AI and ML
But doing so doesn’t necessarily meet business needs
Conventional wisdom holds that artificial intelligence leads to job losses. However, a new survey of more than 21,000 U.S. firms suggests the opposite: companies that invest in AI actually increase headcount, though the effect is delayed.
The study, conducted by Ramp—an AI‑focused finance company—and Revelio Labs, an HR analytics firm, found that firms making substantial AI commitments add jobs at a faster rate than those with low‑intensity AI adoption. The employment boost, however, does not appear until roughly six to twelve months after the investment.
While one might assume the lag reflects the time needed to address AI‑related errors, the Ramp report attributes it to the time required for best practices to propagate through an organization.
“Firms that adopt AI grow headcount by 10.2% over the two years following adoption, but these gains are driven entirely by high‑intensity adopters,” the report states. “Low‑intensity adopters see no statistically significant change.”
High‑intensity adoption is defined here as average AI spending of about $33.67 per employee per month during the first three months after implementation, rising over time. By contrast, low‑intensity adopters spend roughly $2.78 per employee over the same period.
This spend level is modest compared with the roughly $86,000 in severance and restructuring charges Oracle recorded for each of the 21,000 employees it laid off last year as part of its AI capital‑expenditure adjustments.
In a social‑media post, Ara Kharazian, Ramp’s lead economist, cautioned that skepticism is warranted because companies that adopt AI tend to be faster‑growing already. He added that the analysis controls for this by comparing early adopters against firms that have not yet adopted AI, whose growth trajectories are assumed to be more comparable.
“Entry‑level headcount grows even faster, at 12% over two years,” Kharazian noted. “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—people who know how to use AI effectively.”
Entry‑level workers, especially recent graduates and college students, are a natural source for these roles.
Nevertheless, other data paint a less optimistic picture for new graduates. The unemployment rate for recent college graduates in March 2026 stood at 5.6%, compared with 4.3% for all workers, according to the Federal Reserve Bank of New York.
Overall U.S. unemployment remained flat since May at 4.3%, the Bureau of Labor Statistics reported, with nonfarm payrolls adding 57,000 jobs in June and the unemployment rate at 4.2%.
Despite the positive employment signal, some businesses are expressing doubts about AI investments, citing concerns over cost and control. In a recent CNBC interview, Palantir CEO Alex Karp noted that both military and private‑sector enterprises share skepticism about how frontier AI providers such as OpenAI and Anthropic conduct business.
“Technical customers want control over their compute, models, data stack, and investment alpha,” Karp said. “They want to own the means of production.”
He argued that the AI industry must rebuild trust by answering fundamental questions about data ownership, storage, and prompt security. While acknowledging that this stance serves Palantir’s own interests—promoting mobile, application‑layer, and compute solutions—Karp highlighted an unresolved issue for frontier model providers.
Government agencies and enterprises cannot afford to depend on a volatile service provider, especially if AI models become unavailable due to regulatory restrictions, refuse to answer queries, or become prohibitively expensive.
Investing in AI creates a job for model providers—making AI available, controllable, affordable, and worthwhile—a role that still needs to be filled.