Want to work in AI? Here are the skills to master, economist says

NYU Stern economist Robert Seamans argues AI will permeate most occupations and create demand for new roles such as 'AI explainers/translators,' 'AI auditors' to check bias and accuracy, and corporate instructors to train employees on AI tools. For investors, this implies incremental corporate spending on AI governance, training and testing services and potential growth opportunities for firms offering enterprise AI tools, auditing, and workforce-upskilling solutions as companies integrate generative AI into operations.

Analysis

Market structure: winners are AI infrastructure and cloud-platform leaders (NVDA, MSFT, GOOGL, AMZN) plus observability/MLOps vendors (DDOG, SNOW, PLTR) and training/edtech providers (COUR, UDMY) who can monetize ‘AI explainer’ and auditor services. Losers include low-skill staffing, legacy on‑prem software (ORCL, some ERP incumbents) and chip incumbents with weak GPU roadmaps (INTC); expect 6–18 month share shifts as customers prefer cloud-native, model‑enabled stacks and pay 5–15% premium for integrated AI features.

Risk assessment: tail risks include fast-moving regulation (EU/US transparency or liability rules within 6–24 months), major model safety incident triggering lawsuits, or GPU supply shocks causing 20–50% cost spikes. Immediate (days) effects are sentiment-driven; short-term (months) is adoption ramp/pilot-to-prod conversion; long-term (2–5 years) is structural capex into data centers and persistent margin divergence. Hidden dependencies: power/electricity constraints, data‑label supply and skilled “translator” labor; these create second‑order inflation in wages for mid‑skill roles. Key catalysts: large enterprise AI procurement cycles closing in next 3–12 months and major chip cadence announcements.

Trade implications: establish 2–3% long NVDA and 2% long MSFT/GOOGL each (3–9 month horizon) to capture infrastructure and cloud monetization; implement NVDA 3‑month call spread ATM→+15% to cap premium. Pair trade: long SNOW (6–12 months, 2%) vs short ORCL (2%) to express cloud-first vs legacy divergence. Hedge: buy 3–6 month INTC puts (0.5–1% portfolio risk) as downside insurance.

Contrarian angles: consensus underprices the commercial opportunity for AI governance/auditing — small-cap MLOps/obs vendors or staffing/education platforms could re-rate if they land large enterprise contracts; consensus may be overbought on NVDA’s near-term multiple (risk of mean reversion if guidance slips). Historical parallel: internet-era middleware winners consolidated quickly — expect 1–3 takeover targets among MLOps players. Unintended consequence: heavy reliance on a few foundation‑model providers increases counterparty risk and regulation exposure.

Key Decisions for Investors

• Establish a 2–3% portfolio long in NVDA (infrastructure play); implement a 3‑month call spread from ATM to +15% to capture upside while limiting premium outlay ahead of earnings and chip cadence announcements.
• Allocate 2% longs in MSFT and GOOGL (1% each) as cloud + AI stack exposure over 6–12 months; buy 6–9 month 0.5% portfolio protective puts if entering >2% position size to hedge macro risk.
• Run a pair trade: long SNOW (2% position, 6–12 months) vs short ORCL (2%) to express cloud-native data platform adoption; trim if SNOW/ORCL relative outperformance exceeds 25%.
• Purchase INTC 3–6 month puts sized to 0.5–1% portfolio risk to hedge semiconductor/CPU downside from GPU displacement or supply‑chain shocks.
• Initiate 1–2% exposure to COURSERA (COUR) or UDMY (1% each) for 6–12 months as a contrarian play on monetization of corporate AI training; monitor contract wins and enterprise adoption weekly and scale up to 3% if revenue retention >90% over two quarters.