China Internet Sees Only Selective Rotation; AI Monetization and Earnings Validation Remain Key Hurdles

Overview

Based on marketing feedback from Singapore investors, JPMorgan believes sentiment toward China internet has improved, but there has been no broad strategic rotation. Investors primarily view large-cap China internet stocks as tactical alternatives after weakening Asian semiconductor earnings momentum, with focus concentrated on Alibaba and Tencent. Further sector re-rating requires upward earnings forecast revisions from more companies, broader fund flows, and empirical evidence that AI investment is translating into attributable revenue, margins, and cash returns.

Core views

Valuation is not investors' primary entry point; the market places greater emphasis on whether AI investment can generate verifiable revenue and reasonable returns before costs weigh on earnings and free cash flow. Alibaba's near-term catalysts come from its cloud business, while Tencent's key lies in AI commercialization disclosure; JD is seen as a short-term earnings recovery trade at a low valuation; Meituan has greater long-term value potential but also greater uncertainty from food-delivery competition. Among model companies, Zhipu is tactically preferred over MiniMax, but both face validation pressure regarding model iteration, low-price competition, revenue quality, and cash burn.

Analysis framework

Through discussions with Singapore long-only, hedge fund, and regional allocation investors, the report compares market expectations for different companies in terms of earnings forecasts, fund flows, AI commercialization, returns on capital expenditure, competitive landscape, and valuation sustainability, and establishes verifiable observation criteria through September and November.

Methodology notes

• Investor researchMarketing feedback and holdings analysis

  Assess whether investors have developed an industry overweight and their holding preferences

  Covers long-only investors, hedge funds, and long-term allocators, distinguishing 3- to 6-month earnings revision trades from long-term AI commercialization allocation logic.

• Fundamental validationAI monetization and capital return assessment

  Test returns on AI investment through revenue attribution, margins, and cash conversion

  Focuses on whether cloud, agent, or AI product revenue can be disclosed and attributed, as well as the impact on free cash flow and shareholder returns after increases in capital expenditure, depreciation, and operating expenses.

• Market rotationEarnings forecast revision and fund flow tracking

  Validate whether rotation from semiconductors to internet has breadth and sustainability

  Requires positive internet earnings revisions and fund flows to expand from Alibaba and Tencent to at least two other large internet companies, while semiconductor revisions weaken.

Asset mapping & comparison

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

• Alibaba Group Holding Limited (BABA)

  Viewed by investors as a relatively preferred large-cap China internet name over the next 3 to 6 months

  Strengths

  External cloud growth and MaaS/Qwen monetization provide a relatively near-term path to validation.

  Weaknesses

  Non-AI businesses are weak, while losses in All Others and visibility on group free cash flow remain limited.

  Comparison

  Versus Tencent, the market believes its cloud business is more likely to provide near-term earnings validation.

  Risks

  Compute supply constraints, disputes over accounting presentation, partial pricing-in of cloud re-rating, and margin improvement falling short of expectations.

• Tencent (0700)

  Suitable for accumulation over 6 to 12 months rather than a short-term AI trade

  Strengths

  Has product distribution advantages, and the market continues to recognize its core businesses and long-term AI potential.

  Weaknesses

  Work Buddy has not disclosed paying users, retention, ARPU, or product revenue.

  Comparison

  Near-term catalysts are weaker than Alibaba's, but its long-term platform and distribution capabilities are recognized.

  Risks

  Rising capital expenditure, depreciation, and operating expenses may pressure free cash flow and buybacks; returns on AI capital remain unproven.

• JD.Com, Inc. (9618)

  A low-valuation earnings recovery trade over 3 to 6 months

  Strengths

  A valuation of approximately 8 to 9x earnings supports tactical positioning, and the near-term earnings beat thesis is relatively clear.

  Weaknesses

  The market is unwilling to assign significant terminal value to the delivery business, limiting its long-term investment appeal.

  Comparison

  Versus Meituan, its short-term earnings recovery thesis is cleaner; Meituan has greater terminal value potential.

  Risks

  Whether cost discipline can be maintained after earnings recover, and whether founder-led spending increases again.

