AI Workloads Expected to Drive 78% Surge in Edge Computing Demand Over Next Year

A recent Deloitte survey of 120 market operators reveals a significant shift in computing infrastructure as organizations accelerate their artificial intelligence capabilities. Edge computing is projected to see a 78% increase in workloads over the next 12 months, significantly outpacing traditional on-premises data center growth by a ratio of six to one. This trend underscores the growing necessity for low-latency processing and distributed architectures to support real-time AI inferencing and the deployment of agentic AI.

Deloitte’s research, led by principal Chris Thomas and senior manager Diana Kearns-Manolatos, highlights that AI-driven workloads—including model pretraining, reinforcement learning, and large-scale inferencing—are increasing demand across all tech environments by at least 20%. However, edge computing platforms and emerging AI cloud providers are poised for the most dramatic growth. Specifically, 78% of surveyed market operators expect edge workloads to spike, a rate that outpaces on-premises data center growth by approximately six to one. This shift is driven by the need for high-performance computing (HPC) that can handle the complex, distributed nature of modern AI systems while maintaining the low-latency responses required for real-time applications.

The survey, conducted between March and April 2025, included 120 market operators such as data center providers, energy providers, and distributors to gauge how infrastructure strategies are evolving. While emerging AI cloud providers lead with an 87% expected increase, the rise of edge computing reflects a strategic move toward 'AI factories' capable of handling standard inferencing and HPC at scale. Conversely, nearly a third of respondents plan to decrease their reliance on mainframes and traditional on-premises workloads. Organizations are instead reconfiguring existing facilities, reactivating decommissioned sites, and partnering with hyperscalers or niche providers to optimize GPU and AI token utilization.

For the edge computing sector, these shifting demands necessitate a focus on building resilient and efficient environments that prioritize data integrity, real-time fault detection, and robust runtime security. As AI systems scale, infrastructure leaders must address the challenges of synchronous and asynchronous coordination across orchestration layers while managing an expanded attack surface. The move toward edge is largely motivated by the requirement for low-latency responses and the need to process high volumes of quality data closer to the source. This evolution requires new engineering-led approaches to control, tooling, and management to navigate the currently uncharted territory of AI consumption and utilization patterns.