Kingsoft Cloud Q2 2026 Earnings Live Stream
[AI Key Takeaways]
Financial Performance
- Total revenue reached RMB 3.07 billion, setting a new quarterly historical high, with a year-on-year increase of 31%
- AI cloud billing revenue reached RMB 1.33 billion, up 82% year-on-year, accounting for 56% of public cloud revenue
- Adjusted gross margin rose to 15.4%, an increase of 2.4 percentage points quarter-on-quarter
- Operating profit turned positive for the first time, with the adjusted operating profit margin reaching 4%, a historical high
Business Progress
- Mars business revenue showed strong growth, with Q2 Mars revenue increasing 12-fold compared to Q1
- The Xingliu platform has deployed 120 models and onboarded over 230 clients
- Officially launched the Xingyuan Agent Kit platform, providing a comprehensive infrastructure suite covering security sandboxes, knowledge and memory management, and evaluation governance.
- Signed an agreement with the Nanjing Communications Administration of the Yangtze River to build Jianghai Cloud, and reached a strategic cooperation agreement with the Wuhan Municipal Data Bureau.
Next Quarter Guidance
- Maintained the full-year capital expenditure baseline expectation at RMB 15 billion.
- Capital expenditures for the first half of the year (including capitalized asset leasing arrangements) reached RMB 6.2 billion, accounting for over 75% of last year's total capital expenditures.
opportunity
- The unprecedented prosperity of the open-source large model ecosystem has brought huge development opportunities to Chinese cloud vendors.
- AI is being implemented across various industries, developing through a combination of Model-as-a-Service (MaaS), intelligent service systems, and FDE.
- Shareholders approved an increase in the cap on related-party revenue from Xiaomi, reaching a combined total of RMB 10 billion for 2026 and 2027, a 39% increase from the previous cap.
Risks
- The Mars business is influenced by various factors such as market price competition volatility and new model substitution, resulting in variable profit margins.
- Supply chain tightness is expected to persist for an extended period, becoming the new normal.
[AI Conference Transcript]
Operator
Good morning, ladies and gentlemen, and thank you for standing by for Kings of Cloud's Second Quarter 2026 Earnings Conference Call. All participants are currently in listen only mode. Following management's prepared remarks, we will open the call for questions. Please note that today's call is being recorded.
I will now turn the call over to Mr. Jackie Zor, Senior Director of Capital Markets at Kings of Cloud. Jackie, please go ahead.
Jackie Zor
Thank you, operator. Hello, everyone, and thank you for joining us today. Kings of Cloud second quarter 2026 earnings release was issued earlier today and is available on our IR website and through PR Newswire.
Joining us today are Mr. Zhoutao, Chairman and CEO, Miss Lii, CFO, Mr. Liu Tao, Senior Vice President, Mr. Ken Kai Yan, Senior Vice President, Miss Yujin, Vice President, Mr. Joe Raylone, Associate Vice President and Mr. Clark, Ken Sport, Secretary and Associate Vice President.
Mr. So will discuss our business performance and key developments, followed by Miss Lee with a review of our financial results. Management will then take your questions. Consecutive interpretation will be provided for convenience and for reference only. In the event of any discrepancy, management statements in the original language will prevail.
Before we begin, I would like to remind you that today's call contains forward-looking statements made under the Safe Harbor provisions of US Private Secretary, Securities Litigation Reform Act of Magnet VIBE. These statements involve risks and uncertainties and actual results may differ materially from those expressed or implied by the forward-looking statements.
Additional information concerning factors that would cause actual results to differ materially is included in the company's volume with the USSBC. The company undertakes no obligation to update any forward-looking statements except as required by applicable law. Unless otherwise stated, all financial figures discussed on today's call are denominated in renminbi.
With that, it is my pleasure to turn the call over to our Chairman and CEO, Mr. Zhou. Mr. Zhou, please go ahead.
Mr. Zhou
Hello everyone, and welcome to Kingsoft Cloud's second quarter 2026 earnings conference call. I am Zhou Tao, CEO of Kingsoft Cloud. This quarter, we have witnessed further evolution in the landscape of AI cloud services. The unprecedented prosperity of the open-source large model ecosystem has created significant growth opportunities for Chinese cloud providers. Our long-held conviction that AI will be implemented across various industries is now materializing through a combination of Model-as-a-Service (MaaS), intelligent system services, and FDE.
In the face of profound industry transformation, Kingsoft Cloud remains committed to its technology-driven foundation and high-quality sustainable development strategy. We have comprehensively advanced our AI cloud services, Mars business, and FD business, achieving encouraging results.
First, AI business continues to drive rapid revenue growth for the company. Total revenue for the quarter reached RMB 3.07 billion, a record high for a single quarter, representing a 31% year-over-year increase. Among this, AI cloud billings reached RMB 1.33 billion, up 82% year-over-year, accounting for 56% of public cloud revenue. The Mars business saw robust growth, with Q2 revenue increasing twelvefold compared to Q1.
