After years of migrating operations to public clouds, a growing number of enterprises are now bringing selected workloads back in-house, reversing what had seemed like an unstoppable shift, according to a new report from InformationWeek. The trend isn't driven by cloud failure but by a push toward optimization, improved performance needs, and more strategic control, the report finds. CIOs and IT leaders are repatriating certain workloads from public cloud infrastructure back to on-premises or private cloud environments.
The cloud never delivered the cost savings it promised, according to Scott Miserendino, CTO at security services firm DataBee. Any efficiencies gained were captured as profit by the hyperscalers, he explained. Reliable cost estimation for cloud workloads has always been challenging, but with more workloads processing larger data sets spread across various cloud environments, the gap between cost estimates and actual costs is widening more dramatically than ever. This is particularly true for egress costs, which are the fees charged for moving data between workloads hosted by different cloud providers or within private cloud environments. For the past decade, "cloud-first" was the automatic approach, said Kelly Raskovich, an associate director at Deloitte Consulting, but that's shifting not because organizations are abandoning the cloud, but because they're becoming more intentional about where workloads operate.
Cost predictability, or its absence, is a major factor in transition decisions, Raskovich noted. Many enterprises have faced sticker-shock from variable cloud operating expenses, especially data egress charges and the ongoing cost of running steady-state workloads in a dynamic cloud environment, she said. For applications with consistent demand, an on-premises infrastructure can deliver greater cost visibility and control, the report states. However, moving workloads back on-premises brings obstacles that must be weighed against long-term benefits, including recent memory and RAM shortages that have driven up hardware costs significantly, plus one-time migration expenses such as building or expanding infrastructure, potential downtime during migration, data transfer costs, application remediation, and IT training. Ran Aroussi, technical co-founder and CTO at VarOps, advised teams to move the workload without redesigning it: "Lift first, measure second and optimize third," he said, warning that teams combining migration with an architecture rewrite can never determine which change caused which outcome.
The shift back to on-premises fundamentally changes the skills teams need, Raskovich warned. Over recent years, engineers have been trained in cloud architecture, FinOps, and cloud-native services, but repatriation demands a revival of traditional infrastructure engineering, physical networking, and hardware lifecycle management. This can create anxiety among team members who may view stepping away from the cloud as a career setback. Hybrid architecture is the future, and engineers who understand both cloud-native and on-premises environments are now among the most valuable talent in the jobs market, she said. Self-hosting generates substantial new work, and infrastructure management teams have grown accustomed to using the easy button, Miserendino noted. Self-hosting also requires more thought about business continuity and technical failure resiliency, and for some critical workloads, resiliency requirements necessitate multiple, geographically dispersed data centers that can be difficult to manage and keep synchronized.
Cloud repatriation doesn't signal the cloud's demise but reflects a more mature market and a more disciplined approach to infrastructure thinking, Raskovich stressed. What's emerging now is the normalization of a strategic hybrid model in which workloads are positioned in the environment that makes the most sense financially, technically, and operationally. Raskovich believes the future lies in a three-tier model: the cloud for elasticity, on-premises for consistency, and the edge for immediacy. CIOs are no longer treating the cloud as an automatic destination but rather as an operating model, she said, finally positioning workloads where the technology best supports the use case. Cloud computing was never the universal silver bullet, though it was often promoted as one, Aroussi said, and the workload is now finding the place it should have been at all along—for a lot of AI work, that place was never the public cloud. The question that should be asked now isn't on-premises versus cloud but what data a system needs access to, Aroussi said, and answering that honestly usually makes the hosting decision self-evident. The repatriation movement suggests that enterprises are maturing past blanket cloud adoption and learning to match infrastructure choices to actual business requirements rather than following industry hype. Leaders who can navigate this hybrid reality will likely find themselves better positioned to balance innovation speed with cost discipline than those still wedded to a single infrastructure philosophy.