No, AI isn't inevitable. We should stop it while we can. | Opinion

David Krueger, an AI professor and founder of a nonprofit focused on AI risk, urges an international halt to advanced AI development, arguing the pursuit of 'superintelligence' poses existential risks and has already driven large-scale job cuts. He calls for concrete measures including ceasing production of advanced AI chips (naming TSMC and ASML as critical nodes), banning or restricting data-center construction, and negotiating international agreements in 2026 to prevent unchecked scaling of AI.

Analysis

Market structure: A credible political push to ban advanced AI chips or data centers would directly hurt foundry/equipment leaders (TSM, ASML) by cutting near-term wafer starts and deflating multi-year capex; buyers of cloud/AI services (NVDA, AMZN, MSFT) could see slower revenue growth but retain pricing power on scarce GPU supply. Competitive dynamics favor vertically diversified or fabless players if fabs hit regulatory headwinds; concentrated suppliers currently enjoy pricing power but also single-point regulatory risk that could compress industry multiples by 20–40% if realized. Cross-asset: equity volatility in semiconductors would spike (VIX-like jump concentrated in SOX), bid bonds (10y yields down 10–40bps), strengthen safe havens (USD, JPY), and lower industrial metals/power demand trajectories by mid-single-digit percent over 12–24 months.

Risk assessment: Tail risks include coordinated export bans or an international moratorium on advanced-node production (low prob, high impact) that could cut TSM/ASML revenue >25% in 12 months and trigger supply shocks. Immediate (days): sentiment-driven selloffs; short-term (weeks–months): legislative proposals, mayoral moratoria and CPI-like headlines; long-term (years): structural capex rerouting or geographic reshoring. Hidden dependencies: downstream demand (hyperscalers) can lobby against bans; enforcement depends on chokepoints (ASML EUV shipments, TSMC fabs). Key catalysts: US Commerce export rules, Senate bills, ASML shipment notices; watch 30–90 day windows after such announcements.

Trade implications: Tactical hedges and relative-value trades are preferable to naked directional shorts. Use 3–6 month put spreads on TSM/ASML to cap cost, rotate EM/FX exposure away from TWD/EUR toward USD/JPY, and lengthen duration by 1–3yrs on conviction of near-term growth slowdown. Sector rotation: underweight pure-play foundries/equipment, overweight defensive utilities/REITs and select software/AI services providers with GPU exposure.

Contrarian angles: Consensus treats an AI ban as binary; political friction, enforcement difficulty and hyperscaler lobbying make a total moratorium unlikely — demand may simply shift geographies or accelerate on-shore investment, creating a multi-year capex trough then rebound. Market may over-penalize high-quality names: >15% drops in TSM/ASML without legislative follow-through would be buying opportunities. Historical parallel: 2018–19 trade tensions caused temporary semiconductor derating then stronger recovery once supply contracts re-priced. Unintended consequence: accelerated domestic subsidy programs that ultimately benefit non-Taiwan/NL suppliers and reshape market share in 24–48 months.

Key Decisions for Investors

• Establish a 2–3% notional 3–6 month put spread hedge on TSM: buy 6-month 10% OTM puts and sell 6-month 20% OTM puts to protect vs regulatory/export-control shock while limiting premium outlay; reassess after 60 days post any legislative filing.
• Place a 2–3% duration hedge via long TLT or 10y futures for 3–12 months to capture a potential 10–40bp drop in yields if AI capex is curtailed and growth risk rises.
• Implement a dollar-neutral pair: long NVDA (1–2%) vs short TSM (1–2%) over 6–18 months — NVDA benefits from durable GPU demand and fabless flexibility while TSM is levered to advanced-node capex and regulatory chokepoints.
• Avoid initiating outright long positions in ASML or TSM until a 60-day post-announcement window passes after any federal moratorium/export-control bill; if either stock falls >15% without binding legislation in that window, scale into 1–2% buys.