Second, profitability improved significantly. The adjusted gross margin rose to 15.4%, an increase of 2.4 percentage points quarter-over-quarter. Operating profit turned positive for the first time, with the adjusted operating margin reaching a record high of 4%. These are gratifying results achieved by seizing the AI wave, improving revenue quality, and implementing cost reduction and efficiency enhancement measures in parallel.
Third, customer structure continues to optimize, accelerating the realization of business opportunities both within and outside our ecosystem. Within the ecosystem, revenue from Xiaomi and Kingsoft ecosystems reached RMB 810 million this quarter, a 28% year-over-year increase, accounting for 26% of total revenue. Outside the ecosystem, revenue from the top five customers grew 51% year-over-year. Our AI cloud business now covers a wide range of industries, including cutting-edge internet AI labs, embodied intelligence, autonomous driving, AI for science, fintech, gaming, and audio-video services.
A diversified customer structure and business layout not only drive continuous growth in revenue scale but also enable us to allocate computing resources more flexibly, enhancing our bargaining power and risk resistance capabilities.
Hello, everyone, and welcome to Kingsoft Cloud's second quarter 2026 earnings call. I am Sotal, CEO of Kingsoft Cloud. This quarter, we witnessed further evolution in the AI cloud market, with the rapid growth of the open-source model ecosystem creating significant opportunities for neutral cloud service providers.
At the same time, our long-held vision of bringing AI to every industry is becoming a reality through a combination of Model-as-a-Service, Agent-as-a-Service, and FDE services. Against this backdrop, Kingsoft Cloud remains committed to technology leadership and high-quality sustainable growth. We are accelerating the development of our AI cloud Mars and SDE businesses with encouraging progress.
First, AI continues to drive strong revenue growth. Total revenue reached a record RMB 3.07 billion, up 31% year-over-year. AI cloud billings increased 82% to RMB 1.33 billion and accounted for 56% of public cloud revenue. Mars revenue also grew strongly, with Q2 revenue up more than 12 times from the Q1 level.
Second, profitability improved significantly. Adjusted gross margin rose to 15.4%, up 2.4 percentage points quarter-over-quarter. Operating profit turned positive for the first time, with adjusted operating margin reaching a record high of 4.0%. This reflects our continued efforts to capture AI opportunities, improve revenue quality, and drive greater operating efficiency.
Third, our customer mix continued to improve with stronger momentum both within and outside our ecosystem. Revenue from the Xiaomi and Kingsoft ecosystems reached RMB 810 million, up 28% year-over-year and accounting for 26% of total revenue. Revenue from our top five non-ecosystem customers grew 51%.
Our AI cloud business now serves a broad range of sectors, including Internet Services, Frontier AI Labs, Embodied AI, autonomous driving, AI for science, fintech, gaming and online video, to name a few. This diversified customer base supports continued growth while allowing us to allocate computing resources more flexibly and strengthen our pricing power and business resilience.
Mr. Zhou
Next, I will provide a detailed overview of our business progress in the second quarter of 2026. In the public cloud segment, revenue reached RMB 2.36 billion this quarter, representing a substantial year-over-year increase of 45%. Firstly, Xiaomi's AI capabilities are comprehensively empowering its 'human-car-home' ecosystem. Meanwhile, WPS AI continues to advance its operations. As the exclusive strategic cloud platform for the Xiaomi and Kingsoft ecosystem, we see unprecedented growth potential for cloud services driven by the AI era.
In June 2026, the shareholders' meeting formally approved an increase in the annual cap for related-party transactions with Xiaomi. The combined caps for 2026 and 2027 now total RMB 10 billion, representing a 39% increase from the previous limit. In the first half of the year, public cloud revenue from Xiaomi and Kingsoft grew by 54% year over year. Secondly, the service capabilities of our Star Flow platform have been further strengthened.
The Star Flow model API services continue to be refined, offering multi-model support and enterprise-grade integration capabilities. Currently, the Star Flow platform has deployed and launched 120 models, ensuring that major new models are listed synchronously with their market release. We have onboarded more than 230 clients. Thirdly, we have deepened cooperation with leading customers in emerging sectors. We completed the delivery of large-scale computing clusters for top-tier embodied AI and autonomous driving clients, providing stable support for their efficient model iteration. We have also established deep partnerships with leading clients in the 'AI for Science' sector to facilitate the rapid expansion of their new businesses.
Now let me walk you through our business progress. In the second quarter of 2026. In public cloud, revenue reached RMB 2.36 billion, up 45% year over year. First, Xiaomi continues to expand AI across its human car home ecosystem, while WPSAI continues to advance as the only strategic cloud platform for the Xiaomi and Kingsop ecosystem. We see substantial AI driven growth opportunities.
In June, our shareholders approved a further increase in the annual caps to connect the transactions with Xiaomi. The combined caps for 2026 and 2027 now total RMB ten billion, 39% higher than before the adjustment. In the first half, public cloud revenue from Xiaomi and Tinsoft grew 54% year over year.
