A colleague of mine interrupted me during my coffee break. He asked about an AI design tool his team was using. It's $3,000 a year, and he wanted to renew. Should he try to get a better deal? My answer was: go ahead and try, but keep in mind it sits under the threshold where any vendor would likely offer a discount. He nodded and moved on. I kept thinking about it.
The thing is: nobody had to approve the AI subscription his team was running. It came out of a company credit card, under a threshold, filed as software. From what I understand, the tool does something we would otherwise have to hire someone for. Hiring the person it's quietly doing the work of would have required a headcount request, three approvals, and a business case defending why the role should exist at all. But for the AI subscription, he just pulled out his company credit card and bought it without anyone looking. I certainly hadn't noticed - we're already in our second year with this particular tool.
That asymmetry isn't an accident. It's the default shape of how every company's budget works. Capital and tools tend to move through frictionless lanes. People move through a lane built to slow decisions down, because an additional person on the payroll is a recurring, contestable, politically visible cost. A tool subscription and its token usage are just a rounding error - until maybe they aren't anymore.
So the AI tool doesn't have to be better than the person to win. It just has to sit in the lane with less friction. Multiply that by every team quietly running the same math, and your company's staffing decisions are being made by which expense category has fewer approval gates, not by anyone assessing what the work actually needs.
Two takeaways follow from this: