Kingsoft Cloud Holdings Limited (KC) Q2 FY2025 Earnings Call Transcript & Summary

Good day, and thank you for standing by. Welcome to the Kingsoft Cloud Second Quarter 2025 Earnings Conference Call. [Operator Instructions] Please be advised that today's conference is being recorded. I would now like to hand the conference over to your speaker today, Nicole Shan. Please go ahead.

Thank you, operator. Hello, everyone, and thank you for joining us today. Kingsoft Cloud's Second Quarter 2025 earnings release was distributed earlier today and is available on our IR website at ir.ksyun.com, as well as on PRNewswire Services. On the call today from Kingsoft Cloud, we have our Vice Chairman and CEO, Mr. Zou Tao; and CFO, Ms. Li Yi. Mr. Zou will review our business strategies, operations and other company highlights, followed by Ms. Li, who will discuss the financial performance. They will be available to answer your questions during the Q&A session that follows. There will be consecutive interpretation. Our interpretations are for your convenience and reference purpose only. In case of any discrepancy, management's statement in the original language will prevail. Before we begin, I'd like to remind you that this conference call contains forward-looking statements within the meaning of Section 21E of the Securities Exchange Act of 1934 as amended and as defined in the U.S. Private Securities Litigation Reform Act of 1995. These forward-looking statements are based upon management's current expectations and current market and operating conditions and relate to events that involve known or unknown risks, uncertainties and other factors, all of which are difficult to predict and many of which are beyond the company's control, which may cause the company's actual results, performance or achievements to differ materially from those in the forward-looking statements. Further information regarding this and other risks, uncertainties or factors are included in the company's filings with the U.S. SEC. The company does not undertake any obligation to update any forward-looking statements as a result of new information, future events or otherwise, except as required under applicable law. Finally, please note that, unless otherwise stated, all financial figures mentioned during this conference call are denominated in RMB. It's now my pleasure to introduce our Vice Chairman and CEO, Mr. Zou. Please go ahead. Thank you.

