The Signal · Column 055

AI budgets get approved faster than they get staffed

Andrei van Noordt · 16 September 2026 · 2 min read

Mercer has put a name to the thing every AI budget meeting is quietly circling. Its argument is that the hard part is no longer approving the spend, it is turning that AI investment into impact, and that what sits in the gap is people rather than technology. I see the same pattern in the briefs that cross my desk. The money is signed off. The platform is chosen. The impact is still being described in the future tense.

The reading most people miss is that investment and impact run on completely different clocks. Buying a capability takes a quarter and a signature. Building one takes hiring, rewriting how work flows, retraining the people doing it, and absorbing the months where output dips before it climbs. Most companies budget carefully for the first and improvise the second, then conclude the tools underdelivered. The tools rarely underdeliver. What underdelivers is an organisation that purchased a capability and then staffed it like a pilot project.

Nobody has ever bought their way to an AI capability; they have only ever hired their way there.

This is where the market reprices, and it is not repricing what people assume. The scarce profile is no longer the person who can build the model, because that skill has broadened quickly. It is the person who can carry a model into a live process and make the business trust what comes out of it: the engineer who owns deployment and monitoring, the data lead who makes the inputs defensible, the manager willing to redesign a workflow around a system instead of bolting a system onto the workflow. Those are the people who convert spend into outcome, and there are far fewer of them than there are licences already being paid for.

For hiring leaders, the practical read is simple. Before you approve the next expansion of AI spend, name the specific people who will be accountable for the impact, and check whether they already work for you. If they do not, that hire is the real investment and everything else is equipment. The companies seeing returns this year began hiring for adoption at the same moment they began paying for the tools, not two budget cycles later when the board asked what the money bought.