The marketable signal is dispersion, not broad China beta. Earnings power is concentrating in export-linked and AI-adjacent industrial chains, which implies margin expansion for a narrow set of names while the rest of the China complex keeps earning a lower multiple because domestic demand is still the binding constraint. That is usually more durable than a one-off data print because it shifts capital allocation toward capex, networking, power equipment, and materials rather than households.

For NVDA, this is supportive only indirectly: it reinforces that the AI buildout is still converting into supplier profit pools even in a weaker macro backdrop. The second-order effect is that the trade increasingly rewards earnings revision momentum in semis, networking, and optical interconnect rather than just headline AI enthusiasm; if hyperscaler capex stays firm over the next 1-3 months, those names should keep outperforming, but any slowdown in cloud spend or export restrictions would hit the multiple quickly.

The contrarian read is that consensus may be overestimating the breadth and speed of any China stimulus. Fiscal support can stabilize the margin floor for heavy industry, but it does not repair household balance sheets or property wealth on a 1-3 month horizon, so consumer-facing and property-sensitive equities remain vulnerable to value traps. Watch for a package large enough to move retail sales, credit growth, and property transactions; absent that, this is a selective industrial upcycle, not a clean China recovery.