30% of Americans Worry That AI Will Make Their Jobs Obsolete

70% of Americans say AI will decrease job opportunities generally and 30% fear their own job could become obsolete (up from 21% a year earlier). Significant public concerns include opposition to local AI data centers (65%), belief the government isn’t doing enough on AI regulation (74%), 55% saying AI will do more harm than good in daily life, and 51% opposing military use of AI for target selection. Poll of ~1,400 US adults conducted March 19–23; margin of error ±3.3 percentage points.

Analysis

Local resistance to new AI infrastructure is a direct tax on marginal supply: slower permitting and higher community opposition will push out multi-year data center build schedules, raising effective construction and financing risk for owners of greenfield sites. That creates a transient scarcity premium in colocated capacity and power-constrained regions, which benefits hyperscalers with locked-in campuses and incumbents who can flex existing capacity while penalizing developers reliant on continual new builds.

Public discomfort with military use and opaque governance increases the probability of procurement delays and tougher contracting terms for boutique AI defense contractors; prime contractors with large, diversified backlog and established compliance teams are better positioned to absorb pauses. Expect a multi-quarter visibility hit on niche systems integrators and a re-rating pressure on early-stage defense-AI vendors that lack institutional contracting experience.

Negative sentiment among the electorate also raises the odds of near-term regulatory initiatives (local moratoria, state-level siting restrictions, disclosure mandates) that can compress multiple sectors unevenly: software vendors face demand uncertainty, cloud infra owners face capex timing risk, and chipmakers see order phasing that can amplify share price volatility. Paradoxically, constrained greenfield capacity and shifting timelines can sustain pricing power for GPUs and other AI compute components for several quarters, supporting a hardware outperformance thesis versus pure SaaS plays.

Key reversals would be clear federal/state permitting guidance, large-scale corporate retraining commitments, or a marquee public–private build that normalizes data-center siting; conversely, ballot measures, export-control escalation, or a high-profile adversarial AI incident are credible tail risks that would materially widen liability and compliance costs across the stack within 6–24 months.

Key Decisions for Investors

• Pair trade (3–9 months): Long NVDA (buy 6–9 month calls) / Short DLR or EQIX (buy 3–6 month puts). Rationale: hardware scarcity and continued AI demand vs near-term permitting/capex deferral in data center REITs. Target 2.5:1 upside/downside; stop-loss 25% on option premium.
• Trade (6–12 months): Long CRWD or PANW (40–60% position) — calldated exposure or buy-and-hold — to capture upgraded spending on governance, monitoring, and safety tools as regulation/oversight tightens. Expect 20–40% upside if enforcement/standards pick up; downside capped to single-digit if market reprices growth.
• Event/defense trade (9–18 months): Long LMT (buy 12-month calls or stock) for resilience amid procurement volatility; avoid small-cap defense AI primes. Upside from re-accelerated DoD programs post-policy clarity; downside limited by diversified backlog — target asymmetry ~1.5–2x over 12 months.
• Tactical hedge (0–6 months): Buy short-dated puts on mid-cap SaaS/AI names with rich multiples prone to sentiment swings (select names via desk list). Small allocation (2–4% portfolio) to protect against rapid derisking if regulatory headlines or ballot initiatives surface.