• Meituan (3690)

  Offers greater long-term terminal value potential but with higher uncertainty

  Strengths

  If competition eases, improving food-delivery unit economics could provide earnings upside.

  Weaknesses

  Investors have widely divergent expectations for steady-state profitability in the delivery business.

  Comparison

  Versus JD, short-term earnings certainty is weaker, but long-term value upside may be greater.

  Risks

  Competitor subsidies are an exogenous variable; sustainable positive delivery economics and high-value order share need to be validated.

• Z AI Co Ltd - H (2513.HK)

  Viewed as the more preferred tactical listed large-model exposure

  Strengths

  Investors focus on its enterprise demand, coding capabilities, pricing, and recurring revenue potential.

  Weaknesses

  The market is unwilling to pay a premium for a permanently independent leading position.

  Comparison

  Continues to outperform MiniMax among model companies; DeepSeek sets capability and low-price boundaries, while Kimi reduces scarcity.

  Risks

  The frontier-model position changes rapidly; unstable API demand, pricing, or inference efficiency would weaken the valuation thesis.

• MiniMax Group Inc - H (0100.HK)

  Less preferred among model companies than Zhipu and Kimi

  Strengths

  Multimodal capabilities may still offer a path to differentiation.

  Weaknesses

  Disclosure on recurring revenue, gross margin, cash burn, and dilution is insufficient.

  Comparison

  Faces DeepSeek's capability and pricing boundaries, while ByteDance and Kuaishou possess stronger distribution, data, and monetization capabilities in similar areas.

  Risks

  If the next two major product releases fail to return to the frontier while sustaining paid demand, the investment thesis will be challenged; no public release dates are available.

Key data

• Alibaba external cloud growthApproximately 40%Level disclosed as of May; investors want to see growth remain close to 50%.
• Alibaba MaaS annualized recurring revenueMore than RMB 8 billionManagement's target is to exceed RMB 30 billion by year-end.
• Alibaba preferred time horizon3 to 6 monthsCloud business is viewed as an AI monetization path closer to validation.
• Tencent preferred time horizon6 to 12 monthsThe market recognizes its distribution advantages but awaits evidence of AI revenue and capital returns.
• JD valuationApproximately 8 to 9x earningsInvestors position it as a short-term earnings recovery trade.
• Meituan high-value order share observation targetMaintain in the high-60% range or higher in 3Q26At the same time, the food-delivery business needs to demonstrate sustainable positive unit economics.

Impact & implications

In the short term, if Asian semiconductor earnings momentum slows, capital may selectively rotate toward large China internet stocks with cloud and AI catalysts, prioritizing Alibaba and Tencent; however, this does not equate to a reversal in the industry trend. Investment decisions should focus more on company-level revenue quality, cost discipline, free cash flow, competitive subsidies, and repeatable AI commercialization disclosure, rather than relying solely on valuations or Southbound fund support.

Risks

• China internet earnings forecast revisions remain negative, preventing selective rotation from broadening.
• Asian semiconductor earnings momentum reaccelerates, causing capital to rotate back into semiconductors rather than internet.
• AI revenue cannot be clearly attributed, or revenue growth is insufficient to cover capital expenditure, depreciation, and operating costs.
• Cloud compute supply, price competition, and rapid model iteration compress AI business margins.
• Subsidy competition in the food-delivery industry persists, weakening delivery-business unit economics.
• Company-level free cash flow, buyback capacity, cost discipline, or disclosure falls short of market expectations.

What to watch

• By September, whether at least two large internet companies beyond Alibaba and Tencent show positive earnings forecast revisions and improved fund flows.
• Whether semiconductor earnings forecast revisions continue to weaken or reaccelerate.
• By November, whether major platforms disclose attributable cloud, agent, or AI product revenue, together with evidence on margins or cash conversion.
• Whether Alibaba external cloud growth can remain close to 50%, margins can improve, and All Others losses can be controlled.
• Whether Tencent can disclose attributable monetization from Work Buddy or Xiaowei while maintaining shareholder returns.
• Whether JD's new-business losses decline without a new round of aggressive investment, and whether Meituan's food-delivery economics and high-value order share improve.
• Subsequent model releases, paid demand, API pricing, and inference-efficiency performance of Zhipu and MiniMax.