Analysis-With tariffs stalled, Trump's China policy drifts

Trump's China strategy is described as adrift, with tariffs rising to around 145% before easing into an uneasy détente and no coherent Plan B emerging after a Supreme Court ruling invalidated many duties. The article highlights contradictory policy signals, including approval of Nvidia H200 AI chip sales to China shortly after national security warnings and a brief Pentagon blacklist of Chinese tech firms that was later withdrawn. The trade deficit with China fell 32% to $202 billion in 2025, but the broader effect has been policy confusion rather than a clear strategic reset.

Analysis

The market implication is less about a clean China escalation and more about policy optionality collapsing into noise. That usually benefits the most state-protected parts of the semiconductor stack in the short run, but it raises the equity risk premium for the entire AI supply chain because customers cannot underwrite capex, export permissions, or unit economics when rules can reverse intra-day. NVDA is the most exposed not because of one headline, but because China demand, product mix, and regulatory uncertainty now interact to cap multiple expansion even if shipments remain partially allowed.

Second-order winners are non-China replacement ecosystems: domestic tooling, defense electronics, and non-China industrial automation names that can absorb reshoring and export-control leakage. The more important loser is not just Chinese end demand, but U.S. and allied OEMs that planned on China as a margin bridge while U.S. manufacturing capacity ramps slowly; that bridge is now politically unstable. Over 6-18 months, this should widen dispersion between companies with low China revenue and those with meaningful AI/data-center exposure to Chinese enterprise spending.

The key risk is that the administration eventually settles into a managed-trade détente that looks dovish for semis but still keeps strategic restrictions in place. If that happens, the immediate headline downside to NVDA fades, but the valuation ceiling remains because investors will not pay peak-growth multiples for a business subject to recurring licensing shocks. The stronger tail risk is a sudden re-tightening after a rare-earth or Taiwan-related flare-up, which could hit China-exposed semis and networking names within days, not quarters.

Consensus may be underestimating how much the policy incoherence itself suppresses re-shoring and capex decisions. That is bearish for equipment suppliers and bullish for incumbents with existing installed base because customers delay large commitments until the regime stabilizes. In other words, this is less a clean punitive trade than a volatility regime: earnings can survive, but multiple compression becomes the dominant channel.

Key Decisions for Investors

• Trim/underweight NVDA into any policy-led rallies; use a 1-3 month horizon and prefer selling upside via call spreads rather than outright shorts because headline reversals can gap the stock higher on any détente signal.
• Pair trade: long defense electronics / domestic industrial automation basket vs short China-exposed semiconductor leaders over 3-6 months; the thesis is not revenue collapse, but lower terminal multiples for names with recurring China uncertainty.
• Buy NVDA downside hedges via 3-6 month put spreads around the next policy milestone; risk/reward favors defined-risk hedges because implied vol typically lags actual policy jump risk.
• Add exposure to U.S. supply-chain substitution beneficiaries on pullbacks, with a 6-12 month view; these names should see better order visibility as firms de-risk China even if the macro backdrop stays muddled.
• Avoid broad semicap exposure until there is evidence of a durable rules framework; the asymmetry is poor because downside comes from multiple compression, while upside is capped by lingering export-control risk.