Citi says AI inference capacity constraints are pushing enterprises to ration spending

Citi says rising pricing complexity should add value at the emerging orchestration layer, with Apple WWDC and next week’s Databricks Summit potential sources of updates on on-device inference and enterprise adoption. Enterprises are responding by rationing AI spend, with Citi citing Walmart’s cap on its internal AI tool, while pricing segmentation across frontier models is accelerating. Citi says the AI trade remains shaped by strong enterprise demand colliding with limited inference capacity, with the pressure showing up through higher costs.