Second, we further strengthen the mass capabilities of our Star Flow platform. Star Flow now supports 120 models with major new models launched on the platform In Sync with their market release and serves more than 230 enterprise customers. Third, with different cooperation with leading customers in emerging sectors, we delivered large scale computing clusters to leading embodied AI and autonomous driving customers supporting rapid model iteration and expanded our corporation with the leading AI for science customer to support the growth of this new business.
Mr. Zhou
In the industry cloud segment, revenue reached RMB 710 million this quarter. In the public services sector, we signed an agreement with the Nanjing Communications Administration of the Yangtze River to build the 'Jianghai Cloud,' creating a dedicated digital infrastructure tailored to the needs of Yangtze River shipping. We also established a strategic partnership with the Wuhan Data Bureau and Wuhan Cloud, collaborating on areas such as computing power grid integration, smart computing applications for digital government, and industrial ecosystem development, thereby solidifying the digital foundation for Wuhan's government and industrial ecosystem.
In the digital health sector, we are leading a special project under the National Key R&D Program on the integration of biology and information technology to build a cloud-based integrated virtual surgery platform. The results have been applied in more than 30 hospitals nationwide, setting an industry benchmark for national-level scientific research collaborations in medicine, engineering, and information technology led by cloud providers. In the enterprise services sector, we have established deep cooperation with Yunshang Gansu to jointly build and operate the Gansu Provincial Government Cloud Platform under an integrated investment, construction, and operation model.
In Enterprise Cloud, revenue reached RMB 710 million. In public services, we signed an agreement with the Yangtze River Communications Administration to build Danghai Cloud, a dedicated digital infrastructure platform for Yangtze River shipping. We also formed a strategic partnership with the Wuhan Municipal Data Bureau and Wuhan Cloud across computing, resource interconnection, digital governance, intelligent computing, applications, and ecosystem development.
In Digital Health, we are leading a project under the National Key R&D Program on Biology and Information Integration to develop a cloud-based virtual surgery platform, which has been deployed in more than 30 hospitals nationwide. In Enterprise Services, we deepened our cooperation with Yunshang Gansu to jointly build and operate the Gansu Provincial Public Services Cloud under an integrated investment, construction, and operations model.
Mr. Zhou
In terms of product technology, we are closely aligning with the needs of computing implementation and AI application upgrades to enhance our full-stack AI capabilities. This quarter, the Xingliu Mars platform continued to optimize model deployment for high-concurrency inference requirements, significantly improving the throughput of several core models and supporting refined management based on permission usage and model dimensions.
We officially launched the Xingyuan Agent Kit platform, providing a comprehensive foundation covering secure sandboxes, knowledge and memory management, and evaluation governance, to help enterprises quickly build production-grade AI agent applications. Meanwhile, we are integrating AI capabilities into our peripheral cloud products, transforming components such as database storage into platform-type products that can be easily invoked by agents.
We optimized the Xinliu Xuntui platform to enhance the flexibility of resource scheduling. Focusing on training and fine-tuning scenarios, we added capabilities for borrowing and reclaiming physical queue resources and flexibly configuring computing power, which greatly improved resource utilization under multi-task variations and lowered the threshold and operational costs for R&D teams' training environments.
To meet the demands for privatization and domestic computing, the Galaxy Platform has completed deep adaptation for multiple mainstream domestic computing chips and achieved visualized management throughout their entire lifecycle.
In products and technology, we continued to upgrade our full-stack AI capabilities for intelligent computing and AI application deployment. This quarter, we further optimized model deployment on the Xingliu Mars platform for high-concurrency inference, significantly improving throughput for several core models and enabling more granular access, usage, and model-level management.
We also launched the Agent Kit, providing secure sandbox, knowledge and memory management, and evaluation and governance tools to help enterprises build production-grade AI agents. At the same time, we are making general-purpose cloud products such as databases and storage easier for agents to access and use.
We enhanced the staff load training and inference platform with more flexible resource scheduling, sharing, and allocation for training and fine-tuning workloads, improving utilization and reducing development and operating costs for the private deployment of domestic AI infrastructure. Our Galaxy Stack platform has achieved deep integration and full lifecycle visual management for multiple mainstream domestic AI chips.
Mr. Zhou
Moving forward, we will continue to tap into significant business opportunities both within and outside our ecosystem, constantly optimize the operational efficiency of our computing assets, and strengthen our profitability and self-sustaining cash generation capabilities amidst the wave of technological evolution and application implementation. We aim to create long-term, sustainable value for our customers, shareholders, and society through solid performance. Next, I invite our CFO, Mr. Li Yi, to present the financial results for the second quarter. Thank you.
Looking ahead, we will continue to capture opportunities both within and outside our ecosystem, improve the operating efficiency of our computing assets, and strengthen our profitability and cash generation capability. Amidst the tailwinds in the AI industry, we remain committed to creating long-term sustainable value for customers, shareholders, and society. With that, I will hand the call over to our CFO, who will review our second-quarter financial results. Thank you.
