Why IT Teams Are Reclaiming Control From the Public Cloud

For more than a decade, public cloud has given many enterprises a faster way to build, scale, and experiment. But as cloud use has matured, the trade-offs have become harder to ignore, including less predictable costs, more complex environments, and more services for teams to manage.

For OpenMetal CTO Jamie Tischart, that reassessment is less about abandoning public cloud and more about deciding where dedicated, private, and open-source infrastructure models can give teams better control.

That does not mean public cloud is going away. Synergy Research Group reported that enterprise spending on cloud infrastructure services reached $129 billion in the first quarter, up more than $35 billion from the same quarter a year earlier.

According to Tischart, IT leaders are becoming more focused on cost volatility and operational complexity.

“The cost structure and inconsistency in billing have become difficult for teams to manage. As companies have become more focused on profitability, variability in infrastructure costs is no longer sustainable for many organizations,” Tischart said. 

Where public cloud still fits and where it falls short

Public cloud remains valuable when teams need speed, elasticity, or access to specialized managed services. The challenge comes when every workload is treated the same way.

Steady, always-on systems such as databases, internal platforms, production applications, and analytics environments may not need the same elasticity that public cloud provides. Over time, those workloads can become expensive under consumption-based pricing models, especially when usage is predictable, but billing remains variable.

For predictable, always-on workloads, dedicated or private cloud infrastructure can make budgeting easier because teams are not tying core production systems to highly variable consumption-based billing.

As cloud environments grow, many organizations are putting more structure around how they manage them. Flexera’s 2026 State of the Cloud Report found that Cloud Center of Excellence adoption rose from 69% to 71% year over year, while the share of organizations with a FinOps team reached 63%.

Those numbers show how much cloud management has changed. Cloud has become important enough, expensive enough, and complex enough that many organizations now need dedicated teams to manage it.

Tischart said the economics can become especially difficult when cloud spending begins to exceed roughly 15% of revenue, before even accounting for the management and development time required to optimize architecture around cost controls.

“For large-scale cloud environments, teams can spend many valuable hours each week just trying to optimize potential costs. That time could be spent in more strategic areas of the company,” he said. 

Ongoing optimization work is one reason more teams are reassessing where certain workloads should run. Public cloud can support them, but for predictable, always-on systems, it may not always be the most cost-effective option.

VMware changes and what comes next

Cost is only one reason teams are rethinking infrastructure. VMware changes have also pushed many organizations to revisit decisions they may have delayed.

Following Broadcom’s acquisition of VMware, VMware eliminated perpetual licenses and moved entirely to a subscription-based licensing model, according to Acronis. Acronis also reported that VMware’s minimum licensing requirement increased to 72 CPU cores per purchase as of 2025, compared with a previous baseline of 16 cores.

For teams with large VMware environments, those changes can shift the economics quickly. A platform that once felt stable may now require a broader review of licensing, workload placement, infrastructure design, and vendor exposure.

For Tischart, the VMware moment is also forcing a larger conversation about vendor lock-in. When infrastructure strategy depends heavily on one proprietary platform, pricing or licensing changes can quickly become a business constraint rather than a purely technical concern.

Tischart said organizations should avoid treating this only as a short-term migration question.

“The factors everyone needs to consider are return on investment and switching costs. I encourage people to look at this over multiple years, because it can become a major savings and productivity benefit,” he stated. 

That multi-year view matters. Moving away from an established platform requires planning, retraining, testing, and operational confidence. But if a change reduces licensing exposure, improves infrastructure economics, and frees staff for higher-value work, the business case can become stronger over time.

Evaluating alternatives to traditional virtualization

As organizations look beyond traditional virtualization platforms, OpenStack and Proxmox are getting more attention from teams that want more control, less lock-in, and a path toward open-source infrastructure models. OpenMetal’s comparison of Proxmox and OpenStack explains how both options can support organizations rethinking their virtualization strategy.

But this is not only a technical decision. Teams have to weigh the short-term effort of learning new tools against the long-term value of flexibility, cost control, and independence from proprietary licensing models.

