AI Pulse · ANZ AI Watch

A cheap AI pilot can create a false sense of security. You might run a useful proof of concept with a small group, see a manageable monthly bill, and assume the cost will stay broadly in line as you expand it. Datacom CEO Greg Davidson is warning New Zealand enterprises not to make that assumption. Consumption-based AI pricing can compound quickly as more staff use the tool, more work flows through it, and usage becomes part of day-to-day operations.

The practical issue is not whether AI delivers value. It is whether you know what that value will cost once the pilot becomes a company-wide service. A pilot rate is often a starting point, not a permanent commercial arrangement. If AI spend has not been budgeted as an ongoing operating cost, finance teams can be left explaining a material new bill after the business has already become dependent on the service.

Davidson’s advice is sensible: avoid treating one vendor’s pilot pricing as a long-term commitment. Build in contractual exit rights and portability, and consider a multi-vendor approach where it makes sense for your business. Before scaling, ask what drives consumption, who owns the budget, what costs look like at wider adoption, and how easily you can move models, data, or workflows if pricing changes. This is the kind of thing we build into an AI command centre, giving leaders a clearer view of AI usage, cost, and operational dependency.