Pulse Report: AI Cost of Compute

Senior operators are deploying AI cost controls faster than they can see, track, or forecast the spend those controls are meant to govern.

AI cost management is running on instinct, not instruments. Only 1 of 10 operators can see AI spend as a clean line item. Six of 10 do not know which way that spend has moved in six months. Half do not track costs against forecast at all. The gap between action and visibility is the clearest signal in this data. Six of 10 have set team quotas, yet 9 of 10 still cannot read their spend cleanly. Leaders are capping a number they cannot fully see. Confidence is thin. Only 1 of 10 feels highly confident forecasting spend a year out, while 4 of 10 rate confidence low. Unit economics stay hidden. Three of 10 cannot name their biggest cost driver, and just 1 of 10 knows cost per AI task precisely. This uncertainty is already shaping decisions. Half of operators say consumption pricing changed a project choice. Two killed a project outright. One respondent, a manager in professional services, pointed to the harder problem of separating true AI value from process gains it merely highlights. The takeaway for executives is direct. Before adding more controls, invest in visibility. A clean line item, a simple forecast-to-actual check, and one unit-cost metric would move most teams from guessing to knowing. The sample is small at 10 responses, so treat these as early signals, not settled fact.