“The transition can be unsettling in the short term as teams learn new technologies and stacks. On the other hand, it can be liberating for team growth, removal of vendor lock-in, and better infrastructure unit economics,” Tischart highlighted.

Leaders may assume they have to choose between simplicity and flexibility, but Tischart does not see it that way. In the right environment, an OpenStack-powered private cloud or Proxmox-based environment can simplify operations over time by reducing tool sprawl and giving teams more freedom to adapt.

Depending on the organization’s needs, that broader model could include hosted private cloud, integrated bare metal, dedicated storage clusters, GPU servers, managed services, or a combination of those options. OpenMetal is one provider in this space, with infrastructure options that reflect the broader shift toward more predictable costs, performance control, and operational flexibility.

Rethinking infrastructure as a connected system

Leaders are also starting to look at infrastructure less as a set of separate tools and more as a connected system.

Compute, storage, networking, virtualization, orchestration, security, and support are often discussed separately. In practice, they affect one another. A decision in one area can change performance, cost, reliability, and scalability somewhere else.

Tischart said leaders should think about infrastructure as a system that can be tuned to workload needs while still being flexible enough to change as requirements evolve.

For teams that do not want to manage that complexity alone, providers such as OpenMetal can help bridge the gap between infrastructure control and day-to-day execution through managed private cloud and operational support.

“The gaps come when systems are designed without infrastructure considerations or knowledge of the infrastructure limitations,” Tischart said.

A more connected strategy can help organizations combine private cloud, dedicated servers, storage, GPUs, and managed services into a broader infrastructure model that is easier to adapt over time.

How evolving workloads are changing infrastructure needs

AI, analytics, data-heavy applications, and performance-sensitive systems are adding more pressure to these decisions. These workloads often need predictable performance, strong resource control, and clearer cost visibility.

BCC Research reported that major technology companies are collectively investing $650 billion annually in AI infrastructure, with cloud spending driven by AI workloads expected to surpass $500 billion by 2026.

For IT leaders, AI is not only a software issue — it is an infrastructure issue, too. Teams need to know whether their environments can support GPU demand, data movement, storage performance, security requirements, and predictable cost management.

For organizations building private AI environments, the infrastructure question often extends beyond raw compute. GPU availability, storage performance, data location, network design, and operational support all affect whether AI workloads can scale without creating unpredictable costs or compliance concerns.

Tischart noted CIOs and CTOs should evaluate not only current infrastructure spend, but how that spend may change over the next 24, 36, and 48 months.

“Most importantly, we should all be evaluating the demands of AI adoption on our infrastructure strategy and the impact on overall cost,” he said.

The role of security and compliance

Security and compliance are becoming part of the same conversation. Organizations are paying closer attention to where data lives, how workloads are isolated, who can access systems, and how activity is audited.

Providers such as OpenMetal give teams more control over how they configure and manage private and dedicated infrastructure. This is important for businesses with strict compliance needs, data residency concerns, or confidential computing workloads.

What IT leaders should be evaluating now

IT leaders do not need to start with a major migration. A better first step is to reassess where each workload belongs: which applications need public cloud elasticity, which could run more cost-effectively on private or dedicated infrastructure, and which require stronger control over performance, compliance, or data location.

Leaders should also look beyond today’s bill. Tischart said one common mistake is miscalculating switching costs, or not calculating them at all, because teams often focus on the short-term effort without measuring the longer-term return.

The public cloud era is not ending, but the default-cloud era may be. More organizations are moving toward a deliberate mix of infrastructure models, where public cloud, private cloud, bare metal, open-source virtualization, and managed infrastructure each play a role.

For enterprise leaders, the opportunity lies in making infrastructure decisions more intentional. The organizations that do this well will know which workloads belong in public cloud, which need more control, and how each choice supports cost, performance, and growth over time.

To learn more about infrastructure options for private cloud, bare metal, GPU workloads, and managed services, visit OpenMetal.