Tim Pawlenty: What AI thinks we should do about AI

6-10%: Conservative estimates cited in the piece indicate unemployment in AI-exposed occupations could reach 6–10%, with high vulnerability in programming, customer service, record-keeping, market analysis, sales, investment advice, IT security, law, accounting, engineering and media/entertainment. The author warns policymakers have largely ignored the risk, raising the prospect of mass layoffs that could sharply reduce consumer spending and threaten financial stability, and calls for urgent preparation and policy action.

Analysis

AI-driven labor displacement creates concentrated winners in compute, orchestration and security while producing diffuse losers across labor intermediaries and consumer-facing microbusinesses. Expect cloud compute demand to rise ~20–40% in the first 12–24 months of enterprise model rollouts as firms shift from in-house tooling to hosted inference and fine-tuning; that magnifies capex-to-opex migration and raises gross margins for hyperscalers while compressing legacy on-prem vendors.

Second-order stress will show up in consumer credit and local services: a 1–2% permanent shift of white‑collar labor from payroll to software could reduce regional deposit growth and commercial real estate demand over 2–5 years, concentrating risk in small banks and office REITs rather than large diversified financials. Cybersecurity budgets and compliance services should reaccelerate — attack surface increases and new regulation enforcement create recurring revenue opportunities that can re-rate high‑growth software names over a 12–36 month window.

Regulatory risk is the primary catalyst that can flip outcomes quickly; aggressive data/privacy rules, auditability requirements, or mandated human‑in‑the‑loop policies could slow adoption materially inside 6–18 months and favor incumbents with deep compliance moats. Conversely, lighter-than-expected regulation plus enterprise ROI proofs could concentrate profits at a handful of cloud/AI platform providers, creating a convex payoff for long-duration exposure while making labor‑intensive service models structurally uninvestable.

Key Decisions for Investors

• Long GOOG/GOOGL LEAPs (12–24 month): Allocate 2–3% notional to long-dated call options on GOOGL (buy Jan-2028 LEAPs) to capture convex upside from hosted AI compute demand. Risk: premium decay and regulatory shock; set max position loss = premium paid, trim if single-day negative regulation headline causes >15% drawdown in the underlying.
• Long cybersecurity (CRWD or PANW) equity (6–18 month): Size 2–4% to CRWD or PANW to play sustained security spend; target 30–50% upside if budgets reallocate to tooling, with a 20% stop-loss. Hedge: reduce exposure by buying put protection on 30% of the position if broader tech sell-off exceeds 10%.
• Short staffing / labor-intermediaries (RHI or MAN) (6–12 month): Small, tactical short (1–2% notional) in staffing firms to express margin and demand compression risk as automation substitutes routine tasks. Risk/reward: asymmetric – limited upside if staffing rebounds; use a hard stop at 25% adverse move and pair with longs in automation software to lower net directional risk.
• Pair trade — long INTU (automation/payroll software) vs short RHI (staffing) (12–24 month): Go 1.5x long INTU equity / 1x short RHI to express structural reallocation from manual services to embedded automation in SMBs. Expected outcome: 40–60% relative outperformance for the pair if adoption accelerates; cap downside by reducing net exposure if macro consumer stress indicators (credit card delinquencies or small business payrolls) worsen by >30% year-over-year.