China and America Must Get Serious About AI Risk

In November 2024 the US and China issued a landmark joint statement stressing the need for human control over any decision to use nuclear weapons, following earlier Geneva talks on AI risks. The article warns that accelerating AI adoption raises systemic threats — from cyberattacks, AI-enabled lethal weaponry and disinformation to novel bioweapons and potential AI-driven market crashes — while noting China’s new rare-earth export controls as evidence of intensifying strategic competition. The author urges sustained, senior-level diplomacy and bespoke risk-reduction frameworks (distinct from traditional arms control) to address verification, dual-use and private-sector dynamics, implications hedge funds should fold into geopolitical and supply-chain risk assessments.

Analysis

Market-structure: US–China AI risk diplomacy plus Chinese rare-earth export tightening accelerates bifurcation: winners include US cloud/AI infrastructure (NVDA, MSFT, GOOGL) and cybersecurity firms (PANW, FTNT) which gain pricing power as demand for secure, verifiable AI stacks rises; losers include China-dependent hardware suppliers and broad Chinese internet/AI exporters (BABA, BIDU, KWEB) facing both policy and FX pressure. Supply-side: rare-earth and chip-equipment constraints imply 10–30% higher input-cost volatility for downstream chipmakers over 6–18 months, raising capex and onshoring incentives that favor ASML and Western fabs. Cross-asset: expect modest flight-to-quality into USTs (yields -5–15bps on headline shocks), USD strength vs CNY (2–6% moves on policy shocks), commodity spikes in rare earths and gold (5–25% on escalation), and higher realized equity vol for semis and Chinese tech over 1–3 months.

Risk assessment: tail risks include cascading cyberattacks or market flash crashes from AI-driven trading, broad sanctions on Chinese materials, or rapid military escalation; probability low but impact systemic (value-at-risk shock >20% equity drawdown). Time horizons: immediate (days) — headline-driven vols; short (weeks–months) — export-control implementation and supply re-routing; long (quarters–years) — structural decoupling and onshoring capex. Hidden dependencies: many AI safety and verification vendors are small-cap and leveraged to government contracts; second-order effect — export controls accelerate Chinese domestic investment, creating longer-term competition. Catalysts: bilateral summit outcomes, concrete export-control regulations, and major cyber incidents will accelerate repricing.

Key Decisions for Investors

• Establish a 2–3% portfolio long in NVIDIA (NVDA) via a 6-month call spread (buy 15% OTM, sell 40% OTM) to capture continued AI-infrastructure dominance while capping premium; reassess after NVDA moves +30% or on chip-capacity announcements.
• Add 1.5–2% exposure to rare-earth plays: 60/40 split MP Materials (MP) / Lynas (LYC.AX) via shares; target +30% in 6–12 months on further export tightening, stop-loss at -15% per position.
• Allocate 1.5–2% to cybersecurity names (long PANW shares or 12-month 25-delta calls) expecting a 20–40% revenue uplift thesis for security services if AI-related hacks rise; trim on +40% moves.
• Establish a 2% tactical short via KWEB (or CQQQ) 3-month put spread if (a) China announces additional export controls or (b) CNY weakens >2% against USD within 30 days; unwind if CNY stabilizes or policies soften.
• Hedge tail risk with 1% allocation to GLD (or 3–6 month ATM S&P put protection sized to 1% equity risk) to protect against geopolitical escalation that could cause a >10% equity drawdown within 30–90 days.