Companies Serious About AI Are Hiring. The Rest Are Doing Layoffs.

Two clients sent us job orders in the same week last month. One is scaling a data platform team from six people to nineteen and still not done. The other has not opened a single req since February, months after telling us in January they were “going all in on AI.”

Same size company, same market, both describing an AI strategy. Only one of them was actually running one. We see this split constantly, and it rarely tracks with company size, funding, or even industry the way people assume. It tracks with something narrower: whether a company treated AI as a real capital commitment or as a subscription it could cancel anytime.

The Data Behind the Pattern

A dataset released in June put a number on what we had already been seeing. Ramp and Revelio Labs linked AI spending records to workforce data across nearly 22,000 U.S. companies, splitting them by how much a company actually invested rather than whether it used AI at all.

• The spending gap is real. Token adopters spent roughly $3 per employee per month. The heaviest spenders averaged roughly $34, more than ten times as much.
• The headcount gap follows it exactly. High intensity adopters grew headcount 10.2 percent over two years. Low intensity adopters saw no measurable change at all.
• Entry level roles moved the same direction. Junior headcount at the heaviest spenders grew 12 percent, directly against the assumption that AI eats junior jobs first.

What a Real AI Investment Looks Like From Our Side

We can usually tell which group a client belongs to before we see any data.

Signs of a real investment:

• The req is specific. They ask for engineers who can wire a model into an existing production system, not people who have simply used a chatbot.
• Growth shows up everywhere, not just engineering. Sales, support, and finance open at the same time, because growth in one part of the business becomes growth everywhere.
• New roles get invented. Box’s CEO has talked publicly about thirteen new job categories at his company, including one called a model evaluator, a role that exists purely because AI output now needs a human judging which answer is actually better for a specific use case.

Signs of a company still dabbling:

• The req is vague. Nobody can say exactly what the person will own.
• The budget has not cleared a trial period. Approval is pending on results that were never clearly defined.
• Someone says “we’re still figuring out our AI strategy” on the first call. That is a company telling you honestly that it has not committed to one yet.

The Layoffs Are the Other Half of the Same Story

The companies making headlines for AI linked cuts this year are rarely the heavy spenders. Multiple 2026 estimates put AI attributed layoffs in the ninety to a hundred thousand range so far.

Almost none of the companies behind that number are increasing their AI spend. They are the ones that treated a subscription as a substitute for a real decision, then cut headcount when the substitution predictably did not hold up.

What This Means If You Are Hiring or Job Hunting

If you’re hiring:

• Ask what’s actually expanding. Is the plan growing because AI let the business take on more work, or shrinking because a tool is covering a gap nobody funded properly?

Check the posting data. Tech postings are up 27 percent year over year, and 75 percent now list AI skills, up from 73 percent in May. That is where the real demand is concentrating.

If you’re job hunting:

• Target the high intensity group. That is where the entry level growth actually is.
• Expect a different interview. Production judgment over algorithm recall, and less patience for AI buzzwords with nothing behind them.

We built our placement process around this exact distinction. Every engineer we deploy is vetted for the production judgment that shows up at real adopters, not the AI buzzwords that show up on a resume. If you want help figuring out which side of this split your own hiring plan actually sits on, let’s talk.