.
Yan Zhuang
President of CSI Solar & Director
15% for all the manufacturing.
Shawn Qu
Chairman, President & CEO
For all the manufacturing, the gross margin in Q3 is over 15%. Now if it's only module, it's low, it's below 10% because there are some where we don't have much margin.
Philip Shen
ROTH Capital Partners, LLC, Research Division
Okay. Moving on to the next question. As it relates to your 2026 guide, you gave us some color there, which is great, and you continue to talk about the ramping of U.S. manufacturing. But how -- can you give us color on how you're able to do that even though there's still substantial FEOC risk as it relates to either ownership or just meeting the OBBBA FEOC requirements could be challenging.
Shawn Qu
Chairman, President & CEO
Yes. Philip, I answered this question in the last earnings call. Philip, we believe we can meet the requirement, the OBBBA requirement by doing certain adjustment.
Philip Shen
ROTH Capital Partners, LLC, Research Division
Okay. And then as it relates to the AD/CVD reserve or with the auction case, there could be meaningful retroactive duties. And I was wondering, can you quantify how much exposure that might be? And as a result, do you think you might need to reserve for that situation on the balance sheet?
Shawn Qu
Chairman, President & CEO
Yes, it could be. I would say also could not be, right? So the court process is still moving along and there will be a long -- there will be quite a while before there's a final decision if they go to the appeal court. And we have discussed this with our external lawyers and the -- also the audit firms. We don't think we need to book any reserve at this moment.
Operator
The next question will come from Brian Lee with Goldman Sachs.
Brian Lee
Goldman Sachs Group, Inc., Research Division
Maybe just a follow-up to Phil's question. I know you guys are wanting to see the AD/CVD process through litigation and the case is still pretty early on. But in the event that you did have to accrue a liability or reserve some amount of funds for a potential negative decision, can you help to kind of quantify the range?
I guess, back of the envelope math suggests it could be well over $1 billion if we estimate your U.S. shipments over the past couple of years. I guess, first, is that the right way to think about it? And then second, how would you -- again, just playing devil's advocate, hypothetically, if you had to do that, what would be your sort of funding strategy to finance that amount just given the cash burn and the high degree of net debt you have right now?
Shawn Qu
Chairman, President & CEO
Well, Ben, I guess you are also talking about the auction case. And as I said, when I answered Philip's question, we don't think that we have to make reserve. Therefore, there's no like -- no, I just -- I don't have to do any backup envelope at this moment. This is what my lawyer told me. This is what my auditing firm told me. So I don't want to speculate here.
Why don't you ask the petitioner to speculate how much money they can get or how much money they will be able to get for U.S. government?
Brian Lee
Goldman Sachs Group, Inc., Research Division
Yes. No, I mean, I think there are published research around potential value of the claim here. I don't know how accurate they are, but I do think there is a published number, which again counts into the billions of dollars. But yes, I'll take that offline.
I guess maybe just bigger picture question. We're all just trying to gather more detail. We know there's no finite answer, but it'd just be helpful if you could elaborate, let's say, on the FEOC question as well. You're obviously telling your customers -- your actions you contemplate taking to make sure you're FEOC compliant. Is there any insight into those conversations you can provide to give the financial community the same level of like confidence around what steps you may be taking to make sure that your U.S. manufacturing investments are going to be justified?
Shawn Qu
Chairman, President & CEO
Well, OBBBA has very simple and clear rules, which says, a big picture, it will require 75% from -- not from the FEOC and no more than 25% from the FEOC if there's 2 partners, right? If there's only like 1 plus 1 partner. If there are 2 partners, 2 shareholders from the FEOC countries that the 2 together should take no more than 40%. So there are very clear rules there.
So when I said we will make adjustment to meet the OBBBA, so I think it's quite clear. It's something like a 5-year -- like Grade 5 student or no. As long as you are structured yet with these percentage numbers, then you are OBBBA compliant. What else do I have to tell you?
Brian Lee
Goldman Sachs Group, Inc., Research Division
Okay. No, that's helpful color. And then last question for me, I'll pass it on, is on the asset sales. It sounds like that's definitely going to ramp up in '26, which is a bit of a reverse from the past couple of years as you've been moving towards this IPP model. It sounds like it's focused on cash generation and delevering. Can you quantify kind of what volume megawatts, megawatt hours you anticipate monetizing through asset sales as opposed to keeping on the balance sheet for '26? And what kind of delevering potential that might result in for the balance sheet next year?
Shawn Qu
Chairman, President & CEO
Well, as I said, we'll continue to build IPP portfolio. However, given the current market condition, we are going to keep the balance a little bit to be a little bit more cautious. And also, as I said when I answered Colin's question, I haven't brought my -- let my Board approve my 2026 AOP annual operation plan yet. So I will give you more details in March, because typically, our Board approved the AOP in February.
So what I can say now is that given the current market situation, we are going to be a little bit more cautious. And I also said that we have enough operational projects, high-quality projects, which we can cash in. But see, every year, we always sell some projects. And Ismael mentioned that we sold some high gross margin projects in Q3 that helped to boost our gross margin in Q3, right, overall gross margin. So I don't have the number yet, but I will let you know. What I can let you know now is that we will be a little bit more cautious. So we are going to recycle more cash.
