Brian Gruttadauria, CTO Hybrid Cloud, HPE.gettyFor a long time, virtualization was treated as a solved problem. Platforms were stable, operating models were familiar and service providers built around that consistency. Over the past few years, due to changes in the industry, that stability has shifted.This shift is not a routine refresh. Costs have changed, and workloads have evolved. Assumptions that business leaders held for the past decade no longer apply. The question is not whether to move, but whether to rethink the platform or carry forward the same constraints into the next cycle.The Pressure Is BuildingFor most providers, the pressure is first felt through cost. Licensing changes are creating real strain, particularly for MSPs operating within tight margins. At the same time, there is growing discomfort with dependency as many environments were built over years on a single stack. That approach made sense when the ecosystem was predictable. It’s harder to justify now.​Workloads are also changing. AI, data pipelines and high-performance applications do not align cleanly with traditional virtualization models. Running a virtual machine is no longer the challenge. Positioning it close to the data it depends on, and doing so across environments, is.​If the conversation stays focused on cost or vendor replacement, it misses the broader shift: The role of infrastructure itself is changing.Service providers are not all starting from the same place. Those operating more like cloud service providers are already managing workloads across public and on-prem environments, which makes this transition more immediate and more complex. Others, particularly those focused on private cloud, may face fewer integration challenges today but still need to plan for hybrid extension over time.Infrastructure’s Role Is ChangingVirtualization was originally about efficiency. Dividing physical resources to run more workloads on the same infrastructure. Today, it is increasingly about coordination.Providers are moving toward a platform model where compute, networking, data and security operate as a unified system. Observability is embedded rather than added later; security is integrated from the start. Operations cannot rely on manual processes when workloads are distributed and continuously changing.In more advanced environments, infrastructure behaves less like a set of independent systems and more like a connected set of services. These systems exchange information, trigger actions and surface issues automatically. That shift changes what virtualization needs to support. With the industry shifting, the role of virtualization needs to change.Waiting Makes It HarderThere is a natural instinct to wait, let the ecosystem mature, watch how others handle the edge and avoid disruption. That logic does not hold here.This transition spans multiple years and affects architecture, operations and skills. Some applications will not migrate smoothly, and some teams will find it difficult to change considering the deep experience they have tied to existing platforms. These challenges do not diminish with time; they accumulate.Providers that start early have room to work through complexity. They can test workloads, identify gaps and build internal capabilities before the transition becomes urgent. This allows them to absorb disruption in stages. Providers that wait could be setting themselves up to face the same work under tighter timelines and external pressure, where risk shows up in compressed execution, higher costs and fewer options.​The advantage for early movers is control. They are not constrained by decisions made years ago.​This becomes visible in day-to-day operations. Automation replaces manual steps, deployment becomes repeatable and troubleshooting shifts toward proactive identification. It also shows up in service delivery, where faster provisioning, more consistent performance and integrated security become expected rather than differentiated.​Over time, these gains compound. Early movers build internal tooling, refine processes and gain operational familiarity while others are still beginning the transition.​This shift is often framed in terms of cost or vendor strategy. In practice, it is just as closely tied to AI readiness.​AI workloads introduce different requirements. They depend on data locality, high-performance compute and the ability to move data across environments. They also introduce new service models, including GPU-backed infrastructure and hybrid deployments that span edge and core. These models do not align cleanly with legacy virtualization environments. They require platforms designed to support them from the outset. That is what makes the timing relevant. This is not only about replacing existing systems. It is about preparing for new workload demands.What To Evaluate Before MovingDecision-making here needs to extend beyond near-term cost. Total cost of ownership over a two- to three-year period still matters, but it should be evaluated alongside architecture and operating model. The key questions are practical. Can the architecture support hybrid environments and multi-tenancy? Are security and observability integrated into the platform? Can operations shift toward automation? Does the platform support AI workloads without significant redesign?It is also important to evaluate the platform layer itself. A platform-based approach can provide greater flexibility as AI standards, security requirements and workload demands evolve. It also helps create more consistency around governance, observability and operations across hybrid environments, reducing the need for repeated architectural changes as infrastructure requirements shift.​There is also an execution layer. Teams need time to build new skills. Supporting capabilities such as backup, disaster recovery and migration must be in place. These factors determine how quickly a strategy can move from planning to execution.Most infrastructure decisions are incremental. This one is not. Service providers are being pushed to revisit assumptions that have been in place for years. That creates friction, but it also creates an opportunity to reset. ​Handled well, this is a chance to reduce dependency, simplify operations and build for the next generation of workloads. Handled poorly, it becomes a delayed migration that carries existing limitations forward.The difference comes down to timing. Providers that act now will spend the next several years refining their model. Those that wait will spend that same time catching up.​Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?