Enterprises shift to hybrid cloud amid costs, AI and sovereignty

Enterprises are reducing blanket cloud-first strategies and moving workloads across public clouds, private data centers and regional providers. Rising fees from hyperscale providers, unpredictable AI-related consumption charges and data sovereignty concerns are cited by IT leaders as reasons for the change. Gartner projects that 90% of organizations will adopt hybrid cloud infrastructure by 2027.

Companies that built systems on Amazon Web Services, Microsoft Azure and Google Cloud are reassessing where to run specific applications. Primary compute and basic storage fees remain visible, but firms report that secondary charges such as data egress, inter-region transfers, long-term storage tiers and AI inference or token usage now add materially to monthly bills. Large-scale model training and high-volume inference have produced instances of unexpected, rapid cost increases.

Jay Litkey, SVP of cloud and FinOps at Flexera, warned that “the surprises are usually not the obvious things, such as compute or storage. It is often the layers underneath that catch people out,” and pointed to data transfers between regions, wrong storage tiers and duplicate environments as common causes of excess spend.

Quick provisioning in public clouds has created governance gaps. Organizations that lifted and shifted legacy systems without redesign are seeing resource sprawl, unclear ownership and fragmented visibility across engineering, finance and operations teams. Justin Sharrocks, general manager for EMEA at Trusted Tech, pointed to weak workload design and low FinOps maturity as frequent drivers of overspending and added that “cloud pricing hasn’t necessarily increased; it’s become easier to mismanage.”

Some IT teams are embedding cost controls into architecture decisions and moving FinOps earlier in the development lifecycle. James Peet, practice director for cloud and digital transformation at Ensono, said teams that make cost a design consideration change how systems are built and how budgets are managed.

Regulatory and geopolitical factors are also shifting placement decisions. Financial services, healthcare, government and critical infrastructure organizations are seeking clearer guarantees about data residency, access and applicable legal frameworks. Mark Duff, VP international regions at Mitel, reported that security, compliance and digital sovereignty have become board-level priorities and that those concerns are influencing infrastructure strategy.

As a result, hybrid and multi-cloud deployments are becoming the norm. Workloads that need elastic scale and global reach continue to run in public clouds, while sensitive data, latency-sensitive applications and steady-state processing migrate to on-premises or regional solutions. Edge computing is growing in sectors that require local processing of continuous sensor data and real-time decision-making. Krystal Mattich, VP of infrastructure and trust at Brain Corp, explained that physical AI workloads impose requirements for low latency, local processing and safety that make edge or on-premises placement preferable for some functions.

Industry professionals emphasize that most organizations are not abandoning hyperscalers but are selecting where each workload runs based on cost, latency, compliance and resilience. James Brooks, lead for hybrid solutions at Hewlett Packard Enterprise, highlighted secondary charges such as data egress and over-provisioning as hidden causes of rising bills and identified AI workloads as a new focus for cost control.

This article is part one of a three-part series on post-cloud strategies. Part two will examine how sovereignty, security and geopolitics are driving regional and sovereign cloud adoption, and part three will review approaches to hybrid architecture design.