AI Job Growth: New Data Shows AI Creates Jobs, Not Cuts Them
AI Job Growth: New Data Shows AI Creates Jobs, Not Cuts Them
Every few weeks another headline warns that AI is about to hollow out the workforce. Entry-level roles are the ones supposedly most at risk. The reasoning goes that if a model can draft the email, write the code, or answer the support ticket, why hire the junior person who used to do it? A new dataset on AI job growth suggests the opposite is happening.
A new study from Ramp’s Economics Lab, done in partnership with workforce data firm Revelio Labs, puts some real numbers behind the question instead of just vibes. And the numbers say something different from the doom narrative: companies that adopt AI seriously are hiring more, not less, and entry-level hiring is growing fastest of all.
What the study actually measured
Ramp has a unique vantage point here. Its corporate card and bill pay platform sees exactly what companies are spending on AI vendors: OpenAI, Anthropic, GPU cloud providers, coding agents, and the rest of the AI tooling stack. Researchers linked that spending data to Revelio Labs’ workforce records for 21,559 U.S. firms, tracking monthly headcount from January 2021 through February 2026.
A company counted as an “AI adopter” once it hit at least $100 a month in AI vendor spend for three consecutive months, enough to rule out someone expensing a ChatGPT subscription once and calling it a pilot. From there, adopters were split into high-intensity and low-intensity groups based on how much they spent per employee in the first three months after adoption.
The headline result: firms that adopted AI grew total headcount by roughly 10% over the following two years. But that number hides an important split. Almost all of the gain came from high-intensity adopters, the companies that went all in. Low-intensity adopters, the ones dabbling, saw no statistically significant change in headcount either way.
The entry-level number driving this AI job growth
The part of this study that cuts against conventional wisdom hardest is what happened to junior roles. At high-intensity adopting firms, entry-level headcount grew 12%, faster than headcount overall. If AI were quietly automating away the bottom rung of the career ladder, this is exactly the number you’d expect to see shrink. Instead it grew, and the gains showed up broadly: engineering, sales, admin, and customer service all saw increases, not just the technical roles you’d assume AI tools would supercharge first.
The theory the researchers offer is a familiar one to anyone who has actually deployed automation inside a real organization: AI removes friction and grunt work, which lets the company take on more work and grow, rather than doing the same amount of work with fewer people. Growth, not headcount reduction, ends up being the dominant effect, at least for firms that commit to using the tools properly rather than bolting them on at the edges.
The caveats matter, and the researchers are upfront about them
Credit to the authors here: they don’t oversell the result. A few things are worth flagging before anyone treats this as the final word on AI and employment.
First, this isn’t a representative sample of the U.S. economy. It’s Ramp customers who can be matched to Revelio’s workforce data, which skews toward tech-forward, venture-backed, knowledge-work-heavy companies. The AI adopters in this study were already larger and growing faster than non-adopters before they ever touched an AI tool, which is exactly why the researchers compared adopters to other, not-yet-adopted firms rather than to companies that never adopt AI at all. It’s a more careful design than a naive before-and-after comparison, but the underlying selection effect (the kind of company that goes all-in on AI is a different kind of company to begin with) doesn’t fully disappear.
Second, two years is not a long time horizon. The authors are explicit that they can’t rule out reallocation within firms showing up later, job mix shifting even while total headcount holds up. They plan to keep tracking these firms as more data comes in, which is worth watching.
So the honest summary isn’t “AI adoption guarantees growth.” It’s closer to: among the kind of company that commits hard to AI, the fear of an immediate hiring collapse hasn’t materialised. If anything, the opposite has happened so far.
Why this matters beyond the headline
The distinction between high-intensity and low-intensity adopters is arguably the most useful part of this research for anyone actually running an organisation. Buying a few AI seats and hoping for a step change doesn’t move the needle: the study found literally no measurable effect for that group. The gains belong to companies that adopted deeply enough to change how work actually gets done.
That’s consistent with what we see working with regulated organisations at Askelie. The teams that get real value from automation aren’t the ones layering a chatbot onto an existing process and calling it done. They’re the ones that let AI take over the repetitive, rules-based, document-heavy work (the intake forms, the compliance checks, the data reconciliation) and then redeploy the people who used to do that work toward the judgment calls, the relationships, and the exceptions that actually need a human. That’s not a story about fewer people. It’s a story about the same people doing more valuable work, and organisations that grow because they can finally take on more of it.
The Ramp/Revelio data is early, and it’s one dataset from one part of the economy. But it’s a useful counterweight to the assumption that AI adoption and job growth are naturally at odds. For now, at least among the companies willing to commit to it, they aren’t.