For all the anxiety around AI-driven job losses, the labor market is not yet showing evidence of broad displacement. The more interesting signal may be where employment is moving, rather than whether it is disappearing.

The ecommerce era offers a useful precedent. Since 2000, retail employment has remained broadly flat, while transportation and warehousing employment has risen by roughly 50%. Technology disrupted the composition of labor demand without eliminating the need for labor itself.

There are signs of stress at the entry level. Unemployment among young workers and recent graduates has moved higher, raising questions about whether AI is beginning to compress demand for junior knowledge workers. But the trend started before the GenAI boom, meaning monetary tightening and weaker hiring remain important competing explanations.

The stronger evidence is emerging on the productivity side. Workers are already spending meaningful portions of their week using AI—and the industries using it most are also reporting the largest time savings.

Information services leads the pack, with AI used for roughly 14% of weekly work hours and associated time savings of approximately 2.6%. Professional services and finance show a similar relationship.

The occupational data reinforces the point. Computer and math, management, and business and finance roles are among the clearest early beneficiaries.

For PE, this changes the underwriting question. The near-term opportunity may not be replacing employees—it may be generating more output from the same headcount. In Solow terms, AI raises A: total factor productivity, allowing growth without proportional increases in labor or capital.

Bottom line: If AI can lift portfolio-company productivity before materially disrupting employment, the first macroeconomic dividend could arrive through margin expansion and faster EBITDA growth—not mass layoffs.