AI billings growth of 82% drives Kingsoft Cloud's second-quarter beat, with MaaS emerging as a new revenue and profit growth engine

Overview

The report reviews Kingsoft Cloud's 2Q26 results and focuses on the impact of its AI cloud, MaaS, and GPU computing power businesses on its revenue mix and margins. Goldman Sachs believes that expanding AI demand, growth among non-ecosystem customers, and the ramp-up of MaaS drove revenue and earnings above expectations. It therefore raises its 2026E-2028E forecasts and target price, while highlighting risks related to chip supply, competition, customer AI investment, and financing capacity.

Core views

Kingsoft Cloud generated 2Q26 revenue of RMB3.1bn, up 31% yoy and 3% and 1% above Goldman Sachs' forecast and Visible Alpha consensus, respectively, primarily driven by 45% yoy growth in public cloud revenue. Total AI billings increased 82% yoy to RMB1.3bn, accounting for 43% of total revenue and 56% of public cloud revenue, up from 37% and 50% in 1Q26, indicating the continued rise in the importance of AI within the revenue mix. In terms of customer mix, revenue growth from the Xiaomi and Kingsoft ecosystems slowed from 69% in 1Q26 to 28%, but revenue from the top five non-ecosystem public cloud customers still increased 51% yoy, indicating that growth is not entirely dependent on ecosystem customers. Enterprise cloud revenue declined 1% yoy to RMB717mn, which the report attributes to revenue recognition seasonality skewed toward 2H26E and the company's proactive shift in business mix toward higher-value AI projects. The improvement in profitability outpaced revenue growth. The 2Q26 gross margin was 15.4%, up 0.4 percentage points yoy and 2.4 percentage points qoq, above Goldman Sachs' forecast of 14.3%, mainly due to higher profit contributions from the AI cloud business, particularly non-ecosystem customers. Adjusted EBITDA increased 171% yoy to RMB1.1bn, 21% above both Goldman Sachs' forecast and consensus; the adjusted EBITDA margin rose 18.5 percentage points yoy to a record high of 35.8%. Non-GAAP operating profit turned positive at RMB124mn, significantly above Goldman Sachs' forecast of RMB13mn; the operating margin increased 11.1 percentage points yoy and 6.2 percentage points qoq to a record high of 4.0%. The company's 2Q26 capital expenditures, including capitalized assets under lease arrangements, reached RMB3.2bn, compared with RMB3.0bn in 1Q26, while Goldman Sachs forecasts full-year 2026E capital expenditures of RMB16.2bn. The report views MaaS as a new growth and margin driver. Its 2Q26 revenue rose to 12x qoq, reaching RMB120mn and accounting for approximately 9% of AI revenue according to Goldman Sachs' estimates. Goldman Sachs believes MaaS has a higher margin than the GPU computing power business, so its rising revenue contribution should support gross margin expansion. However, MaaS is also more volatile than GPU computing power, with performance affected by token pricing, new model releases, and operating efficiency. The company believes its strengths include a neutral platform position, the ability to provide third-party large models based on customer needs, control over lower-cost computing power, collaboration with large-model providers to improve inference efficiency, and the provision of stable services under service-level agreements. The StarFlow training and inference platform already supports up to 120 large models and serves more than 230 enterprise customers. The company has also launched AgentKit, which provides secure sandboxes and knowledge and memory management tools for enterprise AI agents. The GPU computing power business constitutes a relatively stable pillar of AI growth. In addition to 51% yoy revenue growth from its top five non-ecosystem customers, the company has delivered large computing clusters to leading customers in embodied intelligence, autonomous driving, and AI for Science. Management believes GPU computing power contracts have fixed terms and that utilization reaches 100% after chips are leased, resulting in greater revenue stability than MaaS. However, tight high-end chip supply is viewed as a long-term constraint, and the company is pursuing supplier diversification, particularly by increasing the use of domestic chips for AI inference. In terms of pricing, Kingsoft Cloud has raised storage and computing power prices, which the report believes demonstrates its ability to pass through costs and expand margins. The company has also launched asset-light managed services to provide operations and support for customers' AI clusters. Based on the second-quarter performance, Goldman Sachs raises its 2026E-2028E revenue forecasts by 2%-5% and its adjusted EBITDA forecasts for the same period by 12%-15%. The new revenue forecasts are RMB12.8164bn, RMB16.4287bn, and RMB19.9886bn, respectively, compared with previous forecasts of RMB12.5066bn, RMB15.7022bn, and RMB19.0400bn. Adjusted EBITDA forecasts are RMB4.5410bn, RMB6.4434bn, and RMB8.3677bn, respectively. EPS forecasts were also raised from RMB-2.03 in 2026E, RMB-1.24 in 2027E, and RMB-0.52 in 2028E to RMB-1.28, RMB-0.22, and RMB0.77. The report expects 2H26E revenue to increase 34% yoy, above the 31% growth recorded in 2Q26. AI-related revenue contribution is expected to rise from 31% in 2025 to more than 40% in 2026E, driving total revenue growth of approximately 34% and 28% in 2026E and 2027E, respectively. The annual adjusted EBITDA margin is expected to increase from 24.4% in 2025 to 35.4% in 2026E, 39.2% in 2027E, and 41.9% in 2028E. The report continues to emphasize the medium-term growth visibility provided by the Xiaomi and Kingsoft ecosystems. Goldman Sachs expects the combined revenue contribution from Xiaomi and Kingsoft to rise from 27% in 2025 to 38% in 2028E, with a combined revenue CAGR of 43% in 2025-2028E, compared with a 21% CAGR for other customers. Xiaomi revenue is expected to record a 46% CAGR over the same period. Goldman Sachs believes Kingsoft Cloud's AI revenue contribution of 31% in 2025 is the highest among Chinese cloud service providers and that the company will benefit from Xiaomi's continued increase in AI investment, as well as incremental, primarily training-related demand from other internet and AI companies. In terms of valuation, Goldman Sachs maintains its Buy rating and raises its DCF-based 12-month target price from US$17 to US$18, while maintaining a WACC of 10.3% and a terminal growth rate of 3%. The new target price implies 2026E and 2027E price-to-sales ratios of 3.0x and 2.3x, respectively. Kingsoft Cloud currently trades at 1.7x next-12-month sales, below one standard deviation under its historical mean since 2025. Based on the current price of US$12.37 used in the report, the target price implies 45.5% upside.

