China AI has entered a new stage of application and profit realization

Overview

This report is a thematic study covering the China AI ecosystem, macro impact, and Greater China equity opportunities. It argues that China AI has entered an “AI 2.0” stage: market focus is shifting from frontier training capability to inference, application rollout, and profit realization. Instead of simply chasing global frontier models, China is pushing AI to spread quickly across real-economy areas such as energy, manufacturing, internet platforms, consumer, urban infrastructure, robotics, and autonomous driving through low-cost models, an open ecosystem, a large number of use cases, manufacturing supply-chain strength, and policy coordination.

Core views

Core views include: first, China AI’s competitive focus is moving from model leadership to deployment speed, cost efficiency, and system-level integration; second, the financial impact of AI adopters is starting to show, mainly through cost efficiency, operating leverage, and margin expansion rather than an immediate revenue breakout; third, AI can become a medium-term productivity lever for China, but short-term net contribution to aggregate growth may be limited because capital expenditure and early efficiency gains are partly offset by labor-adjustment frictions; fourth, investment opportunities are expected to expand from early AI enablers to the application layer, foundation models, semiconductor localization, data centers, power, server supply chains, and AI adopters with clear monetization paths.

Analysis framework

The report combines top-down thematic selection with bottom-up analyst validation. Morgan Stanley uses the Global AI Mapping Survey to build an initial pool of AI exposure among stocks in its Greater China coverage universe, focusing on AI exposure categories, changes in materiality, and next 12-month EPS impact, then industry analysts evaluate how AI can translate into revenue growth, cost efficiency, or margin expansion, and exclude “AI washing” ideas that have narrative but lack credible monetization paths.

Methodology notes

• Theme stock screeningChina AI 65

  China AI thematic stock framework covering the full AI value chain

  This framework is used to identify companies with substantive and monetizable AI exposure in power, semiconductors and hardware, infrastructure, foundation models, and the application layer, and maintains a fixed constituent set to compare the performance of different thematic baskets.

• Survey and mappingGlobal AI Mapping Survey

  Mapping AI exposure and financial materiality

  The report uses a global AI mapping survey to track how AI adoption is spreading across Morgan Stanley’s coverage universe and to compare whether AI exposure and materiality have increased across different survey waves.

• Macro estimationAI productivity impact framework

  Impact of AI on total factor productivity and potential GDP

  The report treats AI as a medium-term productivity tool and estimates its cumulative impact on China’s total factor productivity and 2035 potential GDP levels, while emphasizing that short-term J-curve adjustment costs exist.

Asset mapping & comparison

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

• MiniMax (0100.HK)

  Chinese AI foundation model provider

  Strengths

  The report identifies MiniMax as one of China’s key foundation model providers, benefiting from the shift of Chinese AI from training to inference and applications.

  Weaknesses

  Foundation model commercialization is still in an early stage, and enterprise-side monetization pathways may be uneven.

  Comparison

  Listed alongside Z.ai as one of the key foundation model providers mentioned in the report.

  Risks

  Model competition, inference cost, commercialization pace, and the regulatory environment could all affect valuation.

• Z.ai (2513.HK)

  Chinese AI foundation model provider

  Strengths

  The report identifies Z.ai as one of the key companies in China’s foundation model space.

  Weaknesses

  Needs to prove sustained model capability, customer adoption, and revenue monetization.

  Comparison

  Together with MiniMax, it represents the core Chinese foundation model theme.

  Risks

  Rapid technology iteration, changing competitive dynamics, and enterprise-side payment uncertainty.

• ALIBABA GROUP HOLDING LTD (BABA.US)

  Full-stack AI platform

  Strengths

  The report considers Alibaba the most advantaged full-stack AI platform in China, with synergies across cloud, models, platforms, and application ecosystems.

  Weaknesses

  AI cloud and model investments require sustained capex, and margin trajectory may be affected by competition.

  Comparison

  The report sees Tencent as stronger in the application layer, while Alibaba is stronger in full-stack capabilities.

  Risks

  Cloud competition, capex payback cycle, model commercialization, and broader macro-consumer conditions.

• Beisen (9669.HK)

  AI adopter

  Strengths

  The report expects its AI ARR to grow from over RMB 6 million in FY25 to over RMB 60 million in FY26, with significant incremental profit potential and room to gain market share.

  Weaknesses

  The revenue base remains small, and sustained AI product paid growth needs to be proven.

  Comparison

  It is highlighted alongside Meitu and Beijing Roborock in the AI adopter risk-return screen.

  Risks

  Enterprise software budgets, AI feature adoption rates, and intensifying competition.

• Meitu (1357.HK)

  AI adopter

  Strengths

  The report expects monetization through photo, video, and design applications, with paid conversion potentially doubling to 8% to 10% by 2028.

  Weaknesses

  Monetization in consumer applications still requires ongoing validation.

  Comparison

  Along with Beisen and Beijing Roborock, it is among the AI adopters with relatively better risk-return.

  Risks

  User growth, subscription conversion, competition in AI content tools, and regulation.

• Beijing Roborock Technology (688169.SS)

  AI adopter and consumer robotics

  Strengths

  Advanced AI algorithms can support a premium positioning, faster replacement cycles, and margin expansion, with broader consumer service-robot potential.

  Weaknesses

  Demand for consumer hardware and product upgrade cycles can be volatile.

