China internet’s cheap valuation has become a consensus, but capital returns still require earnings and AI monetization evidence

Overview

This report is based on a week-long roadshow feedback session with J.P. Morgan and U.S. investors, discussing the allocation barriers for the China internet sector after valuations becoming cheap became a consensus. The core conclusion is that U.S. investors have not rejected China internet, but broadly deferred transactions because AI hardware, storage, semiconductors, and private or newly listed AI assets still have clearer near-term earnings-revision or scarcity narratives. The sector needs to shift from 'cheap' to 'verifiable': stocks that can deliver cloud margin, AI revenue, unit economics, capital returns, or listing-related events that are observable in the next few quarters are more likely to attract incremental capital.

Core views

The report does not advocate betting on sector-wide beta; instead it supports stock-level opportunities with clear timing and quantifiable validation points. Alibaba is viewed as the clearest large-cap tactical target because cloud margin is closest to earnings-report verification, while Baidu has catalyst density from Kunlunxin and capital returns but still needs to prove AI platform quality and parent-company earnings quality. Tencent is well-regarded for quality, but lacks sufficient urgency; investors want disclosures rather than narratives, including how AI KPIs connect to revenue, margins, or user engagement. Meituan benefits from regulation, but unit economics still require validation in 2Q26 and 3Q26; JD may represent a cleaner earnings-revision expression of food-delivery rationalization. Zhipu’s relative preference over MiniMax is reinforced, but commercialization proof has replaced model quality as the key; Kuaishou’s Kling is in focus, and the debate is whether its value can truly accrue to shareholders of the listed company.

Analysis framework

The report applies an investor-feedback and stock-level catalyst mapping approach: first identify allocation constraints and opportunity costs, then categorize the sector into three groups, including stocks with near-term underwritten validation points, higher-quality names that need disclosure to restore urgency, and names with attractive assets but unproven equity transmission. The report also compares U.S. and Asian investor feedback, noting that the core difference is not regional disagreement; it is that investors are more willing to believe reportable evidence than macro narratives.

Methodology notes

• Investor feedback frameworkSelective positioning framework driven by validation

  Cheapness is not sufficient; capital re-entry requires verifiable catalysts

  The report redefines China internet from a valuation-rerating trade into a validation-driven trade, with emphasis on near-term earnings revisions, AI commercialization KPIs, cloud margins, unit economics, capital returns, and listing events.

• Relative positioning frameworkOpportunity cost and capital rotation framework

  China internet competes with AI hardware, storage, semiconductors, and scarce AI assets for capital

  U.S. investors recognize China internet is cheap, but when AI hardware and scarce AI assets still offer cleaner re-rating narratives, they are not willing to rotate purely on cheapness.

• Stock catalyst frameworkCalendar-based catalysts and reportable-variable framework

  Catalysts with timing, disclosure, and financial transmission are more likely to change behavior

  The report stresses that cloud margin, AI KPIs, unit economics, IPO/listing events, lock-up release, capital returns, and revenue disclosure are more important than broad sector valuation levels.

Asset mapping & comparison

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

• Tencent

  High-quality core platform, J.P. Morgan maintains Overweight

  Strengths

  Strong compounded earning power, controlled AI investment risk, depressed valuation multiple, and investor recognition of asset quality.

  Weaknesses

  The AI revenue path still appears longer-term, with insufficient urgency; investors prefer disclosures over narrative.

  Comparison

  Compared with Alibaba, Tencent’s AI validation points are more KPI- and revenue-transmission-based, less direct than cloud margin.

  Risks

  If Hunyuan, WeChat agents, WorkBuddy, cloud APIs, and ad monetization fail to link to reported revenue, margins, or engagement, AI optionality may remain excluded from core earnings.

• Alibaba

  The clearest large-cap tactical expression, J.P. Morgan maintains Overweight

  Strengths

  Cloud margin is a near-term, reportable, and quantifiable variable; external cloud growth is close to year-over-year 40%; first-time MaaS ARR disclosure improves visibility.

  Weaknesses

  The cloud margin long thesis is already understood by many investors; incremental edge lies in judging margin-improvement slope and reinvestment leakage.

  Comparison

  Compared with Tencent and Baidu, Alibaba’s cloud margin is more directly tied to near-term financial report variables.