[Interpreted] Hello, everyone. Thank you, and welcome all for joining Kingsoft Cloud's Second Quarter 2025 Earnings Call. I am Zou Tao, CEO of Kingsoft Cloud. During the past 3 years, the company firmly implemented high-quality and sustainable development strategy, fully embraced AI opportunities and our business fundamentals have taken on a completely new look. We have not only achieved growth in both revenue and profitability, but also extensively upgraded our IaaS and PaaS cloud service capabilities for the generative AI era. This quarter, our business sustained growth capability was once again verified. The high-speed growth of AI intelligent computing business has driven incremental demand for basic cloud services, further accelerating revenue growth. First, our Q2 revenue reached RMB 2.35 billion, representing a year-over-year growth of 24%, a significant acceleration from previous quarter's 11% year-over-year. Both, public cloud and enterprise cloud achieved year-over-year growth, among which public cloud increased significantly by 32%, reaching RMB 1.63 billion. Second, our embracement of AI continued to unleash favorable momentum. This quarter, AI gross billing reached RMB 728 million, representing a year-over-year increase of over 120% and a quarter-over-quarter growth of 39%, accounting for 45% of public cloud revenue. In other words, over the past 2-plus years, while successfully driving high-quality development in our basic cloud business, we have also built an intelligent computing cloud business of nearly equivalent scale. The rapid development of GenAI itself and the demand for its implementation across diverse industry verticals have lifted the ceiling of cloud services market. We will continue to embrace AI, enhance our technical capabilities, refine our intelligent computing products and be a leading player in the era of GenAI. Third, as the sole strategic cloud platform of the Xiaomi and Kingsoft ecosystem, we firmly grasped the enormous demand of ecosystem clients and pursued symbiotic growth and mutual success with the ecosystem. This quarter, revenue from Xiaomi and Kingsoft ecosystem reached RMB 629 million, up 70% year-over-year, with its contribution to total revenue further increased to 27%. In the first half of 2025, revenue from Xiaomi and Kingsoft ecosystem reached RMB 1.13 billion, accounting for 40% of the total annual cap of related party transactions in 2025. Benefiting from the continued prosperity of the Xiaomi and Kingsoft ecosystem and ever-expanding business opportunities, we are fully confident in further growth of ecological business collaborations in the second half of this year. Now let me walk you through the key business highlights for the second quarter of 2025. In public cloud space, revenue reached RMB 1.63 billion this quarter, representing a year-over-year increase of 32%. The development of intelligent computing cloud and basic cloud has been mutually reinforcing. With cloud consumption growth from both ecosystem's internal and external clients advancing in parallel. In terms of intelligent computing cloud, the solid demand for training computing power services and the gradually growing demand for inference computing power services have laid a solid foundation for the sustained development of intelligent computing cloud. On one hand, the implementation and application of AI across various industries have begun to emerge. Customers such as large language model companies, Internet audio/video services, real-time communications, online travel agencies and gaming have added incremental demand for AI inference. On the other hand, the growth in data volume driven by AI has boosted the growth of basic cloud services and growth in such high-quality basic cloud services has offset the revenue pressure caused by our proactive scaling down of low-margin services. In terms of ecological customers, we deepened our cooperation with Xiaomi, integrating the advantageous resources of both parties, providing long-term, stable and high-performance cloud computing services for Xiaomi. We also ensured the smooth launch of [ Sword Hero State Zero ] by Seasun Games, a subsidiary of Kingsoft by providing products and services such as database and cloud elastic compute. In enterprise cloud space, revenue reached RMB 724 million this quarter, representing a year-over-year increase of 10%. In terms of public services sector, we partnered with Kingsoft Office to take the lead in officially releasing the Kingsoft government AI all-in-one server, increasing investment in AI plus public services scenarios and providing full stack AI capabilities, including intelligent computing services, platform services and large language model services. Through the strong cooperation between Kingsoft Cloud and Kingsoft Office, we have successfully integrated AI technology with practical applications in public services sectors. This provides scalable and replicable solutions for customers' digital transformation. In health care sector, drawing on our industry-leading capabilities in top-level planning, comprehensive data governance, robust technology foundation and AI capability, we're continuously building the information capabilities for regional health care systems and hospitals in the field of digital health. In this quarter, we won the bid for the Changchun Municipal Public Health Information Platform project, facilitating the interconnection and sharing of local health big data. In addition, we're also working on the construction of the data lake project of Zhujiang Hospital of Southern Medical University and the cloud-native hospital information system of Zhongnan Hospital of Wuhan University. In the field of enterprise services, we are committed to implementing AI in multi-industry scenarios and complex business models, truly improving enterprise operational efficiency and achieving process intelligence. This quarter, we took the lead in cooperating with a major state-owned bank to advance the credit report automation project, building a benchmark project in the banking industry. We applied AI capabilities end-to-end in complex core business scenarios, enabling intelligence throughout the entire process, including data collection, analysis, decision-making and report writing, which has significantly improved the work efficiency of bank account managers. In terms of product and technology, we uphold the principle of building success based on technology and innovation, focusing on delivering best-in-class customer experiences across our core product offerings. We set up basic cloud and intelligent computing cloud R&D team separately but collaboratively to improve the AI plus PaaS capabilities on top of our robust public cloud infrastructure. This quarter, we continued to optimize our AI suite. Using cloud container services as the foundation, we provide out-of-the-box cloud-native components such as heterogeneous resource management, AI workload scheduling, intelligent operation and maintenance and resource monitoring, offering full life cycle support for large language model scenarios. We continue to optimize the capabilities of the StartFlow training inference integration platform, providing data sets, simulation service capabilities required by the AI industry as well as full life cycle management of data processing tasks. Our intelligent computing cloud technologies help customers improve model training performance and save unit costs through optimized resource management, data governance, network communication and protection. In addition, we released a new version of Kingsoft Cloud Galaxy Stack, supporting multiple mainstream processor platforms and domestic operating systems and improving its compatibility for private deployment scenarios. Overall, AI is injecting new momentum into cloud computing, while cloud computing in turn, is essential to support rapid model training and adoption. We continue to build and upgrade our intelligent computing cloud resources and at the same time, leverage our expertise to enhance AI implementation into key sectors like public services, financial services, health care and other fields. AI adoption is accelerating in various sectors and its substantial ability to improve productivity, enhance user experience and generate additional revenue streams has been verified. Looking ahead, we firmly believe the market opportunities brought by AI revolution has just begun. We will leverage on the comprehensive capabilities we have accumulated over the past few years by embracing AI, continue to deepen our commitment to the Xiaomi and Kingsoft ecosystem and high-quality external customers and focus on polishing our core products and solution capabilities to create long-term value for our customers, shareholders, employees and other stakeholders. In addition, this quarter, I would like to extend a warm welcome to the company's new CFO, Ms. Li Yi. I will now pass the call to Li Yi to go over our financials for the second quarter 2025. Thank you.