  Comparison

  Shares AI-enabled perception, navigation, and robot upgrade benefits with Ecovacs Robotics.

  Risks

  Competition, pricing pressure, overseas demand, and hardware innovation cycles.

• Ecovacs Robotics (603486.SS)

  AI adopter and service robotics

  Strengths

  Upgrades in AI perception and navigation can support product up-tiering and TAM expansion.

  Weaknesses

  Realizing premiumization depends on product experience and consumer willingness to pay.

  Comparison

  Along with Beijing Roborock, it belongs to the consumer-robot AI application benefit theme.

  Risks

  Industry competition, demand weakness, inventory, and pricing pressure.

• Cambricon, Iluvatar, NAURA, AMEC, ACMR, SMIC

  AI semiconductor localization enablers

  Strengths

  The report lists these as key AI enablers benefiting from China’s long-term semiconductor localization trend.

  Weaknesses

  Advanced process nodes, EDA tools, and manufacturing capacity may still remain bottlenecks.

  Comparison

  Compared with application-layer companies, these firms benefit more directly from AI capex and localization of the supply chain.

  Risks

  Technology bottlenecks, export restrictions, capacity constraints, valuation volatility, and the capex cycle.

• CATL, Yingliu, Sieyuan

  AI power and energy infrastructure beneficiaries

  Strengths

  The report highlights AI power demand as a key theme, with growth in data-center demand and stored, flexible power value potentially creating incremental opportunities.

  Weaknesses

  The release pace of power and storage demand is linked to data-center build cycles.

  Comparison

  Compared with semiconductor and model companies, the benefit is more in AI infrastructure adjacency.

  Risks

  Power infrastructure approvals, demand forecast errors, pricing risk, and policy risk.

Key data

• Long-term impact of AI on China’s total factor productivityCumulative increase of about 3 percentage points over the next ten yearsThe report estimates AI can partially offset aging demographics and shrinking labor pressure.
• 2035 potential GDP impactabout 3.5 percentage points higher versus a no-AI adoption scenarioThis effect is a medium-term productivity story, not short-term cyclical stimulus.
• China AI semiconductor self-sufficiency rateabout 41% in 2025, estimated about 86% in 2030Higher self-sufficiency supports more resilient and cost-efficient AI deployment.
• enabler/adopter penetrationfrom about 43% to about 51% over two yearsIndicates AI adoption continues to spread across covered companies.
• 1H26 China CIO Survey47% plan to launch their first AI project in the next 12 monthsHigher than 40% in 2H25, reflecting stronger willingness to launch AI projects.
• AI adopter NTM EPS performanceup about 62% over the past two yearsSignificantly above MSCI China’s roughly 10% performance.
• AI adopter EBIT marginexpected to expand by 12 to 13 percentage points to 16% to 17% by 2027EThe report views AI currently as mainly delivering efficiency gains and operating leverage.
• China AI cloud growthabout 72% CAGR from 2024 to 2029ECloud and data-center infrastructure is a key support for inference and application diffusion.
• Major technology companies' AI-related capexestimated RMB 5970 billion in 2026 and RMB 7110 billion by 2030The report corresponds to about 10% CAGR and expects AI-related capex to contribute about 0.2 to 0.3 percentage points to real GDP growth over the next two years.

Impact & implications

From an investment perspective, the report believes AI enablers and foundation models remain the core themes at this stage, but as AI adoption spreads, the application layer and AI adopters with clearly monetizable paths are expected to offer catch-up opportunities. Infrastructure, power, semiconductor localization, AI server supply chains, and data centers benefit from the capex cycle; among internet platforms, Alibaba is viewed as the best-positioned full-stack AI platform, while Tencent has advantages in application-layer monetization and ecosystem monetization. In foundation models, MiniMax and Z.ai are cited as key Chinese foundation model providers. At the macro level, AI is expected to improve medium-term productivity, but in the next 2 to 3 years it may face employment substitution, earnings pressure, and deflation risk, requiring policy buffering.

Risks

• Enterprise-side AI monetization is still in an early and uneven stage, and revenue growth may lag behind adoption growth.
• Constraints in advanced chips, EDA tools, and manufacturing nodes may continue to limit China’s AI frontier innovation.
• AI adoption may first create job displacement and earnings pressure over the next 2 to 3 years, with productivity gains lagging behind.
• Data transparency varies by industry; the screening framework may bias toward companies with impacts that are more visible in the short term and underweight early-stage opportunities.
• Rapid technological change may alter competitive positioning, exposing current leaders to displacement risk.
• Semiconductor constraints, infrastructure supply, policy changes, and macro demand weakness could all affect theme performance.

What to watch

• The share and pace of Chinese enterprises moving AI projects from pilots to production environments.
• Whether AI adopters’ NTM EPS, EBIT margin, and ROE continue to improve relative to MSCI China.
• Whether AI capex for AI cloud, data centers, and major technology companies is on track with the report’s forecast.
• Progress in localization of Chinese AI chips and semiconductor equipment, especially whether downstream bottlenecks ease after SMIC expansion.
• Whether power, storage, and data-center supply can support rising AI inference demand.
• The extent of policy cushioning for AI-related job displacement, and the adjustment speed in services and cognitive roles.
• Commercialization progress of platforms and model players such as Alibaba, Tencent, MiniMax, and Z.ai.