Ramp: Heavy AI Adopters Grew Headcount 10% Over Two Years

The most useful thing in the current AI-and-jobs argument is that nobody can agree what the data says, and Fortune lays that out plainly. On one side is a recent study by financial services firm Ramp of more than 21,000 US firms that found the heaviest AI spenders, what Ramp calls its "high-intensity" adopters, expanded overall staff by 10% and boosted entry-level hiring by 12% over two years, while the bottom two-thirds of adopters saw no headcount growth at all. On the other side is the latest US jobs report, which showed employers unexpectedly cut 23,000 positions in July, alongside Microsoft laying off nearly 5,000 people in early July while pouring billions into AI data centers, Amazon cutting about 30,000 jobs between the end of 2025 and the start of this year, and Oracle telling regulators in a June filing that its layoffs of thousands during the past year were tied to AI.

Underneath the split is a measurement problem more than a story problem. UCLA economics professor Till Von Wachter told Fortune that "it's been notoriously hard to pin that down," meaning the extent to which any observed layoffs are really driven by AI rather than post-pandemic corrections. Researchers even have a name for the noise, "AI washing," where companies attribute layoffs to AI to seem forward-thinking or, conversely, hide their AI usage for fear of public outcry. A June California Policy Lab study found no statewide spike in unemployment insurance claims among AI-exposed roles like software developers and customer service reps since ChatGPT's release in late 2022, which cuts against the loudest displacement narratives. But a 2025 report co-authored by Stanford economist Erik Brynjolfsson, using ADP workforce data, found workers aged 22 to 25 in AI-exposed roles such as software engineering suffered a 16% relative employment drop compared to less-exposed peers.

The honest caveat is that these studies are measuring different things with different proxies. Ramp's numbers cannot tell you whether its heavy AI spenders were already the faster-growing firms that would have hired anyway, and the Stanford result is a slice of one age band in a handful of exposed occupations, not the whole labor market. Even the loudest CEOs have softened their claims: Sam Altman said in May that OpenAI had been wrong about how much "people would continue to be at the center of everything," and Anthropic's Dario Amodei wrote in June that his earlier comments about AI eliminating jobs were not meant to make him a "prophet of doom" but a call for policymakers to plan and adapt.

For anyone running a company or a policy desk, the useful move is probably to stop asking whether AI is destroying jobs in the aggregate and start asking which cohorts and which firms it is helping or hurting, because that is the level at which the current data can actually answer.