Q4 2024 Earnings Summary
- Kingsoft Cloud achieved a 30% year-over-year revenue growth in Q4 2024, exceeding top players in the industry, driven by strong performance across both public cloud and enterprise cloud services.
- The company has shifted its strategy from traditional IT cloud services to AI cloud and solution services, leading to higher margins compared to traditional CPU-centric IaaS services, and providing higher profitability potential due to increased demand for GPU AI services.
- Kingsoft Cloud is poised to benefit from the accelerated adoption of large language models in China, such as DeepSeek, which is lowering deployment costs and increasing demand for AI applications across industries, particularly within the Xiaomi and Kingsoft ecosystem, where AI adoption is expected to be very strong.
- Slow actual business volume for AI inference outside of Xiaomi and Kingsoft ecosystem: Although there is a clear trend of AI and large language models penetration, the company has not yet seen significant actual business volume from customers outside of Xiaomi and Kingsoft ecosystem, especially in terms of large orders (e.g., clusters of 128 or 256 units). This suggests that growth in AI inference revenue outside of their ecosystem may take time to materialize.
- Management cautious about sharing pricing strategy and margin details, indicating uncertainties: In response to questions about pricing strategies for GPU cloud revenue amid falling prices for high-performance AI servers, management declined to provide specifics due to business sensitivity. This reluctance to share details may indicate uncertainties or pressures on pricing and margins.
- Significant capital expenditures and potential financial risks: The company has invested around RMB 5-6 billion in capital expenditures on computing equipment over past few quarters, and expects to invest around RMB 10 billion in total AI investment for full year 2025, stating it's "just a start". They are utilizing leasing and off-balance-sheet arrangements to finance this. This significant investment and reliance on alternative financing may introduce financial risks and pressure on cash flows.
| Metric | Period | Previous Guidance | Current Guidance | Change |
|---|---|---|---|---|
Revenue Growth | FY 2025 | no prior guidance | Accelerated revenue growth on an annual basis for FY 2025 | no prior guidance |
Profitability | FY 2025 | no prior guidance | Positive non‐GAAP operating profit for the full year FY 2025 | no prior guidance |
AI Investment | FY 2025 | no prior guidance | Investment of RMB 10 billion in AI‐related expenditures (OpEx and CapEx) for FY 2025 | no prior guidance |
Revenue Contribution from Xiaomi/Group | FY 2025 | no prior guidance | Revenue from connected parties expected to reach RMB 11.3 billion over FY 2025–FY 2027 | no prior guidance |
| Topic | Previous Mentions | Current Period | Trend |
|---|---|---|---|
Consistent strong AI revenue growth and strategic investment in AI | Q1–Q3 discussions detailed AI revenues growing at triple‐digit year-over-year rates, significant infrastructure investments (e.g., high CapEx directed to GPU servers and proprietary platforms), and evolving revenue contributions from AI services | In Q4, AI gross billings hit RMB 474 million with nearly 500% year-over-year growth and AI now contributes 34% of public cloud revenue, demonstrating a deepened strategic investment in AI | Continued strong growth with increasing commitment to AI, reinforcing a bull sentiment |
Dependence on key ecosystem partners (Xiaomi and Kingsoft Group) | Across Q1–Q3, the ecosystem provided between 19% and 36% of revenue, with discussions of deepening relationships and forward-looking revenue commitments (including a proposed RMB 11.3 billion cap over three years) | Q4 emphasized a 78% year-over-year increase from the ecosystem, with revenues reaching RMB 493 million (22% of total revenues) and highlighted its strategic, long-term revenue role | Growing importance and strategic reliance on ecosystem partners, with an increasingly favorable tone |
Rising capital expenditures and financing risks for AI initiatives | In Q1–Q3, heavy investments were noted (e.g., RMB 1.2 billion in Q1, large CapEx allocations in Q2, and significant GPU server investments in Q3) with clear discussions on financing strategies (leasing, bank borrowing, external partnerships) | Q4 continued to show significant AI investments (e.g., depreciation costs of RMB 343.1 million due to GPU acquisitions) but provided less detailed commentary on financing risks | Ongoing high CapEx spending with a slight de-emphasis on financing risk details, suggesting cautious optimism |
Transition from traditional IT/cloud services to AI cloud and solution services | Q1–Q3 highlighted a deliberate pivot—scaling down low-margin services like CDN, establishing a dedicated AI subsidiary, investing aggressively in AI infrastructure, and enhancing pricing models | Q4 confirmed the strategic shift with a new higher pricing model for AI services and a continued increase in AI’s contribution to revenue, signifying the transformation is complete | A clear and committed transformation to AI-based services continues, with a strongly positive outlook |
Profitability improvements with ongoing concerns over margin transparency | Q1–Q3 discussions showed improving adjusted EBITDA and gross profit margins from shifting to high-margin AI services and effective cost controls, though detailed margin breakdowns remained elusive | Q4 reported record non-GAAP gross profit, a 16.1% EBITDA margin, and significant operating profit improvements, yet management maintained caution over full margin transparency | Sustained profitability gains are evident, while concerns over granular margin transparency persist—a cautiously optimistic view |
Adoption and integration of large language models and AI training services | Q1–Q3 detailed investments in high-computing clusters, the launch of AI platforms, and integrations with projects such as legal AI and public services, showcasing active LLM adoption and diversified training support | Q4 further advanced LLM integration with the launch of the Singular platform and highlighted DeepSeek’s role in reducing deployment times and driving ecosystem-wide AI adoption | Accelerated LLM adoption and streamlined AI training integrations underline robust and optimistic growth prospects |
Emerging competitive pressures in the AI market | In Q1, there was explicit mention of pricing competition from ByteDance, Alibaba, and Baidu—indicating a maturing competitive landscape and acknowledgement of robust competitors | Q4 did not mention competitive pressures from these players, suggesting the focus has shifted away from this concern | The theme of competitive pressure has faded from recent discussions, implying either stabilization or deprioritization of this risk |
Downscaling of non-core businesses (e.g., CDN) | Q1 through Q3 repeatedly noted the strategic reduction of low-margin CDN services—from over 50% at peak to significantly lower percentages—to focus resources on higher-margin opportunities | Q4 confirmed further downscaling, with CDN’s contribution dropping to between 10% and 15%, reinforcing the focus on core, high-margin services | Continued and deliberate downscaling of non-core operations points to a steady transformation towards higher profitability |
-
Revenue Growth Outlook
Q: What's the expectation for 2025 revenue growth?