Analysis framework

The report first compares 2Q26 revenue, margins, and adjusted EBITDA with Goldman Sachs' forecasts and market consensus, then breaks down the sources of growth and differences in profitability by public cloud versus enterprise cloud, ecosystem versus non-ecosystem customers, and MaaS versus GPU computing power. It subsequently incorporates management's commentary on product capabilities, customer demand, chip supply, and pricing to update its 2026E-2028E revenue, adjusted EBITDA, and EPS forecasts. Finally, it determines the target price using DCF and cross-checks it against next-12-month price-to-sales and historical standard-deviation ranges.

Methodology notes

• Valuation methodologyDCF discounted cash flow

  DCF discounted cash flow valuation

  Goldman Sachs discounts forecast-period cash flows and terminal value to present value, using a WACC of 10.3% and a terminal growth rate of 3% to calculate Kingsoft Cloud's value, and accordingly raises its 12-month target price from US$17 to US$18.

• Valuation methodologyPS valuation

  Comparison of forward price-to-sales ratios and historical ranges

  The report compares the target price's implied 2026E and 2027E price-to-sales ratios of 3.0x and 2.3x with the current next-12-month price-to-sales ratio of 1.7x. The current multiple is below one standard deviation under the historical mean since 2025.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Kingsoft Cloud (KC.US)

  The report views it as a beneficiary of Chinese AI cloud demand, Xiaomi's continued increase in AI investment, and expansion among non-ecosystem internet and AI customers.

  Strengths

  AI revenue accounted for 31% in 2025, which the report says is the highest level among Chinese cloud service providers. MaaS has higher margins, GPU computing power contracts are stable, and StarFlow supports up to 120 large models and serves more than 230 enterprise customers.

  Weaknesses

  Enterprise cloud revenue declined 1% yoy in 2Q26, MaaS revenue is more volatile than the GPU computing power business, and business expansion requires substantial capital expenditures and continued financing support.

  Comparison

  The current next-12-month price-to-sales ratio of 1.7x is below one standard deviation under the company's historical mean since 2025. The target price implies 2026E and 2027E price-to-sales ratios of 3.0x and 2.3x.

  Risks

  High-end chip supply, peer competition, weaker-than-expected AI investment by key customers, insufficient capital expenditure financing, and dilutive financing.

