Who's selloff was it, anyway?

Last week’s tech rout (briefly) erased more than $1 trillion in market value. The headlines fixated on the obvious trigger: Hyperscalers announced combined capital expenditures approaching $650 billion for 2026. We’ve all heard of FOMO; now we are getting a first-hand look at FOLS—fear of losing shirt. Markets rebounded sharply by Friday with the Dow crossing 50,000 for the first time, but the bounce shouldn’t obscure what the selloff revealed. Beneath the capex panic was the potential for accelerated de-skilling of the knowledge economy.

The deskilling fear was on clear display in the so-called “SaaSpocalypse.” When Anthropic released legal research, financial analysis, and data marking plugins for its Claude Cowork platform, investors saw a demonstration of how AI might begin exercising domain-specific expertise in fields previously considered safely human territory. Thomson Reuters and LegalZoom each dropped more than 15 percent. RELX and FactSet posted double-digit declines. The WisdomTree Cloud Computing Fund, an ETF tracking emerging cloud software companies, has fallen 20 percent over the past six weeks, including roughly seven percent last week alone.

A second, structural threat compounded the SaaS panic: the rapid maturation of platforms like OpenClaw that enable radically decentralized, AI-powered software development. OpenClaw and its offspring, Moltbook, enable individuals and small teams to plan and execute complex software projects with minimal human intervention. The aggressive democratization of tools for building sophisticated software suggests the competitive moats protecting legacy software giants may be drying up. If so, AI is calling into question not just the jobs within these firms but the viability of the firms themselves.

Bank of America’s Vivek Arya pointed out that the capex and SaaS fears were mutually exclusive. If demand for AI is so strong it threatens the existence of software firms, then the Magnificent Seven’s capex plans are justified. That these two fears operated in tandem suggests a stampede rather than well-considered second- and third-order analysis. Panics are rarely logical, but the underlying anxiety about the possible loss of high-paying jobs can be instructive.

For decades, the professional, scientific, and technical services sector has powered America’s high-wage job growth. According to the Bureau of Labor Statistics (BLS), this sector accounts for nearly 15 percent of total employment, and BLS projects it to grow at more than double the rate of overall employment through 2034. Legal services, management consulting, financial analysis, and software development have helped millions of workers join not just the middle class but become wealthy. What if these projections aren’t just wrong but wildly off?

MIT researchers have found that current AI technology—not the systems an increasingly recursive AI development cycle are speeding into production—can already automate tasks accounting for nearly 12 percent of the US labor market, representing up to $1.2 trillion in wages. The most exposed jobs in the MIT study are concentrated in white-collar professions.

In a January 2026 essay, Anthropic CEO Dario Amodei argued that unlike previous automation waves that tended to hit individual sectors and industries at irregular intervals, AI’s “cognitive breadth” means it will simultaneously pressure finance, consulting, law, and tech. While capital and firm-level adaptive capacity constrain adoption and will mitigate the speed of change the underlying message remains the same: it isn’t just companies or jobs that will be under stress, but white-collar skills themselves. The greater resiliency of these workers may not be enough if entire skill categories are suddenly swept away.

Labor market data offer some faint early signals of the slowdown in white-collar job creation. Employment in professional and business services has experienced low hiring activity over the past two years. The unemployment rate among college graduates in AI-exposed fields has edged higher.

This is the selloff beneath the selloff. Capex exposure is one thing, but dramatic restructuring of the ways work gets done, which people get paid for doing it, and how restructuring will be absorbed by the economy and the workforce is another entirely.

As I’ve noted before, we have some recent experience with sudden, rapid deskilling. The automation and trade shocks of the past several decades, while decidedly a good thing for the economy and most Americans, hit skills, some workers, and their communities hard. We never planned for those effects or envisioned their consequences, which have been significant. Nothing is learned from the second kick of a mule, especially a bigger, more powerful mule. America needs a plan.

---

Brent Orrell is a senior fellow at the American Enterprise Institute, where he studies workforce development and AI’s impact on employment.

Originally published at AEI.