[Foreign Language] Thank you all for joining the call today. It's my pleasure to join Kingsoft Cloud. I'm looking forward to working collaboratively with the team to navigate the challenges, capitalize on opportunities and do my best to contribute to the execution of high-quality and sustainable development strategy. Thanks for the trust of the Board, company and the stakeholders. Now I will walk you through our financial results for the second quarter of 2025. This quarter, our AI strategy continued to be successful, driving business expansion across all products. Total revenues for this quarter were RMB 2,349.2 million, reflecting a 24.2% year-over-year increase. Of this, revenues from public cloud services were RMB 1,625.3 million, up 31.7% from RMB 1,234.5 million in the same quarter last year. This growth was primarily fueled by a surge in AI-related business with gross billings hitting over 120% year-over-year increase to RMB 728.7 million. Revenues from enterprise cloud services reached RMB 723.9 million, up 10.1% from RMB 657.2 million in the same quarter last year, primarily driven by high demand for IT delivery services and steady progress on our external enterprise project delivery. Total cost of revenues was RMB 2,010.4 million, up 27.8% year-over-year, which was mainly due to our investment into infrastructure to support AI business growth. IDC costs increased by 10.3% year-over-year from RMB 728.2 million to RMB 803.1 million this quarter, which was mainly due to our increased purchase catering to our new AI cluster demand. Depreciation and amortization costs increased from RMB 265.9 million in the same period last year to RMB 552 million this quarter, mainly due to depreciation of newly acquired high-performance servers to expand our AI business as well as the regular CPU servers to support the computing and storage demand ordered by AI-related customers as the data set exploding. Solution development and services costs rose by 14.8% year-over-year from RMB 491.1 million to RMB 563.7 million, driven by expansion in solution architecture and delivery personnel to support revenue growth. Fulfillment costs and other costs were RMB 25.8 million and RMB 65.8 million this quarter, respectively. Our adjusted gross profit for this quarter was RMB 350.6 million, increased by 8.4% year-over-year and 7% quarter-over-quarter. It was mainly due to the expansion of our revenue scale and the enlarged contribution from AI business. Adjusted gross margin was 14.9% in this quarter compared with 17% in the second quarter of 2024 and 16.6% last quarter. Our adjusted gross margin has been negatively impacted by the higher cost of server along with the expansion of our AI business and upfront costs incurred for certain customers for future revenue activity as well as the price pressure of certain large-scale clusters. On the expense side, excluding share-based compensation costs, our total adjusted operating expenses were [ RMB 560.7 million ], increased by 1% year-over-year and 31.2% quarter-over-quarter, of which our adjusted research and development expenses were RMB 183.1 million, decreased by 8.5% from same quarter last year. The decrease was mainly due to the decrease of personnel resulting from our strategic adjustment for research team. Adjusted selling and marketing expenses were [ RMB 109.5 million ], decreased by 6.8% year-over-year. Adjusted general and administration expenses was RMB 268.1 million, increased by 12.8% year-over-year due to the increase of credit loss resulting from prepayments made to suppliers related to the procurement of certain servers. Our adjusted operating loss was RMB 166 million, narrowed by 11.7% from RMB 188.5 million in the same period last year. The improvement was mainly due to the share-based compensation adjustment. However, the adjusted operating loss increased compared with RMB 55.8 million from last quarter. It was mainly due to the increase of credit loss caused by the prepayment made to certain service providers. Our non-GAAP EBITDA profit was RMB 406 million, increased by 5.7x from RMB 66 million in the same quarter last year. Our non-GAAP EBITDA margin achieved 17.3% compared to 3.2% in the same quarter last year. It was mainly due to our strong commitment to AI cloud computing development, strategic adjustment of business structure and our strict control over cost and expenses. As of June 30, 2025, our cash and cash equivalents totaled RMB 5,464.1 million, providing a strong liquidity position to support operations and AI investments. The increase was mainly due to our public equity offering and private placement with Kingsoft Corporation as well as the prepayment we received from strategic customer, which will be used to support its further cloud construction. This quarter, our capital expenditures, including those financed by third party reached RMB 1,135 million and right of use assets obtained in exchange for finance lease liabilities were RMB 1,665.8 million. Looking ahead, AI technology has created a wealth of opportunities for cloud computing, not only the computing demand brought by modern training and inferencing, we also help enterprise to adopt AI capabilities into their complex business scenario. Our company as the enabler of AI provides cutting-edge technology and compute resources to all kinds of customers, help them to leverage sophisticated AI models and platforms without need for extensive in-house infrastructure and large amount of capital expenditures, significantly lowering the barriers to entry and accelerating technological break across various sectors. Thank you all.