Ms. Li
Thank you, Mr. Zhou and Mr. Tian, and thank you all for joining the call today. I will now discuss the second-quarter financial results in RMB. Before we walk through the details of the financial results for the second quarter, I would like to highlight the following aspects.
First, our quarterly revenue exceeded RMB 3 billion for the first time in the company's history, marking year-over-year growth for several consecutive quarters. In particular, our AI cloud gross billings increased by 82% year-over-year to RMB 1.33 billion, accounting for over 43% of our total revenue, compared to 31% a year ago. This reflects a continued structural shift in our business mix towards AI.
Second, our profitability has improved. Our adjusted gross margin was 50.4%, up 2.4 percentage points quarter-over-quarter and 0.5 percentage points year-over-year. Our adjusted EBIT margin reached 36%, up from 17% in the same quarter last year and 32% last quarter.
Notably, we returned to break-even at the operating income level this quarter and recorded an adjusted operating profit margin of 4%. These outcomes validate our ability to convert strong demand in our AI business into healthy profit growth.
Third, we continue to invest to accelerate the build-out of our AI computing capacity. Capital expenditures, together with right-of-use assets obtained through third-party financing and finance leases, reached RMB 3.3 billion this quarter, versus RMB 2.9 billion in the previous quarter and RMB 2.8 billion in the same quarter last year.
Ms. Li
Now, let me walk you through our financial results for the second quarter of 2026. Total revenue for the quarter was RMB 3,072 million, up 31% year-over-year and 40% quarter-over-quarter. Of this, revenue from public cloud services was RMB 2,358 million, up 45% from RMB 1,625 million in the same quarter last year.
Revenue from enterprise cloud services reached RMB 740 million, compared with RMB 724 million in the same quarter last year, a slight decrease of 1% year-over-year. Total cost of revenue was RMB 2,606 million, representing a 30% year-over-year increase, mainly due to continued investment in AI cloud infrastructure.
IDC costs increased by 23% year-over-year, from RMB 803 million to RMB 990 million this quarter. The increase was mainly driven by higher rack service costs. Depreciation and amortization expenses increased by 75% year-over-year, from RMB 552 million in the same quarter of 2025 to RMB 964 million this quarter, largely due to depreciation of newly acquired AI infrastructure, including servers and network equipment.
Solution development and service costs increased by 4% year-over-year, from RMB 564 million in the same quarter of 2025 to RMB 586 million this quarter. The modest increase was mainly due to higher costs incurred in AI transformation for solution development and delivery. Sales and marketing expenses and other costs totaled approximately RMB 66 million this quarter, compared to RMB 92 million in the same quarter last year.
Ms. Li
Our adjusted gross profit for the quarter was RMB 472 million, an increase of 35% year-over-year and 34% quarter-over-quarter. The adjusted gross margin was 15.4%, up from 14.9% in the same quarter last year and from 13% in the previous quarter. This increase was driven by higher gross margins in the public cloud business, thanks to strong AI demand.
Excluding share-based compensation expenses, our total adjusted operating expenses were RMB 391 million, a decrease from RMB 761 million in the same quarter last year and from RMB 455 million in the previous quarter, mainly reflecting our disciplined cost and expense control.
Research and development expenses were RMB 884 million, up 101% year-over-year. Adjusted sales and marketing expenses were RMB 102 million, down 7% year-over-year. Adjusted general and administrative expenses were RMB 105 million, down 61% year-over-year, largely due to lower credit loss expenses.
Our adjusted operating profit was RMB 124 million, turning profitable from an adjusted operating loss of RMB 166 million in the same period last year. This improvement was primarily driven by revenue scale expansion, higher gross margins, and enhanced operating efficiency. The adjusted operating profit margin was 4% this quarter, compared with -7.1% in the same period last year and -2.2% in the previous quarter.
Ms. Li
Our adjusted net loss was RMB 61 million, down from RMB 300 million in the same quarter last year and RMB 237 million in the previous quarter. Our non-GAAP net income was RMB 1.1 billion, an increase of 171% from RMB 406 million in the same quarter last year.
Our non-GAAP gross margin reached 36%, compared with 70% in the same quarter last year and 82% in the previous quarter. This decline was primarily due to higher depreciation costs in our cost base as we accelerate the build-out of our AI computing capacity, despite improvements in gross profit.
As of June 30, 2026, our cash and cash equivalents totaled RMB 4,674 million, compared with RMB 4,904 million as of March 31, 2026. The modest decrease was mainly due to our continued investment in AI infrastructure to support business growth.
Looking ahead, we aim to capitalize on the explosive growth in AI demand by further investing in infrastructure, expanding our product and service offerings, managing credit and liquidity risks, and improving operating efficiency. We remain committed to our 'All-in AI' strategy and continue to deliver high-quality growth to our shareholders. Thank you all.
This concludes our prepared remarks. We will now begin the Q&A session. If possible, please ask your questions in both Mandarin and English. Operator, please proceed.
Operator
Thank you. We will now begin the question and answer session. If you wish to ask a question, please press *11 on your telephone and wait for your name to be announced. To withdraw your question, please press *11 again. We will now take our first question. Your first question comes from Li Pingzha from CICC. Please go ahead. Your line is open.
