VentureBeat Pulse Research surveyed 107 organizations with 100+ employees in Q2 2026, finding a significant compute gap: enterprises are investing aggressively in AI infrastructure while lacking visibility into its economics. Only 21% run AI in production at scale, yet 45% plan to evaluate AI-specialized clouds—a layer almost none use today—and 64% intend to switch or add infrastructure providers within 12 months.
The compute already deployed runs underutilized: 83% report GPU utilization of 50% or less, and fewer than half (44%) can rigorously track what their AI compute actually costs. When choosing infrastructure, enterprises prioritize integration with existing stacks (41%) and total cost of ownership (35%), not headline price—cost per million tokens is the deciding factor for just 8%.
The coming shift from GPU compute to memory bandwidth as inference scales is barely on enterprises' radar, with roughly one in five either unaware of it or yet to address it. The survey suggests that despite high churn intent and rapid investment in new infrastructure, visibility and measurement capabilities remain a foundational challenge for enterprise AI deployments.