BlackRock says the China AI play is stock-specific, not a regional trade

BlackRock Investment Institute expects most AI “winners” to be U.S. stocks, with only select China names, and kept its view on Chinese equities neutral while staying overweight the U.S. The Nasdaq is up a little over 12% YTD versus the mainland China ChiNext up more than 20%, but the broader MSCI China index is down over 10% as profit visibility remains challenged by slower growth and competition. BlackRock argues “cheap, open-source AI” may boost adoption but not necessarily AI-provider profitability, while highlighting opportunities in “physical AI” (AI embedded in robotics/hardware) and preferring stocks exposed to scarce inputs like infrastructure. Overall stance tilts to buying U.S. AI leaders given U.S. strength in chips, frontier models, and capital markets.

Analysis

This is less a new fundamental signal than a reinforcement of an existing capital-allocation trend: global AI beta continues to concentrate in U.S. ecosystems where earnings visibility, margin protection, and liquid financing are strongest. The market implication is that broad China tech exposure remains a value trap unless a company has a clearly exportable moat in semis, batteries, or industrial hardware; otherwise, cheaper multiples can stay cheap if revenue quality and buyback capacity lag.

The more interesting second-order trade is not software vs. software, but monetization vs. adoption. If open-source AI keeps compressing model-level economics, the profit pool shifts toward scarce inputs: advanced chips, packaging, networking, power, data-center infrastructure, and robotics/automation hardware. That argues for relative strength in U.S. and select Taiwan/Korea supply-chain names, while broad Chinese AI providers may see user growth without corresponding EBITDA expansion.

For China, the biggest risk is that policy support accelerates usage but not equity returns: with slower nominal growth, intense domestic competition, and limited pricing power, incremental AI spend can simply deepen the race-to-the-bottom. The contrarian angle is that some of the best asymmetric exposure may be in "physical AI" names and industrial suppliers, not headline AI software winners; if that thesis is right, the current market is still underweight the enablement layer. Falsifier: a credible step-up in Chinese AI monetization, margin guidance, or foreign inflows into select China tech over the next 1-3 quarters.

Key Decisions for Investors

• Relative-value trade: long QQQ or SOXX vs short FXI or KWEB for 1-3 months; thesis is continued U.S. earnings durability vs China multiple compression. Cover if MSCI China re-rates on a sustained earnings revision cycle or policy-driven revenue acceleration.
• Express the 'picks-and-shovels' view with a basket long in infrastructure/semis over AI software: long SMH, ASML, ANET, and VRT versus a basket short of unprofitable AI application names; best entry is on any post-earnings pullback in the winners.
• Treat BLK as a sentiment beneficiary, not a direct AI winner: modest long only if flows into U.S. large-cap and thematic ETFs accelerate; otherwise the signal is too soft to justify a standalone position.
• Fade broad China AI enthusiasm unless confirmed by earnings: short-rally sell FXI/KWEB into policy headlines, with a stop if top-tier Chinese internet names show meaningful margin expansion or ad/revenue inflection over the next 1-2 quarters.