Li Pingzha
Good evening, Mr. Cui and Ms. Li. Thank you for taking my questions. First, congratulations to the company for achieving profitability at the operating level this quarter. I have two questions, both regarding our MaaS (Model-as-a-Service) platform. Recently, there has been significant discussion about the improved capabilities of open-source models. Could you please discuss the impact of this trend on the company's MaaS business? Based on your observations of your own platform, what is the current trend in usage growth, and what are the primary typical use cases driving this growth?
My second question is: Considering that the MaaS platform may have a shorter investment payback period, will the company allocate more resources to this business segment?
First, how significantly do improvements in open-source model capabilities impact the company's MaaS (Model-as-a-Service) business? Based on your observations, what is the current growth trend in usage, and which use cases are the primary drivers? Second, given that the payback period for the mass-market business might be shorter, will the company allocate more resources to it? Thank you.
Mr. Liu Tao
Alright, let me address this question. First, let's look at the changes brought about by enhanced model capabilities. On one hand, the domestic market for web coding demand is substantial, with significant volume. Traditionally, we have observed that a large number of customers tend to use overseas models.
As the capabilities of models like Kimi and others improve, we are seeing a substitution effect where domestic models are replacing overseas ones in the web coding sector, driving demand for domestic models. On the other hand, the penetration of agentic workflows is increasing. More scenarios are adopting agents, with many of our clients' agents moving into production, which has led to the launch of our Agent Key product.
Regarding agent usage, we observe a bifurcation in demand. For complex problem-solving, there is a tendency to use newer models like K3 or GM-5.3. However, for routine tasks, customers tend to choose the most cost-effective models, as seen in the recent price cuts.
Consequently, we are seeing growth in call volumes for models like V4 Flash and Mimo Flash in daily work scenarios. This addresses the first part of your question regarding the main drivers of this growth trend.
On another note, the MaaS business and the computing power business each have distinct characteristics. For the computing power business, when we sell a machine, the utilization rate is effectively 100%. Furthermore, we typically sign long-term contracts for security reasons, making this business relatively safe with high and stable profit margins.
The MaaS business, however, is subject to fluctuations in market price competition, substitution by new models, operational efficiency, and other multifaceted factors, resulting in variable profit margins. Therefore, we view these two businesses as a balanced strategic choice. We aim to develop both while using them as backups for each other.
Specifically, if our computing power business faces special circumstances, we can utilize it for inference tasks. Conversely, when inference demand spikes, we can shift inference workloads to the computing power infrastructure. Overall, this represents a strategy of balanced development. Thank you.
OK. So just to quickly translate, this answer comes from our SVP, Mr. Liu Tao. In relation to your first question, the development in open-source large language models has mainly three impacts. #1 is that we're seeing very big demand coming from web coding. Traditionally, foreign models have been taking the lead in this area.
However, following the launch of high-performance models such as GLM and K3, we are seeing increasing adoption of these China-made large language models by users from Mainland China. Secondly, the growing use of agentic scenarios has also brought changes to our business. In response, as mentioned in our prepared remarks, we have launched the Agent Kit product to meet this demand.
It is certainly worth noting that for day-to-day routine tasks and workloads, the preference is usually for cost-effective models, which are essentially Chinese models. This is why the development of open-source large language models is actually beneficial to our business.
Regarding your second question on the balance between our MaaS (Model-as-a-Service) business and computing power: we essentially have different business models for these two segments. For the computing power business, once we sell the capacity, utilization is inherently 100%, and we typically secure this through long-term contracts to ensure sustained utilization over an extended period.
Therefore, it is relatively safe, so to speak. However, the MaaS business is subject to several factors, including fluctuations in token prices, the launch of new models that customers may prefer, and the operational efficiency we can achieve in running the MaaS business.
Consequently, we generally balance these two business models, aiming for them to complement each other. We dynamically evaluate both businesses to determine the optimal allocation of resources. Thank you.
Lipping Zal
Thank you. That is very helpful.
Operator
We will take our next question. The next question comes from Wenden Vu from CISA. Please go ahead. Your line is open.
Wenden Vu
Good evening, Mr. Zhou, Mr. Li, and members of the management team. Thank you for giving me this opportunity to ask a question, and congratulations to the company on its strong performance. I have two questions. First, regarding chip procurement progress from June to the current month, what is the status? Also, what are our latest expectations for full-year development?
The second question concerns the enterprise cloud segment. We have observed a deceleration in revenue growth for this segment over the past two quarters. How should we forecast the full-year growth rate for enterprise cloud, also known as industry cloud? Additionally, what is the company's strategy regarding AI transformation and its medium-to-long-term positioning for this business line?
The first question is that since June, how has chip procurement progressed in recent months, and what is your latest CapEx guidance? The second question is about the enterprise cloud segment, where revenue growth has decelerated over the past two quarters. How should we view the full-year growth outlook for enterprise cloud, and what are the plans for AI transformation and medium-term positioning for this segment? Thank you.