  Risks

  If cloud margin improvement stalls, or is offset by local-services and consumer AI losses, current tactical appeal may weaken.

• Baidu

  Event-driven opportunity, J.P. Morgan maintains Overweight

  Strengths

  Kunlunxin, capital returns, Hong Kong primary listing, Stock Connect inclusion, first dividend, and potential special distributions offer relatively dense catalysts.

  Weaknesses

  Investors question whether this is an asset-monetization trade or a sustainable AI platform play; traditional search pressure and communication plus capex scale also affect confidence.

  Comparison

  Compared with sector beta, Baidu is more of an event-path trade; compared with Alibaba, long-term platform narrative still needs more proof.

  Risks

  If parent-company earnings quality, AI cloud revenue sustainability, and capital-return framework are unclear, long-only capital positioning may be constrained.

• Meituan

  Regulation beneficiary but Neutral

  Strengths

  SAMR subsidy-rule guidance is seen as directionally supportive for sector economics; management has a unit-economics target of about CNY1 per order.

  Weaknesses

  Regulatory support alone is not enough for an upgrade; the market needs confirmation that improvement reflects intrinsic strength rather than a temporary external window.

  Comparison

  JD is viewed by several investors as a cleaner earnings-revision expression of food-delivery rationalization.

  Risks

  If unit economics are closer to the bear-case CNY0.20–0.30 per order, or if JD first reflects earnings revisions, Meituan’s relative expression strength would decline.

• JD

  Alternative expression of food-delivery rationalization, J.P. Morgan maintains Overweight

  Strengths

  Investors actively cited JD as a cleaner carrier of earnings revisions; J.P. Morgan estimates upside above market consensus.

  Weaknesses

  JD was not a flagship name in this systematic roadshow, and feedback confidence is moderate rather than decisive.

  Comparison

  Under food-delivery regulation and rationalization, JD may offer a more direct earnings-revision surprise than Meituan.

  Risks

  If rationalization does not convert into JD earnings revisions, or the market still favors Meituan’s higher-value order share, the relative advantage may weaken.

• Zhipu AI

  Listed model-lab exposure favored over MiniMax, J.P. Morgan maintains Overweight

  Strengths

  Investors credit model quality, pricing power, capital access, and the financing flywheel; relative preference receives relatively good feedback.

  Weaknesses

  Long-term logic cannot rely only on general foundation-model leadership; it must prove paid demand in coding workflows, retention, and enterprise value capture.

  Comparison

  Compared with MiniMax, Zhipu is more favored by investors, but relative to potential assets like Kimi and DeepSeek, scarcity premium could compress.

  Risks

  Without demonstrated commercialization, pricing power, and customer retention, model-quality advantage alone is insufficient to support long-term valuation.

• MiniMax

  Weakest relative feedback, J.P. Morgan maintains Neutral

  Strengths

  If lock-up is absorbed smoothly, the stock may receive short-term tactical support.

  Weaknesses

  Investors focus on failed repricing, price floor formed by DeepSeek, Zhipu’s capital advantage, and lock-up release pressure.

  Comparison

  Compared with Zhipu, MiniMax relies more on scarcity and foundation-model differentiation, with weaker commercialization proof.

  Risks

  Without pricing power, customer retention, and commercialization evidence, scarcity is a fragile foundation.

• Kuaishou

  Kling hidden asset attracts interest, J.P. Morgan maintains Overweight

  Strengths

  Kling draws high investor attention, with investors citing private valuations around US$16–18bn and potential IPO valuations around US$25–30bn.

  Weaknesses

  The core business is not the investment focus; the main debate is how Kling’s value is transmitted to Kuaishou shareholders.

  Comparison

  Compared with Alibaba’s cloud margin and Baidu’s event path, Kuaishou’s key issue is equity linking rather than whether the asset itself is valuable.

  Risks

  If ownership, dilution, listing structure, parent-company discount, revenue trajectory, and model quality are unclear, Kling may remain an interesting asset rather than a clean stock thesis.

