Heavy AI Spenders Add Workers, Not Cuts

Ramp's Ara Kharazian and Ryan Stevens, with Revelio Labs chief economist Lisa Simon, published "A New Look at AI's Impact on Jobs" on June 30, linking Ramp line-item AI vendor spending to Revelio workforce records for 21,559 U.S. firms observed monthly from January 2021 through February 2026. Using a Callaway-Sant'Anna staggered-adoption design, the authors find total headcount 10.2% higher for high-AI-intensity adopters over the first 24 months after adoption, with no detectable change for low-intensity adopters; entry-level headcount rose 12.0% among high-intensity adopters. Gains were broad across engineering, sales, administration, customer service, and finance, and were concentrated in the Information sector.

The authors are explicit about the limits of their own design: AI adopters in the sample were already larger, more technical, and faster-growing before adopting, so the paper compares high- and low-intensity adopters against later adopters in the same eventual-intensity group rather than claiming a clean causal estimate. Notably, the paper's own literature review cites Brynjolfsson et al. (2025), which used ADP payroll records and occupational exposure scores to find a roughly 16% employment decline for workers aged 22-25 in the highest AI-exposure occupations after ChatGPT's release - a different methodology (occupation-level exposure vs. firm-level spending) reaching a more negative conclusion for a narrower slice of the workforce. The two studies are not directly contradictory since they measure different things, but practitioners should not read this Ramp/Revelio result as a full rebuttal of displacement concerns for entry-level, highly-exposed roles specifically.

Public reporting ties the pattern to concrete hiring moves: PYMNTS reports Google is hiring hundreds of forward-deployed engineers to move customer AI projects from pilot to production, that Box leadership has described roughly 13 new AI-driven job categories emerging internally, and that IBM plans to increase U.S. entry-level hiring in 2026 as it adapts early-career roles. For hiring and workforce planning, useful signals to track include AI spend per employee over time, the share of new hires in implementation/evaluation roles versus traditional functions, and the roughly six-to-12-month lag between sustained AI spending and measurable headcount change that the paper documents.

Follow-up research applying stronger causal identification, industry- and role-level breakdowns beyond the Information-sector concentration this paper found, and whether the entry-level hiring gains persist as AI tools mature and require less integration and evaluation labor per deployment.