Mr. Tianhai
Let me address the first question. Supply chain issues are always a major concern during our earnings releases. I believe you have also observed that supply constraints have been accompanying business development since this wave of growth began three years ago.
Moreover, these supply constraints do not appear to be short-term; rather, they are likely to persist for an extended period. Therefore, we should view this as a new normal. Secondly, looking at our development over the past few years, it is evident that China's AI and computing industries have been advancing at a very rapid pace, which is also reflected in our financial performance.
From our perspective, our most important strategy has been to continuously expand our number of partners and suppliers. Simultaneously, we are accelerating our technical adaptation speed to enhance the diversification of our supply chain.
You may have noticed that domestic chip companies have gone public, and their performance, production capacity, and adaptation optimization are improving rapidly. In particular, they have gained widespread application and recognition in the inference sector. Thirdly, due to the structural nature of the supply market, there are instances of large one-time purchases, leading to irregular batch procurement rhythms.
Therefore, if we look at monthly data, the trend is not linear, and procurement amounts fluctuate significantly. However, when viewed on an annual scale, the figures are quite close to our expectations. For specific numbers, please refer to...
That was a bit fast, so allow me to provide a quick translation. The response comes from our SVP, Mr. Tianhai. To summarize three points: First, it has been three years since 2023, and the market has consistently heard concerns about limited supply. I would say this is actually the new normal. Such supply difficulties are a long-term situation.
Secondly, we should also recognize that despite these constraints, the development of China's cloud computing and AI industries has not been restricted. Our approach to addressing this situation is to increase the number of business partners we work with.
We are working to expand our supplier base and enhance the compatibility of chips made in China. As you are well aware, many Chinese-made chips have recently become publicly available and are particularly effective in use cases such as model inference.
Regarding point number three, I would like to note that CapEx figures tend to be quite volatile on a month-to-month basis. This is because purchasing often involves large sums concentrated in a relatively small number of transactions. Therefore, monthly purchase figures do not follow a linear trend.
Thus, I would say that our full-year CapEx estimate remains in line with our expectations. Our CFO should be able to provide you with more details in this regard.
Ms. Lee
Here, our capital expenditure included capitalized asset lease arrangements totaling RMB 6.2 billion in the first half of 2026, accounting for over 75% of our four-year CapEx from last year. While July's status may not fully represent the third quarter's overall performance, it clearly demonstrates an acceleration in tangible growth. Accordingly, we maintain our full-year CapEx guidance at RMB 15 billion.
Wenden Vu
Thank you, Tintin. Thank you.
Operator
We will take our next question. Oh, apologies, we need to address one more question first. My apologies.
Ms. Regent
Regarding Kingsoft Cloud, we acknowledge the slowdown in revenue growth over the past two quarters. However, we believe it is inappropriate to simply extrapolate the full-year operating performance based on the revenue growth rates of just two quarters. We attribute the recent changes in growth rate to three main factors.
The first factor is well-known: since the second half of last year, rising prices for memory and storage have forced government and enterprise clients to repeatedly adjust their budgets. This has extended our contract signing and budgeting cycles compared to previous periods, representing a very significant market fluctuation in the first half of the year.
The second factor involves seasonal fluctuations in revenue recognition for government and state-owned enterprise (SOE) businesses, particularly regarding contract signing, delivery, and acceptance. As many are aware, especially with large-scale projects, revenue recognition for these clients tends to be concentrated in the second half of the year. Therefore, we believe that full-year revenue performance should not be judged solely based on year-over-year comparisons from the first half.
From our perspective, we place greater emphasis on signed but unrecognized contracts, project acceptance milestones, and the coverage of high-certainty business opportunities. The third factor is that, driven by the current AI wave, Kingsoft Cloud has proactively adjusted its revenue structure within the industry cloud segment.
Previously, the industry cloud segment relied heavily on cloud infrastructure construction, system integration, and project delivery, which generated significant revenue scale. However, this project-based model exhibited pronounced characteristics and technical volatility. This year, following the surge in popularity of Large Language Models (LLMs) in the Chinese market, we have proactively reduced projects with low gross margins, low reusability, and poor payment collection quality.
We are concentrating core resources on projects where AI helps clients integrate into core business processes, thereby building proprietary AI product capabilities and generating long-term operational revenue. Consequently, there may be a temporary divergence between short-term revenue growth and metrics such as contract quality, contract duration, and future sustainable revenue due to the three factors mentioned above. This does not indicate an abandonment of growth; rather, it reflects our active restructuring of the foundation for future growth.
OK. So this answer comes from our VP, Miss Regent. So generally, I don't think although we're seeing relatively slow growth in the enterprise cloud segment, I would say it is not the right way to look at it from the linear extrapolation perspective. I will give you 3 reasons.
I think #1 just to explain why we're thinking of looking at relative weakness in this regard is that the upstream supply pricing hiking hike, which changed quite significantly in recent, in recent quarters has affected the our prospective customers. Essentially the SOE companies and also the government agencies to they have to frequently adjust their budgeting quota, which delayed their decision making process. So that's number one.
