Findings from Omdia's survey of 1201 infrastructure decision makers. Individual results may vary.
have regrets about their initial AI infrastructure decisions
run AI workloads in a hybrid mix of cloud, on-premises, and at the edge
have already moved some AI workloads back on-premises
What's driving AI workload hosting strategies
Cloud costs add up quickly
Cloud spending jumps an average of 30% after AI deployment, pushing over half of organizations to scale back their initiatives.
Workload fit drives the shift
54% of organizations cite workload optimization as the top reason for adopting a hybrid strategy.
Security shapes the architecture
Controlling access to AI systems and data is a significant factor in infrastructure decisions for 82% of organizations.
Sovereignty draws the line
For most organizations, data sovereignty isn't a side consideration—76% agree it directly influences where their AI can run.
Build for what's ahead: AI's next era starts with infrastructure
Server bottlenecks are closing in
More than half of organizations (52%) expect their server infrastructure to become an AI bottleneck within 12 months, reinforcing the need for robust, scalable data center capacity.
The network is running out of runway
67% of organizations expect network bottlenecks within a year, a reminder that server and network capacity must scale together, or the bottleneck simply moves.
Edge is going mainstream
Edge inferencing is surging—97% of organizations are deploying or planning to deploy within 12 months (a 2.4x increase from today).
Validated designs are accelerating
90% of organizations are using or plan to use integrated AI solutions in the next 12 months to simplify complexity and speed up on-premises deployments.
Real voices from infrastructure leaders across industries