This concludes our prepared remarks. Thank you for your attention. We are now happy to take your questions. Please answer your question in both Mandarin Chinese and English, if possible. Operator, please go ahead. Thank you.

[Operator Instructions] We will now take the first question from the line of Wen Yu from CLSA.

[Interpreted] I'll translate the question. So the first question is, could management share the outlook and guidance on the revenue outlook for the second half of this year and also the first half of next year? And how is Xiaomi's investment pace in AI and autonomous driving infrastructure? Additionally, what are the AI capacity demand trends in industries like AI [indiscernible] and the others? Do we observe large model vendors [indiscernible] their model iteration and reduce demand for computing consumption and which other industries show strong AI infrastructure demand? And second question is regarding the gross margin. This year, KC has adopted more leasing of compute resources, and this had already impacted gross margin in the second quarter. And looking ahead, do we expect gross margin to continue to decline in the coming quarters as we use more leasing? What is the current proportion of lease capacity in the overall computing resources pool? And what is our target for preferred ratio?

[Interpreted] So let me translate briefly for Mr. Zou's answer to the first question. So generally speaking, since you're asking about the expectation for the second half growth, I would say that the second half revenue growth, we would expect that to be stronger and better than the first half. That's the general holistic revenue top line situation that I would like to share with you. And secondly, since you asked about Xiaomi, what I can say is that we are in the process of delivering an even larger cluster for Xiaomi's computing power demand without further details about the Xiaomi confidentiality concerns. Thirdly, since you asked about the trend, especially in terms of training versus reference, I would say that after the debut of DeepSeek, the different players in the market started to exhibit different patterns in terms of investment. Some of the players continued to invest heavily in the training of models, and that would include Xiaomi as well, right? But some other players actually have, to some extent, decreased some of the investment in computing power demand. However, we have also been seeing some of the other large enterprises since last year continue to have even stronger inference computing power demand for inference. So I guess, generally speaking, it's hard to comment on each player, each customer that we engage with due to customer confidentiality reasons. But I would say that overall speaking, the market demand for AI continue to be very strong. In relation to the question regarding gross profit margin and its future trend, I think this is a very good question. I would like to put that into the context of our growth model since last year. If you recall, since last year, our old model was purely based on self procurement, which comes with a high CapEx level and also comes with a high gearing ratio. We've been asked a lot of questions by the investor community, highlighting to us the potential risk in relation to that model. So since the second half of 2024, given that consideration, we have adjusted and pivoted to some of the new models, which we would call the resource pool model or the profit sharing model, where CapEx level would be relatively lower and also we would benefit from lower gearing ratio. So overall speaking, because of that shift of that procurement model, although there is a slight decrease of GP margin, however, I would say we generally achieved the strategic choice that we made for that changing of procurement model. And therefore, I think it's actually quite a good success. It's a successful result. And since you also asked about the ratio between the self-owned assets versus the profit sharing model, I would say that we didn't -- we haven't disclosed the particular number in that regard. However, so far, I can tell you that the self-owned assets still command the majority of those assets on our balance sheet. In the future, I would say that in addition to the 2 models that we already have, we are already exploring a new model, which we have applied in one of the key customers, which I would call the agent model, which essentially means that we would do on our customers' behalf -- with our help, we would do the procurement, we would do the construction, and we will do the operation on behalf of that customer. So putting this together, we're actually having 3 models and the particular adoption of any one of this model will purely depend on the demand of those particular customers and the overall balance that we would like to achieve in terms of gearing ratio and CapEx and indebtedness level. So I would say that in general, as we continue to run our different models and their combinations in the next few quarters, we will be more -- we'll have a clearer picture as to where the GP margin will stabilize at. But as far as I can see from this point in time, I would say that the GP margin level relatively will stabilize at where we are right now today.

We will now take the next question from the line of Xiaodan Zhang from CICC.

[Interpreted] And my first question is regarding our capital expenditure plans. So could management update on your capital expenditure plan for this year? And what is the expectation for AI computing power that will be ready to use at the year-end? And secondly, the year-over-year revenue growth of industry cloud has reaccelerated from the last quarter. So could you please share some color on the demand and also your delivery pace of the industry cloud clients?