Key data

• 2Q26 revenueRMB3.1bnUp 31% yoy, 3% above Goldman Sachs' forecast and 1% above Visible Alpha consensus.
• 2Q26 public cloud revenue growth+45% yoyThe primary driver of second-quarter revenue growth.
• 2Q26 total AI billingsRMB1.3bnUp 82% yoy, accounting for 43% of total revenue and 56% of public cloud revenue.
• Revenue growth from the top five non-ecosystem public cloud customers+51% yoyRemained strong even as revenue growth from the Xiaomi and Kingsoft ecosystems slowed to 28%.
• 2Q26 enterprise cloud revenueRMB717mnDown 1% yoy due to revenue recognition seasonality and a business mix shift toward higher-value AI projects.
• 2Q26 gross margin15.4%Up 0.4 percentage points yoy and 2.4 percentage points qoq, above Goldman Sachs' forecast of 14.3%.
• 2Q26 adjusted EBITDARMB1.1bnUp 171% yoy and 21% above both Goldman Sachs' forecast and consensus.
• 2Q26 adjusted EBITDA margin35.8%Up 18.5 percentage points yoy to a record high.
• 2Q26 non-GAAP operating profitRMB124mnTurned profitable and exceeded Goldman Sachs' forecast of RMB13mn; the operating margin rose to 4.0%.
• 2Q26 MaaS revenueRMB120mnRose to 12x qoq and accounted for approximately 9% of AI revenue according to Goldman Sachs' estimates.
• StarFlow platform coverageUp to 120 large models and more than 230 enterprise customersThe platform has optimized the throughput of certain core models for high-concurrency inference.
• 2026E-2028E revenue forecastsRMB12.8164bn, RMB16.4287bn, RMB19.9886bnGoldman Sachs raises the forecasts by 2%-5% overall, from previous forecasts of RMB12.5066bn, RMB15.7022bn, and RMB19.0400bn, respectively.
• 2026E-2028E adjusted EBITDA forecastsRMB4.5410bn, RMB6.4434bn, RMB8.3677bnGoldman Sachs raises the forecasts by 12%-15% overall.
• 2026E-2028E EPS forecastsRMB-1.28, RMB-0.22, RMB0.77Previous forecasts were RMB-2.03, RMB-1.24, and RMB-0.52, respectively.
• AI-related revenue contributionMore than 40% in 2026E31% in 2025.
• Combined revenue contribution from Xiaomi and Kingsoft38% in 2028E27% in 2025; the combined revenue CAGR for 2025-2028E is expected to be 43%.
• Valuation parametersWACC 10.3%, terminal growth rate 3%Used for the DCF-based target price, with both parameters unchanged.
• Next-12-month price-to-sales ratio1.7xBelow one standard deviation under the historical mean since 2025.

Impact & implications

The report believes that expanding AI billings, growth among non-ecosystem customers, and the ramp-up of MaaS are simultaneously improving Kingsoft Cloud's revenue mix and profitability. MaaS's higher margin, the stability of GPU computing power contracts, and pricing power support medium-term margin expansion. The Xiaomi and Kingsoft ecosystems continue to provide growth visibility, while other AI customers broaden the sources of demand. However, achievement of the forecasts still depends on chip supply, customer AI investment, capital expenditure financing capacity, and the competitive environment.

Risks

• Supply-chain disruptions or an inability to obtain high-end chips could constrain expansion of the AI computing power business.
• Intensifying competition among cloud service providers could affect revenue growth, pricing, and margins.
• Lower-than-expected AI investment by key customers, including Xiaomi, could weaken demand growth.
• The company may be unable to secure sufficient funding to support capital expenditures.
• Potential dilutive financing activities could affect existing shareholders' interests.

What to watch

• Monitor the sensitivity of MaaS revenue to changes in token pricing, new model releases, and operating efficiency.
• Monitor whether enterprise cloud revenue can be recognized in 2H26E according to seasonal patterns and the progress of the business mix shift toward higher-value AI projects.
• Monitor high-end chip supply constraints and supplier diversification, particularly the adoption of domestic AI inference chips.
• Monitor whether AI investment by Xiaomi and other key customers supports the revenue growth and customer mix changes forecast in the report.
• Monitor the scale of capital expenditures, financing capacity, and whether dilutive financing occurs.