Why the US-China AI race won’t be won any time soon

The US has rolled out a series of high-profile AI initiatives to compete with China, but key structural choke points — notably China's dominance in rare earths and constraints around advanced lithography equipment — mean neither side is likely to secure a decisive lead soon. These supply-chain and technology bottlenecks, reinforced by export controls and industrial policy, point to a prolonged strategic contest with material implications for semiconductor equipment makers, mining companies and firms exposed to trade controls.

Analysis

Market structure: chokepoints (rare earths processing concentrated in China; extreme-lithography supply concentrated at ASML) create durable rents for non-China suppliers who can prove secure supply. Expect non-Chinese rare-earth producers/processors and Western equipment makers to see 20–40% higher realized pricing power over 12–24 months as buyers diversify; Chinese foundries (SMIC) and China-dependent hardware OEMs lose bargaining power and access to high-end chips.

Risk assessment: tail risks include a full export embargo (Dutch/US escalate) or Chinese retaliation (tariffs, raw-material export limits) that could cause a 30%+ shock to affected revenues in weeks. Immediate market sensitivity will be headline-driven (days–weeks); structural reshoring and capacity build-out will play out over 12–36 months. Hidden dependencies: downstream refining tech, permitting/geopolitics for new mines, and ASML political approvals are bottlenecks that determine who wins.

Trade implications: favor materials processors and Western lithography/semiconductor-capex exposure, while trimming China-exposed pure-play fabs and consumer hardware. Volatility will concentrate in semis and materials — use directional equity + option structures to express views and hedge policy risk. Monitor concrete catalysts that will reprice positions: ASML export approvals, CHIPS Act fund tranches, and Lynas/M.P. processing plant commissioning dates.

Contrarian angles: consensus underestimates time-to-scale for non-China rare-earth refining (likely 18–36 months) so near-term price spikes could be larger than models assume; conversely ASML upside may be capped by political gating so multiples could compress if approvals stall. Historical parallel: energy choke points in 1970s show fiscal/strategic responses create long lags before supply eases — plan for multi-quarter illiquidity and policy-driven repricing.

Key Decisions for Investors

• Establish a 2–3% long position in Lynas Rare Earths ADR (LYCFF) within 1 month, scale to full size if rare-earth price index rises >15% or Lynas announces a new non-China processing contract; target 12-month upside 30–50%, stop-loss -20%.
• Initiate a 1.5–2% long in ASML Holding (ASML) via 18–24 month LEAP calls (buy Jan 2027 5–10% OTM calls) to capture secular lithography scarcity; hedge near-term regulatory headline risk by selling short-dated calls up to 30% of position if implied vol >35%.
• Open a 1% short on SMIC (OTC: SMIFF) or 0.5–1% underweight China-dependent fabs for 6–12 months, increasing if new US/Dutch export controls intensify or if TSMC/ASML shipment approvals are publicly tightened; cover on reversal of policy within 90 days.
• Rotate 3–5% portfolio allocation from China internet/consumer names (e.g., BABA, TCEHY) into US defense primes (LMT, NOC) and semiconductor-capex suppliers (AMAT, LRCX) over the next 3 months; re-evaluate if CHIPS Act disbursements exceed $20B in a tranche or rare-earth prices drop >25% from peak.