J.P. Morgan Global China Summit feedback supports thematic allocation to AI, robotics, and Tier-1 CBD real estate

Overview

This is a J.P. Morgan China equity strategy conference feedback report. Based on client discussions, company meetings, and panel views during the Global China Summit, it emphasizes using broad consumption as a funding source to allocate toward themes with greater flexibility such as AI, robotics, and Tier-1 CBD developers. The report provides an integrated assessment of China's AI commercialization, capex spillovers, improving A-share industrial orders, consumer structural divergence, and selective opportunities in real estate.

Core views

The core views include: first, China AI is still in the catch-up phase for capex and commercialization, and continued improvement in domestic chips and domestic LLMs will enhance the attractiveness of China AI assets; second, global AI capex is spreading from 3C and semiconductors into more industrial links, supporting a structural bullish view on factory automation and robotics chains; third, it is not yet appropriate to broadly add to consumption and internet positions, as more tangible domestic macro consumption improvement and more EPS upside surprises are needed; fourth, real estate exposure should remain limited to a selective preference for premium developers in Tier-1 city CBDs.

Analysis framework

The report uses a top-down thematic allocation framework, combining summit client feedback, panel discussions, company meeting observations, A-share capex and inventory data, sector rotation monitoring, and AI and robotics stock screening to judge that capital may rotate from broad consumption toward AI, robotics, and quality real estate developers.

Methodology notes

• Thematic allocationStructural thematic allocation to AI and robotics

  Treat AI capex, domestic LLM commercialization, and improving orders in robotics and factory automation as the main line of China equity allocation.

  The report argues that AI is a long-term technological transformation. If domestic models and infrastructure cross a commercialization inflection point, they may attract incremental capital into China AI tech positions, while second-order capex spillovers support industrial automation.

• Conference feedbackFront-line feedback from the Global China Summit

  Use client discussions, company meetings, and panel sessions to summarize investor focus areas and industry signals.

  The report translates investor discussions on AI, green shoots in real estate, slowing consumption, and internet earnings into sector allocation judgments rather than relying only on historical market performance.

• Factors and screeningStock screening and sector rotation monitoring

  Use indicators such as free-float market cap thresholds, A-share capex and inventory growth, and sector rotation changes to screen theme-related names.

  The report lists A-shares in the AI industry with free-float market cap above US$5bn, as well as A-shares in humanoid and industrial robotics with free-float market cap above US$3bn, while also observing semiconductor leadership concentration and fading liquidity in cyclical sectors.

Asset mapping & comparison

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

• China AI technology and infrastructure-related stocks

  Core overweight theme

  Strengths

  Domestic chips, domestic LLMs, and AI infrastructure continue to improve; if a commercialization inflection point emerges, valuations may become more AI-driven.

  Weaknesses

  Advanced GPU supply is constrained, and domestic substitution still lags in compute throughput, energy efficiency, and high-speed memory capability.

  Comparison

  Global investors' AI exposure is currently more concentrated in the U.S., Korea, Taiwan, and Japan, leaving room for potential incremental positioning in China AI.

  Risks

  LLM commercialization is slower than expected, capex is insufficient, chip restrictions intensify, and the model capability gap widens.

• Robotics and factory automation chain

  Second-order beneficiary of AI capex

  Strengths

  Orders in thermal management, precision machine tools, injection molding, power infrastructure, automotive equipment, and production line upgrades may be driven by AI capex spillovers.

  Weaknesses

  Order realization, delivery pace, and price transmission still require further verification.

  Comparison

  Compared with beneficiaries focused only on semiconductors and 3C, the industrial chain has broader exposure to benefits, but earnings elasticity may depend more on order conversion.

  Risks

  Capex cycles are interrupted, inventory pressure rises again, end demand is insufficient, and competition limits price transmission.

• Consumer and internet stocks

  Source of funds and selective watchlist direction

  Strengths

  There are still structural opportunities in lower-tier city and township consumption upgrades, healthier consumption, and quality high-growth consumer leaders.

  Weaknesses

  Retail sales growth is slowing, and the sector lacks broad upward momentum.

