Alexander Campbell
· Campbell Ramble
· February 25, 2026 at 16:22
· ⏱ 19 min read
| Read on Substack ↗
Summary
The article argues that AI-driven automation will decimate white-collar jobs (9.5 million currently peaking) but massively expand demand for physical infrastructure, energy, and skilled trades. The author models three scenarios showing 6–16% unemployment and severe fiscal pressure, concluding investors should go long the physical layer (energy, construction, minerals) and short the paper layer (IT outsourcing, professional services). No explicit security-level trade recommendations are disclosed.
•Nine white-collar occupations at all-time highs (customer service, software developers, market research analysts, management consultants, etc.) total ~9.5 million jobs in AI's targeting reticle.
•BLS projects 1.8 million new jobs by 2034 in sectors like electricians (+9%), solar installers (+42%), wind turbine techs (+60%), nurse practitioners (+40%), and cybersecurity (+29%).
•Historical pattern: 270 occupations since 1950 Census, only two went to zero (elevator operators and video rental); telephone operators fell 96% but new job categories emerged.
•Klarna replaced 700 customer service agents with AI, cutting resolution time 80%; GitHub Copilot writes 46% of code at adopting firms.
•Skilled trades gap is 650,000 unfilled openings per year; nuclear workforce must roughly double; shipbuilding needs 30,000 additional workers by 2030.
•Three modeled scenarios (mild/base/severe) assume 25%/40%/55% white-collar exposure with 5%/10%/15% annual automation rates, producing 6–16% unemployment and GDP drag of $0.5–$2.8 trillion.
•UBI funding would require dramatic income tax increases (effective rates approaching 50% in severe case) unless alternative taxes on stagnant capital, conspicuous consumption, or corporate profits are implemented.
•Investment frame: long the physical layer (energy, materials, infrastructure, defense industrials) underpinned by $2T+ in IIJA/IRA/CHIPS spending; short the paper layer (IT outsourcing, seat-based SaaS, professional services).