AI budgets aren't new money. They're taken from somewhere
Gartner surveyed 199 customer service and support leaders in April and May 2026. Spending on AI in that function rose 38%. The overall function budget rose 2%. Kim Hedlin, a director analyst at Gartner, put the mechanism plainly: leaders are "redirecting spending away from labour and overhead and instead toward technology," and the challenge is "ensuring those investments produce measurable business value."
What it actually means
Read those two numbers together and this stops being an adoption story and becomes a transfer. The AI is being paid for by taking money out of people and overhead, and it is happening before anyone has proved the return; Gartner's own analyst says so in the same breath.
Three days ago the point here was that a discounted token price is a customer-acquisition budget wearing a price tag. This is the other end of that thread. The per-token price is not what decides your AI bill, and it certainly isn't what decides whether the programme survives. Where the money came from decides that.
So if you're proposing AI spend inside an organisation, name the line it comes out of before someone else names it for you. An initiative quietly funded by not replacing three people has a very short runway the moment somebody asks what it bought. And if you're selling into one of these functions, understand what you're actually up against: not a competing vendor, but headcount that has already been cut to pay for you, and a buyer who now has to justify that trade.
38%rise in customer-service AI spend, against 2% budget growth
Source: Gartner · Thursday 27 August 2026