For much of the past decade, cloud-first was the default starting point for new enterprise workloads. It let organizations modernize quickly and access capabilities that would have been difficult or expensive to build in-house, and for a large share of workloads, public cloud remains the right answer.

Today, the conversations I’m having with business leaders are becoming more nuanced – shifting away from how quickly workloads can move to the cloud and towards a more fundamental question: where should each workload run to deliver the greatest business value? Do we need the robustness of mainframes, the flexibility of cloud, or the control offered by on-prem?

Infrastructure decisions are no longer driven solely by technology teams. They increasingly influence operational resilience, regulatory compliance, financial performance, and board-level risk, making it a broader C-suite conversation.

The consequence is a shift in enterprise infrastructure strategy where public cloud is still pivotal but not the exclusive destination. A workload-first approach ensures every platform is evaluated and each part of the puzzle is justifiable through a clear return on investment (ROI).

Infrastructure strategy is now a business strategy

Arguably, several factors are driving this shift, but digital sovereignty is becoming one of the more significant. For organizations operating across multiple jurisdictions or in regulated industries, sovereignty extends well beyond where data is physically stored.

Boards increasingly want confidence over who can access critical information, which legal frameworks apply, and whether essential systems can continue operating if regulatory conditions change – not to mention guarding against geopolitical risk. As governments introduce new legislation around data governance and operational resilience, these considerations are becoming central to infrastructure planning rather than compliance exercises carried out after the fact.

The EU Data Act is one clear example of this shift, introducing some of Europe’s most technically specific cloud sovereignty requirements. It obliges providers to support functional interoperability through standardized interfaces and transparent data structures, and sets out switching obligations that cap exit fees and define migration timelines. Crucially, it requires providers to demonstrate that foreign jurisdictions cannot compel access to EU-stored data outside EU legal processes.

Organizations are responding by taking a more deliberate approach to workload placement. Some are adopting sovereign cloud environments, others are investing further in private infrastructure, while many continue to rely on on-premises platforms for workloads requiring greater control. The goal? Operational, regulatory, and commercial optimization.

Why workload placement deserves more attention

Cost is often presented as the primary driver behind workload repatriation, but that can oversimplify what’s really happening. A true workload-first strategy looks beyond simple cost reduction to total ROI. Gartner forecasts public cloud services will grow by 21.3 percent in 2026, driven by continued migration, modernization, and AI adoption.

Performance, latency, resilience, security, sustainability and regulatory obligations all influence where workloads deliver the greatest value. AI is adding another layer of complexity, particularly where organizations are developing proprietary models using commercially sensitive data. Questions about where data is processed, how models are trained, and which infrastructure supports inference have become strategic business considerations rather than purely technical ones.

Treating workload placement as a one-off migration exercise no longer reflects the reality facing most enterprises. Infrastructure strategies need to evolve alongside changing business priorities, regulations and technologies.

That is also why hybrid infrastructure is better understood as an operating model than as a stepping stone towards becoming fully cloud-based. Large enterprises increasingly operate across hyperscalers, private cloud, virtualized infrastructure and mainframes, with each environment supporting different workloads and business requirements. The priority now is building the visibility and governance needed to manage these estates effectively and make informed decisions about where workloads should run.

Visibility is becoming the foundation of better decisions

As enterprise technology estates become more distributed, making informed infrastructure decisions depends on understanding where technology investment is creating value. That need for visibility extends well beyond cost alone. IBM’s latest Calculus of AI Sovereignty study found that 91 percent of organizations don’t fully understand their dependencies across AI vendors, models, and infrastructure, highlighting how difficult it has become to make informed decisions across increasingly complex hybrid estates.

That lack of visibility also makes it harder to demonstrate return on investment. In our latest Technology Investment Management Report, nine in ten technology leaders said uncertainty around ROI is influencing investment decisions, increasing pressure to justify where and how technology spend is allocated.

FinOps has helped organizations bring greater discipline to public cloud spending, but cloud costs are only one part of today’s technology estate. Organizations increasingly need a consistent way to evaluate public cloud, private cloud, SaaS, on-premises infrastructure, and mainframe against business outcomes rather than in isolation.

This is where Technology Business Management (TBM) plays an important role. By providing a consistent framework for understanding the cost and value of technology investments, TBM enables organizations to compare deployment options more effectively and make workload placement decisions based on measurable business priorities rather than assumptions.

The next phase of cloud strategy

Cloud isn’t becoming less important, and neither are data centers. The organizations making the greatest progress recognize that infrastructure strategy is no longer about choosing one deployment model over another but striking the best balance.

Cloud-first accelerated enterprise transformation. The next phase will be defined by organizations that combine hybrid infrastructure – embracing everything from the public cloud to the mainframe – with greater visibility into cost, performance, resilience and business outcomes. Those that treat workload placement as an ongoing business decision, rather than a one-off migration exercise, are those who will be better positioned to respond to changing regulatory, commercial and ongoing AI demands.