[Foreign Language]

For this year's capital expenditure, including sales procurement and lease purchase model, as we mentioned last quarter, for the total year is around RMB 10 billion. For the first half of the year, actually, we have spent around RMB 5 billion. As Tao Zou mentioned, now we have 3 models. So we don't -- because we have quite a strong cash position as at the 31st of June, so that is why at this time, we will adjust our procurement process and models according to the customers' demand. So we still think for the whole year, the capital expenditure is around RMB 10 billion.

[Interpreted] So allow me to quickly translate. So first of all, I would say that what you observed is correct, the trend for enterprise cloud revenue in Q2 to grow faster than before. Now the first reason I would say is the advent of DeepSeek. To be quite fair, the event of DeepSeek has a very good and very deep impact in terms of the relatively more traditional industries in China. It's like a customer education process where we're seeing a lot of strong demand coming from the public services, from health care, from education, from financial services, et cetera, et cetera. However, we currently still have this pain point, which is we're still not at the position where we are able to provide our customers with a very easy-to-use application. So the final landing or application of the software or AI solution is still not there yet. So the strategy of Kingsoft Cloud, what we internally tell our management is actually we need to refrain the impulse to actually spread out our work and our energy across too much and too many verticals. Rather, we would ask the company, the people to actually focus on a few focus areas where we have relative competitive advantages. And then working on those solutions to achieve the so-called 0 to 1 breakthrough and then further do the one-to-end spreading out and application. So that's what we're doing now internally in terms of the enterprise cloud and its combination with Generative AI. Now you also asked about the trend for the second half. Generally speaking, for enterprise cloud, the delivery peak had always exhibited the seasonality that the second half to be better than the first half. So we -- as we look at the forecast for second half, the growth rate for revenue for the second half of this year, we do expect that to be significantly better and higher than the first half of this year.

We will now take the next question from the line of [indiscernible] from CITIC.

[Interpreted] So I will translate the question. We see that the current chip supply that is undergoing some changes. H20 resumes supply, B330A and other chips will also be sold. But at the same time, issues such as chip security vulnerabilities may also suggest that the next development needs to be more cautious. Do we have -- made strategy adjustments to our chip [indiscernible] ideas such as embracing domestic production?

[Interpreted] Okay. So actually, to your question, we've been giving this a lot of thinking and contemplation from a strategy perspective in 2023 as we started to venture into the AI -- Generative AI computing cloud business because this is the general broader picture under the final U.S. geopolitical conflict, which gave rise to huge uncertainty in the supply chain. So on one hand -- now given that backdrop, on one hand, we embrace the compliant chips that are supplied in China. And on the other hand, we had also been closely monitoring and following domestic firms for supply chain. To be quite fair, the restriction for H20 actually happened on the day of our equity follow-on issuance. And the ensuing lifting of that restriction happened a few months later. However, we're not too surprised by that because given the backdrop, given the conflict nature. So also, we have been in close business cooperation with suppliers for domestic chips in China, starting from one firm to many firms. And actually, a few of them, we have a very deep collaboration with. So in summary, although there have been back and forth in terms of supply chain uncertainty, however, there is no material impact to our capability to supply and satisfy demand of our customers. Now also zooming into our particular situation, right, because the majority of our customers are large customers, the key accounts, the large customers. And combining the strategy that I talked about, so far, our capacity and all the channels that we have built, both for domestic chips and also for overseas chips are sufficient to supply demand. So that is the situation right now in the short term. However, I do think that in the longer term, if there's going to be, for example, like a killer app GenAI application where the inference demand for our customers experienced explosive growth and then the demand from the industry surge significantly, we do think that there's a chance that in the future, the supply would not be able to meet this demand. So in short, in the future, the balance between supply and demand largely depends on the domestic chips capabilities as well as their performance. Personally, I take a relatively conservative view that I think in the future, if demand should surge like that, there's a chance that the domestic chip supply will not be -- will not be able to meet the demand in the market in China, but that's for the longer term.

I would now like to turn the conference back to Nicole Shan for closing remarks.

Thank you. Due to time, we conclude our earnings call today. Thank you once again for joining us today. If you have any further questions, please feel free to contact our team. We look forward to speaking with you again next quarter. Have a nice day. Thank you. Goodbye.

This concludes today's conference call. Thank you for participating. You may now disconnect. [Portions of this transcript that are marked [Interpreted] were spoken by an interpreter present on the live call.]