Companies that add more AI also add more people

AI leads to job losses, or so the conventional wisdom goes. But a new survey of over 21,000 US firms implies the exact opposite: When companies invest in AI, they add positions, but not immediately.

According to Ramp, an AI finance biz, and Revelio Labs, an HR biz, companies making a significant financial commitment to AI add jobs at a higher rate than low-intensity adopters. But job gains don't appear until six to 12 months later.

One might be tempted to interpret this as the amount of time it takes to assess the resources required to clean up after AI mistakes, but the Ramp study argues that the lag reflects the time required for best practices to filter through organizations.

"Firms that adopt AI grow headcount 10.2 percent over the two years following adoption, but these gains are entirely driven by high-intensity adopters," Ramp's report on the subject claims. "Low-intensity adopters see no statistically significant change."

High-intensity adopter here means average per-employee AI spending of about $33.67 per month in the first three months of adoption (and rising over time), compared to low-intensity adopters spending just $2.78 per employee.

That's far less than the roughly $86,000 in severance and restructuring charges Oracle incurred for each of the 21,000 employees laid off last year as a wage-shedding counterbalance to its AI capex costs.

In a social media post, Ara Kharazian, lead economist at Ramp, cautioned that some skepticism is warranted because companies adopting AI are already faster growing. But he insists that the analysis accounts for that by comparing early adopters against firms that haven't adopted yet, where the growth trajectory is assumed to be more similar.

"Entry-level headcount grows even faster, 12 percent over two years," said Kharazian. "This is our first evidence that high-AI-adopting firms are hiring different kinds of employees. 

"We believe they are selecting for a new set of skills, specifically, people who know how to use AI and use it well. Entry-level workers, especially recent graduates and college students, are a natural place to look."

That may be the case at the companies surveyed, but other sources suggest that the trend hasn't really improved the lot of those entering the job market. The unemployment rate for recent college graduates in March 2026 was 5.6 percent, compared to 4.3 percent for all workers, according to the Federal Reserve Bank of New York.

According to the US Bureau of Labor Statistics, the US unemployment rate remained essentially flat since May, when it was 4.3 percent. "Both total nonfarm payroll employment (+57,000) and the unemployment rate (4.2 percent) changed little in June," the Labor Department said. 

While Ramp's data may suggest some upside to investing in AI, some businesses appear to be having second thoughts, based on concerns about cost and control. In a recent CNBC interview, Palantir CEO Alex Karp argued that military and private sector enterprises share similar skepticism about the way frontier model companies like OpenAI and Anthropic do business.

Technical customers, Karp said, want "control over their compute, their models, their data stack, and their (investment) alpha. They want to know they own the means of production."

Karp argues that the AI industry needs to rebuild trust, which will require answers to basic questions like who owns the data, where it is stored, and whether prompts are secure.

Karp acknowledges that's a self-interested argument because Palantir is pushing a combination of mobile, application layer, and compute. But he's also correct in identifying an unresolved problem with frontier model providers. 

Government organizations and enterprises can't afford to be beholden to a capricious service provider, particularly if its AI models may not be available due to government restrictions, if its AI model may refuse to respond to what's asked of it, or if the price becomes excessive.

When companies invest in AI, they add a job for model providers – make AI available, controllable, affordable, and worthwhile. That work still needs to be done. ®