  Comparison

  The report relatively prefers AI and robotics themes and believes consumption and internet need more EPS upside surprises before positions should be increased.

  Risks

  Macro consumption recovery falls short of expectations, valuation premiums decline, and earnings upside surprises are insufficient.

• Premium real estate developers in Tier-1 city CBDs

  Selective allocation direction

  Strengths

  Prime locations and premium projects may be among the first to benefit from green shoots in the property market.

  Weaknesses

  A broad real estate recovery remains unstable, and investors are still evaluating whether improvement can be sustained.

  Comparison

  The report does not advocate broadly adding real estate exposure, but instead prefers developers focused on Tier-1 CBD areas.

  Risks

  Recovery in real estate sales is interrupted, policy support is insufficient, and household income and consumer confidence weaken.

Key data

• China AI capex relative to U.S. scaleAbout 20%The report says China's annual AI capex is significantly below the U.S., mainly due to advanced chip constraints, the performance gap in domestic substitution, and still-immature demand-side commercialization.
• Domestic LLM capability gapAbout 8 to 12 months behind leading U.S. systemsThe report uses GPT, Gemini, and Claude as capability benchmarks and believes China LLMs have not yet fully entered a self-reinforcing commercialization inflection point.
• DeepSeek V4-Pro price comparisonAbout one-fifth of the blended price of GLM-5.1 / Kimi K2.6The report believes highly cost-effective models help improve the capital appeal of China AI assets.
• 1Q26 A-share capex growthUp 4% year over year excluding financials and real estateThe report views this as an early signal of an upward cycle in A-share capex.
• 1Q26 A-share inventory growthUp 8% year over yearInventory improvement together with rising capex supports expectations for better industrial orders, delivery conversion, and pricing transmission.
• Capex growth excluding utilities and energyUp 7% year over yearThe report believes this measure better reflects improvement in industrial and manufacturing-related demand.
• A-share screening threshold for AI themeFree-float market cap above US$5bnUsed to identify A-share names in the AI industry with scale and liquidity.
• A-share screening threshold for humanoid and industrial roboticsFree-float market cap above US$3bnUsed to identify investable names in robotics and industrial automation chains.

Impact & implications

The investment implication is that China equity allocation should shift from broad consumption and short-term cyclical beta toward AI infrastructure, domestic LLMs, robotics, and factory automation chains with stronger long-term industrial logic; consumption and internet should not be completely avoided, but clearer macro consumption improvement and earnings upside surprises are needed; in real estate, the risk-reward is more concentrated in premium developers in Tier-1 city CBDs rather than a broad real estate recovery trade.

Risks

• China's AI capex remains persistently below that of the U.S., causing infrastructure and model iteration to lag expectations.
• Advanced chip restrictions and the performance gap in domestic substitution raise total cost of ownership for data centers, slowing the expansion of large-scale training and inference clusters.
• The domestic LLM commercialization revenue pool fails to expand quickly enough to form a positive feedback loop between capex and revenue growth.
• Macro consumer data and internet EPS performance fall short of expectations, keeping broad consumption and internet sectors under pressure.
• Green shoots in the property market may prove unsustainable, especially against a backdrop of slowing retail sales, causing repeated reversals in property-related trades.
• Orders, deliveries, and price transmission from AI capex spillovers into industrial chains are weaker than expected.
• A decline in market liquidity or reversal in sector rotation undermines valuation support for AI, semiconductors, and robotics themes.

What to watch

• Whether the gap between China's AI capex and that of the U.S. narrows.
• Progress in domestic chips on compute throughput, energy efficiency, and high-speed memory capability.
• Changes in capability, pricing, and commercialization revenue for domestic LLMs such as DeepSeek, GLM, and Kimi.
• Whether valuations of Chinese hyperscalers begin to be driven more clearly by AI.
• Whether A-share capex, inventories, orders, and earnings revisions continue to improve.
• Whether macro consumption data, retail sales, and internet company EPS show clearer upside surprises.
• Whether sales and prices of premium CBD real estate in Tier-1 cities continue to improve.
• Whether semiconductor leadership concentration and liquidity spillovers from cyclical sectors continue or fade.