Operator
The next question will come from Alan Lau with Jefferies.
Alan Lau
Jefferies LLC, Research Division
This is Alan from Jefferies. I would like to know there's a lot of questions on U.S. projects already. So I would like to have a more overview on the market demand on 2026. What do you think the U.S. installation on solar and energy storage separately?
Shawn Qu
Chairman, President & CEO
Okay. I will ask Yan to share his thought.
Yan Zhuang
President of CSI Solar & Director
So you're asking for the installation demand in the U.S. in 2026, right, on both storage and solar?
Alan Lau
Jefferies LLC, Research Division
Yes.
Yan Zhuang
President of CSI Solar & Director
Yes. I think -- so the demand is there, right? Also, the OBBB compliant -- the safe harbor actually made the storage pipelines there. So I think for 2026, the storage project will be there. And the solar as well, the safe harbor also helped to actually to preserve a lot of demand. But on the solar side, I think the cell supply can be a bottleneck for the total demand.
So although we have a good solution, but does not mean everybody has that. So I think -- I hope U.S. will be -- continue to maintain the similar level compared to this year. That's what I hope. But I do not see any significant growth. But energy storage, I think next year, U.S. will continue to be strong. That's my view.
Shawn, if you...
Alan Lau
Jefferies LLC, Research Division
Yes. So I think investors' focus are concentrated or overwhelmingly concentrated on ESS. So I would like to know what type of growth rate you are looking at? Like is it like 20%, 30% growth or 50% or even in China, I think people are talking about even more aggressive growth rates? And then within that growth, how much do you think is coming from [ AIDC ] demand?
Yan Zhuang
President of CSI Solar & Director
Well, so you're talking about the growth globally or U.S.-China? Sorry.
Alan Lau
Jefferies LLC, Research Division
Mainly U.S.
Yan Zhuang
President of CSI Solar & Director
I think that the data center -- yes, worldwide, you talked about data center worldwide, more than half of the data centers built in the U.S. But I think we see a very strong future demand in our portfolio on data center-related storage demand. But I think in terms of start installation construction, next year is not yet. It's not yet. It's going to be -- you got to wait for a little longer time. So for next year, the storage growth will still come from the harbor projects. So this is a regular storage -- those regular storage projects.
Solar, as I said, I'm expecting flat. That's my hope. But storage, you're talking about growth rate, I don't have the number, but you can check the industry reports. They vary a lot, the industry reports. On average, I think there's a growth. I don't know, it's like 20% growth? Yes. I remember I saw some reports number.
Alan Lau
Jefferies LLC, Research Division
I see. So for demand -- ESS demand related to AIDC, you think probably it's after 2026, right? And then like what type of installation you think will be more relevant? Like is it like 2 to 4 hours of system that is for clipping the peak demand or you are seeing even longer hours acting as some off-grid solution or like the main power supply for the AIDC? Like what type of backlog do you see or request from clients are you seeing?
Yan Zhuang
President of CSI Solar & Director
I think for regular storage, conventional storage projects, you're talking about mostly in the U.S., it's actually load shift -- peak shift. So it's rather like 3, 4 hours, around 4 hours. But for data center, to begin with, I think my knowledge, okay, it's more like 2, 3 hours. It's mainly for smoothing out the load. That's what I -- our study shows.
Of course, longer term -- for longer term, the storage project for data centers will progress into longer and longer period of storage, but the cost is also going up. So the challenge is how do we control costs while increasing the length, the duration. But to begin with, the most important application for data center storage is smooth out the load, smooth out the curve. So that's the most important starting point.
Alan Lau
Jefferies LLC, Research Division
I see. So just to confirm, it's more like there are some rules in ERCOT, maybe like the Senate Bill 6, et cetera, requiring more stability on the load. So the demand you are seeing, at least for now as a start is to cope with that request from the grid, right, instead of having long duration ESS for supplying as the main power supply of AIDC. Is this understanding correct?
Yan Zhuang
President of CSI Solar & Director
Well, I think, as I said, right, for longer term, it will progress, right? But to begin with, I told you, it's more like smoothing up the load, so to stabilizing the supply. And so that's my answer.
Alan Lau
Jefferies LLC, Research Division
That's good. That's good. And then finally, I would like to ask on how much of the 14 to 17 gigawatt hours of shipment is going to be in the U.S.?
Yan Zhuang
President of CSI Solar & Director
Actually, we have well diversified our portfolio, our backlog. So I would say around 2/3 will be outside of the U.S. out of the total guided volume next year.
Alan Lau
Jefferies LLC, Research Division
I see. I see. That's pretty diversified...
Yan Zhuang
President of CSI Solar & Director
Yes, but also a small -- it's small in China and mostly it is between -- outside of China, outside of the U.S. So that's the kind of distribution.
Operator
This concludes our question-and-answer session. I would like to turn the conference back over to management for any closing remarks. Please go ahead.
Shawn Qu
Chairman, President & CEO
Well, thank you very much for everyone to come to our call. And also thanks for your continued support. And if you have any questions or would like to sign up a call, please contact our Investor Relations team. Take care, and have a great day.
Operator
The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.