The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs
View original at venturebeat.comVentureBeat AI - Enterprise Ai Title: The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs Date: 2026-07-16 19:16 Source: https://venturebeat.com/ai/the-ai-compute-gap-enterprises-are-buying-infrastructure-faster-than-they-can-measure-what-it-costs <p>Across 107 enterpris…
What we drew from this source
The claims Via News extracted from this document. We point to the source; we don't replace it.
Overall satisfaction with current AI infrastructure averages 4.0 on a five-point scale, with ease of implementation at 3.8 and value for money at 3.9.
60% confidenceThe single largest planned AI infrastructure evaluation area over the next 12 months is AI-specialized clouds, at 45%, a category almost none of these enterprises use today.
60% confidenceIn VentureBeat's prior April-May 2026 survey wave, the most-cited planned infrastructure strategy change was moving workloads to specialized AI clouds, at 33%; usage of CoreWeave (3%), Lambda (4%) and Crusoe (2%) was equally marginal at that time.
60% confidenceFewer than half of enterprises (44%) rigorously track the cost and return of their AI compute; 39% track only partially, 20% cannot quantify it yet, and 6% have not prioritized it.
60% confidenceThe providers drawing the most switching consideration are Microsoft Azure and Google Cloud (33% each), OpenAI (30%), and Gemini (22%), suggesting near-term movement is mostly incumbents trading share rather than defections to new entrants.
60% confidenceEnterprises choose AI infrastructure providers primarily on integration with the existing stack (41%) and total cost of ownership (35%); cost per million tokens is the deciding factor for just 8%.
60% confidence64% of enterprises plan to switch or add an infrastructure provider within twelve months, and 38% within the next quarter alone.
60% confidence83% of enterprises that operate GPUs report utilization of 50% or less; 49% run at 25% or below.
60% confidenceOnly about one in five enterprises (21%) run AI in production at scale; 76% are still experimenting or running only some workloads in production.
60% confidenceOnly 12% of enterprises clear the 50% GPU utilization mark, and a further 8% do not measure utilization at all.
60% confidenceRoughly one in five enterprises (18%) either do not recognize the shift from GPU compute to memory bandwidth as a constraint or have not begun to address it.
60% confidenceSpecialized AI clouds carry the highest net expansion momentum among infrastructure approaches (+24), narrowly ahead of hyperscalers (+22).
60% confidence
Data points we hold from this source
| OpenAI · switching consideration share | 30 percent |
| Dell Technologies · market share | 31 percent |