And #2 you're all quite aware that the seasonality in enterprise cloud business is quite strong. Usually the delivery and the revenue recognition are concentrated in the second-half of every year. So we have actually a quite a strong pipeline to deliver in second-half of the year.
And thirdly, this is actually a result of a proactive adjustment of our business structure, namely proactively from the project based business model to operating based business model. We're operating business model from a financial reporting perspective is automatically classified into public clouds. So this is not, you know, simply as it was seized as a, you know, weakening of the enterprise cloud business. So that's the three points I'd like to offer. Thank you.
Operator
Thank you. We will take our next question. Your question comes from Timothy Zelle from Goldman Sachs. Please go ahead. Your line is open.
Timothy Zelle
Good evening, Mr. Liu and Mr. Li. Thank you for taking my questions. I have two questions to ask. The first one relates to our MaaS (Model-as-a-Service) business. I would like to understand Kingsoft Cloud's positioning, application scenarios, and competitive advantages compared to other competitors in the market.
Could you provide more details on revenue recognition and profit margins for our MaaS services? That is the first question. The second question concerns pricing changes in our AI cloud services. Could you share the trends in AI cloud service pricing we have observed over the past few months, as well as the general pricing landscape in the industry?
We have announced certain price increases or reductions in discounts. What has been the customer feedback so far? Could you help quantify the impact of these pricing factors on the revenue growth rate of our AI cloud business? I will translate this shortly.
Thank you, management, for taking my question. My first question is regarding the MaaS business. I am wondering how you view Kingsoft Cloud's competitive advantage in MaaS services compared to peers in the market, particularly in terms of application scenarios. Could you also share more about the revenue recognition and profitability profile of the MaaS service business?
My second question is regarding the overall pricing trend in the AI cloud business. Could you share the latest trends over the past couple of months? Also, what feedback have you received from customers following the price hikes or discount reductions announced in recent months? Lastly, are you able to quantify the impact of these price increases on your overall AI cloud revenue growth? Thank you.
Mr. Liu Tao
Let me address these two questions. The first concerns the positioning and advantages of our Qingliu platform. First, our market positioning is quite unique. Unlike some competitors, we do not develop our own large language models (LLMs), yet we are a cloud provider. In the market, we generally see two types of clients engaging in MaaS scenarios.
One type consists of cloud providers that develop their own large models, such as Alibaba and Baidu. The other type includes so-called pure-play 'token factories' in the cloud, such as those recently seen in the market. We believe we differ from both groups.
First, since we do not develop our own large models, our sales and business teams are guided by a policy to sell the best models that customers prefer, such as GPT-4o, Kimi, and potentially video generation models like Kling or MiniMax. Therefore, we tailor our offerings to customer preferences rather than being bound by any mandate to push a specific proprietary model. We believe this flexibility is a key component of our competitive strategy.
Secondly, regarding the cloud-based token factory, a critical factor in token production is computing power. Only companies that control low-cost computing power can generate profits within the entire MaaS (Model-as-a-Service) ecosystem. If a company purely focuses on software production and leases computing power, a significant portion of its profits may be captured by the computing power suppliers. Investors should pay attention to this dynamic. For instance, by comparing the hourly rental cost of overseas NVIDIA GPUs with their selling prices, one can estimate the profit margin for vendors performing inference without owning underlying resources. This highlights our competitive advantage in platform positioning within the MaaS sector.
Regarding your question about our market positioning, we hold a unique position in the mass market business. Unlike some full-stack cloud providers that rely on in-house or proprietary models, we do not have such models. Consequently, we are not constrained to sell large-scale models offered by our affiliated companies.
As a result, we are able to—and indeed encourage our sales team to—sell the models that customers prefer most, such as GRM and others. That is the first point. Secondly, in today's market, possessing proprietary or owned computing power is crucial, as it is the only way to secure significant profitability.
Mr. Liu Tao
Secondly, regarding price adjustments, you may have noticed that we have significantly increased prices in sectors related to storage and computing power. The price increase for storage was substantial. However, in computing scenarios, storage demand grows in tandem with computational needs. Therefore, customers were not particularly sensitive to the storage price hike, and we observed that they readily accepted the increase.
During this price adjustment process, we not only passed through our rising costs but also maintained or even improved our profit margins. The same logic applies to computing power. While the cost of hardware is undoubtedly rising, our unique capabilities allow us to adjust pricing in line with increasing computing costs while maintaining, or even enhancing, our gross margins.
Furthermore, we accept customer-hosted computing resources, which we then organize, operate, and maintain. We have seen such scenarios implemented in the current period, and the gross margin profile for this business segment is evidently higher. That covers these points.
In response to your question about the price hikes, we have increased prices for two main products or solutions: first, storage, and second, computing power. I will address each separately.
Regarding storage, it typically represents an incremental cost associated with intelligent computing demand, constituting a relatively small portion of the overall transaction size. Consequently, the vast majority of customers we negotiated with readily accepted the price increase.