A: In Q4 2024, we achieved a 30% year-over-year top-line growth , exceeding even top players in the industry. This growth spans all verticals, including both public cloud and enterprise cloud, with healthy rates in both AI and non-AI segments. Our growth is built on a solid foundation after completing business strategy adjustments in early 2024, resulting in a healthy client mix. For 2024, the keyword is acceleration, and while AI is a very important driver, other verticals will also contribute to healthy growth going forward. -
Margin Outlook and Profitability
Q: Can you share thoughts on long-term profitability trends?
A: Our gross margin has been improving consistently over the past few quarters. Since Q2 2024, our operating cash flow has been positive. Our EBITDA margin has exceeded industry average and is among the top tier of industry peers. We intend to maintain a healthy, steady pace of margin expansion across gross margin, EBITDA margin, and operating profit margin. EBITDA and operating profit will grow at a faster pace compared to gross margin, leading to improved cash flow and the ability to reinvest in AI. -
Capital Expenditure Plan and Funding
Q: Provide an update on the 2025 capex plan.
A: Our total AI investment comprises two parts: OpEx on data centers and CapEx for computing power. We employ an asset-light model, partnering with telecom and IDC companies, which reduces leverage and increases efficiency. In the past four quarters, we've spent around RMB 2.9 billion on AI data center OpEx and RMB 5–6 billion on CapEx for computing and network equipment, totaling RMB 8–10 billion. For 2025, the RMB 10 billion total AI investment is just a start. With strong support from shareholders like Xiaomi and Kingsoft Group, we've arranged leasing and off–balance sheet financing to support growing AI investment beyond our existing cash. Given our revenue growth, we'll continue to increase investment, potentially above industry average in 2025. -
Impact of AI Trends and DeepSeek
Q: How are AI industry trends impacting the cloud computing market and Kingsoft Cloud?
A: The advent of DeepSeek has accelerated the deployment of large language models in China, benefiting Kingsoft Cloud. It has proven that Chinese technology companies can achieve significant technological heights, leading to a revaluation of tech firms. DeepSeek's optimization in engineering methodologies lowers costs and increases application penetration. As more enterprises and individuals adopt AI, Kingsoft Cloud, as an active participant, will definitely benefit. This broader recognition of the strategic value of large language models and AI is raising awareness of Kingsoft Cloud. -
Pricing Strategy for GPU Cloud Revenue
Q: How will falling AI server prices affect GPU cloud revenue and earnings?
A: We have a higher margin compared with traditional CPU-centric services. Our value proposition has shifted from traditional ICE services to AI cloud and solution services, which are critical to our clients' workflows. Clients are willing to pay higher fees due to the necessity and value of the technology. We offer a mix of computing formats to meet client demand and price our services accordingly. Our pricing model for AI-related services is different and higher than traditional services, supporting our growth and profitability. -
AI Inference Demand Outlook
Q: What's the outlook for AI inference demand and its contribution?
A: The trend of AI and DeepSeek penetration is clear, with increased customer numbers and willingness to adopt AI applications. However, large single-customer demand requiring big clusters hasn't materialized yet outside the Xiaomi and Kingsoft ecosystems. Within these ecosystems, the demand is very strong; all products and services, including cell phones, IoTs, and electric vehicles, will be supported by AI and large language models. Kingsoft Office AI user accounts have been increasing significantly. -
Legacy Public Cloud Business Outlook
Q: Will the legacy public cloud business continue to generate revenue?
A: In late 2023 and early 2024, we forecasted a major shift from traditional ICE to AI cloud and adjusted accordingly. We filtered out certain assets and made provisions for asset valuation. Now, all assets on our balance sheet are healthy, revenue- and profit-generating, including the RMB 5 billion in capital expenditures for computing power acquired in past quarters. We have a good potential to improve shareholder return and profitability going forward.
Research analysts covering Kingsoft Cloud Holdings.