Public cloud spend keeps climbing and so does the waste. A practical guide to figuring out which workloads belong in the public cloud, which belong closer to home, and how to decide without guessing.
Public cloud platforms promised elasticity, scalability, and a shift from CapEx to a predictable OpEx model. For a decade, that promise fueled fast, broad adoption across every sector.
But the reality has proven more complex. Instead of cutting waste, many organizations found a significant share of their cloud spend eaten up by inefficient usage and failures, straining budgets and pulling funds away from higher-value work like customer experience, technical debt reduction, and AI.
Not every workload thrives in the public cloud. Some need costly adjustments to get real value from it. Others need tighter security and control than a shared environment can offer. That mismatch is what’s pushing IT leaders to reassess their infrastructure strategy altogether, not to retreat from the cloud, but to put each workload where it actually performs best.
of IT leaders say at least 20% of public cloud spend is wasteful. More than half put that number above 40%.
in inefficient public cloud spending in 2024 alone, on a worldwide spend base of roughly $600B.
A growing number of organizations are confronting the same hard truth: their swift migration to the public cloud may not have aligned with their operational goals or financial expectations. Click each one to see why it matters.
Enterprises looking to pull workloads back should move carefully. The same enthusiasm that drove rapid cloud adoption can just as easily drive a knee-jerk overcorrection, with its own cost and efficiency fallout.
Repatriation is accelerating, and public cloud spend is still climbing. Both are true at once, because most enterprises aren’t leaving the cloud. They’re building a hybrid stack that puts each workload where it performs best.
“The problem is that growing organizations don’t think about cloud, they fall into it. But they usually find that beyond a certain threshold, running their own infrastructure is more economical, certainly for their core workloads.”
There’s no one-size-fits-all repatriation strategy. Start with a thorough audit of your IT estate, and get honest answers to four questions:
Use the answers to sort your workloads against the infrastructure options below. Click a column header to spotlight it.
| Feature | Public Cloud (Azure) | Private Cloud (ReliaCloud, OpenCloud, VMware) | SaaS |
|---|---|---|---|
| Best for | Scalable, global public cloud workloads | Traditional enterprise or scale-out workloads | CRM, HRM, and similar functions |
| Common use cases | Analytics, cloud-native apps, web applications | Edge computing, HCI, VDI, hybrid environments | Various, vendor-defined |
| Compliance | Dependent on service provider | Very high; full control in regulated sectors | Dependent on vendor |
| Cost efficiency | Pay-as-you-go; can be costly long-term | Moderate to high; committed resources | Flat-rate subscription |
| Customization | Low to moderate | High, with open-source flexibility in US Signal OpenCloud | Low; vendor-dependent |
| Disaster recovery | Azure Site Recovery | Requires in-house strategy; VCDA, Zerto in VMware | Managed by vendor |
| IT management | Managed by service provider | Requires expertise specific to the technology (Nutanix, VMware) | Minimal; managed by vendor |
| Performance | High; can be affected by noisy neighbors | High | Dependent on vendor |
| Scalability | Very high | Moderate to high | Managed by vendor |
| Security | High; shared responsibility model | Strong, enterprise-grade security | Low to moderate; vendor-dependent |
Tip: sort your own workload list against this table before you commit to a repatriation plan.
“Organizations need a sorting mechanism for workloads. The public cloud is just another tool in their toolbox, and it might not always be the best fit for the business. For example, it’s suitable for occasional development tests, but if you have a test environment that operates 24/7, it’s best to use a private cloud.”
Not all workloads are created equal. Some, like cloud-native applications, are built to exploit everything the public cloud offers. Others need the control and security a private cloud provides. Filter the categories below by where they typically belong.
Answer three questions and we’ll point you toward the infrastructure model most likely to fit, based on the same criteria IT leaders use to build a repatriation plan.
Not a substitute for a full audit, but a fast, honest starting point.
As IT waste eats into budgets, streamlining where and how you host every workload is the highest-leverage move available. US Signal offers a full range of private and public cloud solutions to make that transition seamless, secure, and cost-effective, whether you’re planning full repatriation, a hybrid strategy, or a cost-effective alternative.