As a result, we were able not only to pass through the cost increases in some cases but also to improve our profitability in those scenarios. Secondly, regarding computing power, thanks to our specific capabilities—including historical strengths, as well as operations, maintenance, and network capabilities—we were again able to pass these cost increases on to our customers.
In some instances, we have also improved our profitability. This quarter, we have initiated several managed service projects under an asset-light business model. We look forward to seeing more of these contributions reflected in our financial statements. Thank you.
Operator
Thank you. We will now take the next question. The next question comes from Wei Zhong at UBS. Please go ahead; your line is open.
Wei Zhong
Good evening, management team. Congratulations on the strong performance this quarter, and thank you for taking my question. I have one follow-up regarding the differentiated positioning of MaaS (Model-as-a-Service) on our platform mentioned earlier. Considering the proprietary models and user ecosystems of other cloud providers, how should we view our long-term positioning in the cloud industry, as well as the sustainability and achievability of stable profit margins?
Let me quickly translate that myself: Good evening, management. Congratulations on the solid quarter, and thank you for taking my question. Considering the proprietary models and user ecosystems of other cloud providers, how should we think about our long-term positioning in the cloud market and the sustainable margin levels going forward? Thank you.
Mr. Liu Tao
I will address this question. As mentioned earlier, our MaaS offering aims to provide both top-tier models and stable, reliable services. As a neutral vendor, we maintain strong relationships with all original model providers and are able to launch online services for these models promptly. By combining our own resources with those of the original providers, we can offer customers highly available and reliable cloud-based MaaS services. In this regard, we operate without any so-called 'psychological burden' or conflict of interest.
Furthermore, given the total volume of resources at our disposal, we are typically able to provide customers with higher guarantees regarding Service Level Agreements (SLAs). The large-scale infrastructure of cloud vendors offers better support for MaaS operations. From a profitability perspective, we collaborate with original providers—for example, on model inference frameworks and undisclosed weights during model releases—to enhance our inference efficiency.
Additionally, we have been consistently investing in inference optimization. Our team has verified that, across multiple open-source models, we can achieve inference efficiency comparable to, or even approaching, that of the original providers. Overall, regardless of model updates, I believe our team has the capability to keep pace with the inference efficiency of new models and maintain healthy gross margins.
Therefore, we believe it is crucial for a cloud AI service provider to offer the top models preferred by customers, along with stable services. As a neutral cloud player, we maintain good relations with all leading large language model labs and are able to provide the best models according to customer demand.
Furthermore, leveraging our technological capabilities and the SLAs we have signed with them, we are able to provide highly available and highly reliable services. Thirdly, regarding your question on profitability, it is crucial to work closely with LLM companies.
For instance, collaborating with labs to optimize model inference, including working based on undisclosed model weights to enhance inference efficiency. In some cases, we have achieved inference efficiency levels comparable to, or even matching, those of the LLM companies themselves. Thank you.
Operator
Thank you. We will now take our final question. The final question comes from Ying Liu at Morgan Stanley. Please go ahead; your line is open.
Ying Liu
Thank you for the opportunity to ask a question. First, congratulations on the company's strong financial results. My question concerns the return on invested capital (ROIC) under the two business models: computing power leasing and Model-as-a-Service (MaaS). Specifically, what are the current ROIC levels, and has there been any recent marginal improvement or decline in these returns?
Let me rephrase my question. I would like to ask about the ROIC for the two business models—computing power leasing and Model-as-a-Service—and the recent marginal changes in ROIC. Thank you.
Ms. Li
Thank you, Lu Yang. At this stage, we do not disclose separate ROIC figures for MaaS and AI computing power services because our asset turnover varies across projects, driven by payback cycles for margin-fixed assets and depreciation policies. Overall, MaaS currently demonstrates significantly better profitability than AI computing power services.
We have seen continued improvement in operating leverage as our AI business scales up. Costs are being steadily diluted, and our trailing twelve-month adjusted operating profit has turned positive, driving a gradual recovery in our overall return on assets. We adhere to a demand-driven and disciplined AI investment strategy with a strong focus on capital efficiency.
With continuous optimization of our business structure and organization, along with substantial AI commercialization, I believe our overall ROIC will continue to improve steadily. OK, thank you.
Operator
Thank you. There are no further questions. Apologies for the Q&A session. I will now hand back for closing remarks.
Jackie Zor
OK, thank you all for joining us today. If you have any further questions, please contact our IR team. Have a good evening. You may now disconnect.
Operator
Thank you. This concludes today's conference call. Thank you for participating. You may now disconnect.
More details:Kingsoft Cloud IR
Disclaimer: The above content is generated by an AI language model based on public data and third-party automatic subtitles. The above content does not represent any position of Futu and does not constitute any investment advice. Futu Group makes no express or implied warranties or representations regarding the accuracy, timeliness, or completeness of the above content.
Risk Disclaimer: The above content only represents the author's view. It does not represent any position or investment advice of Futu. Futu makes no representation or warranty.Read more
Comments (43)
to post a comment
20
2