Key data

• U.S. investor roadshow timing2026年6月15日至6月22日U.S. feedback came from a one-week roadshow.
• Asian marketing window2026年4月初至6月下旬It is not fully comparable one-to-one with U.S. feedback; the report treats it as directional reference.
• Core sector judgmentCheapness is acknowledged, but urgency still requires proofInvestors focus on whether reportable evidence can compete quickly enough with AI hardware and other hot trades.
• Alibaba Cloud growth and disclosureExternal cloud growth close to year-over-year 40%, first disclosure of MaaS ARRSupports J.P. Morgan’s increased 12-month conviction on Alibaba and makes cloud margin the central validation point.
• Meituan fair valueHKD88The report keeps Meituan Neutral, arguing regulatory upside is not sufficient alone to justify an upgrade.
• Meituan unit-economics debateManagement target about CNY1 per order; bear-case around CNY0.20–0.30 per orderInvestors are focused on whether post-regulation the profit pool is protected or redistributed toward a more rationalized competitive environment.
• Kling investor valuation framingPrivate valuation around US$16–18bn; potential IPO valuation around US$25–30bnThe report explicitly states these are investor statements and unverified, not J.P. Morgan valuation targets.
• MiniMax lock-up focusExpected to unlock around 2026年7月8日Lock-up release, pricing power, customer retention, and model cycles are key investor focus points.
• 3Q26 observation standardAt least two large-cap China internet stocks show positive EPS revision breadthIf earnings revisions appear, valuation discount for capital sources may narrow; otherwise a light sector beta and stock-specific positioning should remain.

Impact & implications

The implication for investment decisions is that valuation support for China internet is real, but insufficient to trigger large-scale capital return on its own. Positioning should shift from sector beta to stock-specific evidence: prioritize whether Alibaba cloud margin improves consecutively, whether Baidu’s event path is delivered, whether Tencent AI KPIs enter revenue and margins, whether Zhipu can prove commercialization in workflow monetization, and whether food-delivery rationalization first converts into earnings revisions at JD or Meituan. If AI hardware, storage, or semiconductors disappoint on earnings, they could become external rotation catalysts for the sector; otherwise China internet must rely on its own reportable performance to win back capital.

Risks

• If AI hardware, storage, and semiconductor trades continue to provide clearer earnings revisions, China internet may remain a capital source rather than a destination.
• Sector-wide cheap valuation, if not accompanied by earnings revisions, AI monetization, or calendar-based catalysts, may fail to trigger broad capital reflow.
• Insufficient disclosure on Tencent AI commercialization may keep AI optionality excluded from core earnings for longer.
• If Alibaba cloud margin does not expand consecutively, or is offset by local-service and consumer AI losses, tactical attractiveness may fade.
• If Baidu cannot prove AI cloud revenue sustainability, parent-company earnings quality, and a capital-return framework, the event-driven narrative may not translate into a long-term repricing.
• If Meituan unit economics fall short of management target, or if JD first delivers earnings revisions, regulatory support may not effectively flow into Meituan’s stock price.
• Model-lab scarcity premium may compress as more assets list or become investable; both Zhipu and MiniMax need commercialization proof.
• MiniMax lock-up release, limited pricing power, and customer-retention uncertainty may create technical and fundamental pressure.
• If Kling’s value cannot be clearly attributed to Kuaishou shareholders, the hidden-asset narrative may fail to support the parent valuation.

What to watch

• By the end of 3Q26, whether at least two large-cap China internet stocks show positive EPS revision breadth.
• Whether Alibaba’s cloud margin improves consecutively and whether local-service and consumer AI losses are controlled.
• Whether Tencent discloses Hunyuan, WeChat agents, WorkBuddy, cloud API, and ad monetization KPIs that connect to revenue, margins, or user engagement.
• Whether Baidu events such as Kunlunxin IPO progress, Hong Kong primary listing, Stock Connect inclusion, first-ever dividends, or special distributions are delivered.
• Whether Baidu AI cloud revenue has durability rather than being one-off or event-driven.
• Whether Meituan’s unit economics in 2Q26 and 3Q26 prove improvement as an internal capability rather than a mere regulatory window.
• Whether JD first shows an earnings-revision surprise from food-delivery rationalization.
• Zhipu’s paid demand in coding workflows, pricing power, customer retention, and enterprise value capture.
• Supply pressure, repricing ability, and customer retention after MiniMax lock-up release around 2026年7月8日.
• Kling’s ownership, dilution, listing structure, revenue trajectory, model quality, and transmission to Kuaishou shareholder value.
• Whether AI hardware, storage, and semiconductors disappoint on earnings, forcing capital rotation back into China internet.