By Greg Bayliss-Lane, Cloud and Security Product Manager, FirstNet Technology Services
Everyone is talking about cloud repatriation, but I think the real conversation is whether businesses are putting applications in the right place.
What frustrates me about most cloud repatriation articles is that they focus on platforms instead of workloads. The question isn’t public cloud or private cloud. The question is where a workload should run and what business outcome it needs to deliver.
Customers think they want a platform, but what they actually want is an outcome: lower costs, improved resilience, predictable recovery, better performance, clearer accountability or fewer operational headaches. Whether that outcome is delivered through public cloud, private cloud, colocation or a combination of the three is often secondary.
That is why I find the repatriation discussion interesting. Not because it proves public cloud was wrong, but because it forces organisations to stop and ask whether their workloads are actually sitting in the right place.
Public cloud is still growing
Cloud repatriation is not a wholesale move away from public cloud. Flexera’s 2026 State of the Cloud Report shows that public cloud remains a significant part of enterprise infrastructure strategy, and that 73% of respondents run hybrid environments, an increase on the previous year.
That tells me organisations are not making a single choice between public and private cloud. They are using both. Some hybrid environments are designed that way from the beginning, while others develop over time through acquisitions, independent application decisions, separate technology teams and inherited platforms.
The challenge is not having more than one platform. The challenge is knowing why each workload is running where it is.
Why organisations are re-evaluating cloud
The early cloud conversation focused heavily on adoption. Businesses wanted to know what they could move and how quickly they could move it. The conversations I have today are different. Customers want to understand what value they are getting, what the service will cost over time and who is accountable when something goes wrong.
In the same report, 85% of respondents named managing cloud spend as a leading challenge. In 2023 Gartner predicted that 50% of critical enterprise applications would reside outside centralised public-cloud locations through 2027. Based on the conversations I’m having with customers, neither surprises me.
Some workloads need cloud-native services and rapid scalability. Others need predictable costs, local hosting, dedicated resources or clearly defined operational ownership. Some workloads continue moving into public cloud; some are being moved back. What matters is not the direction of travel. What matters is whether each workload is running in the right place for the outcome the business is trying to achieve.
That is very different from saying cloud has failed.
What I am seeing
When cloud adoption accelerated, moving to cloud was often presented as the obvious answer, and I expected that decision to become a no-brainer for most organisations. What I have actually seen is that customers are far more cost-conscious than I expected and are often willing to trade flexibility for predictability.
They still see real value in public cloud, but they also want predictable costs, clear accountability, known recovery outcomes and a well-defined exit path. Most customers are not trying to leave cloud. They are trying to make sure their workloads are sitting in the right place.
I also continue to see requirements that focus on the platform before the workload has been properly documented. A customer may specify availability, backup, disaster recovery or a hosting location without a complete workload inventory: server quantities, operating systems, resource allocations, dependencies, data volumes and recovery targets may still be outstanding. Without that information, it’s difficult to know whether the chosen platform is actually solving the right problem.
The exit conversation needs to happen earlier
The thing most customers forget to consider is the exit cost. They know what it costs to run the service. They often do not know what it costs to leave. Exit is not only about egress charges. It includes:
- data formats
- ownership
- access
- retained backups
- licence dependencies
- migration effort
- recovery requirements
- contract terms
- the time required to move safely
A colleague asked me about a customer that wanted to move from its existing Veeam provider to a new platform. In that scenario, the recommendation was to start a new backup chain and retain the historical backup separately as an archive. The customer’s main concern was not the new backup platform. It was ownership of the backups, access to the data and how long the historical copies needed to be retained.
Those questions should be answered when the service is designed, not when the customer decides to leave.
Five questions before choosing a hosting platform
1. What problem are you actually trying to solve?
Do not start with Azure, AWS, private cloud or colocation. Start with the business problem. Are you trying to reduce cost, improve performance, increase resilience, address a recovery gap, retire old hardware or reduce the workload on an internal IT team? If the objective is unclear, the platform comparison will also be unclear.
2. What does the workload need to achieve?
Understand what success looks like for the application: performance, availability, security, recovery, data location and operational ownership. An application that needs rapid elasticity has different requirements from a stable ERP environment, and a development platform has different requirements from a business-critical database. Treat them differently.
3. Do you understand the workload?
You need a proper inventory. Document the servers, operating systems, applications, hypervisors, resource allocations, dependencies, data volumes and expected growth. If you cannot count the workloads or explain their dependencies, you cannot size, price or migrate them confidently.
4. Is the demand steady or unpredictable?
Public cloud is well suited to workloads requiring rapid elasticity, global reach, short-term experimentation or access to cloud-native services. Stable, business-critical workloads deserve a proper private-cloud or colocation comparison, particularly where predictable costs, local hosting and clear accountability matter. The answer will not be the same for every application.
5. What does it cost to operate, recover and leave?
Do not compare only the monthly infrastructure figure. Include connectivity, licensing, security, backup, recovery and eventual migration costs. Then document the exit terms before signing the contract.
My view
In my experience, the wrong conversations start with “Should this run in Azure?” or “Should this be hosted in a private cloud?” The right conversation starts with: what problem are you actually trying to solve?
The most common mistake I see is customers choosing a platform before they have defined the outcome they are looking for. Once that happens, the conversation becomes about technology instead of business objectives. I also think too many people assume moving something to the cloud automatically makes it better. Sometimes it does. Sometimes it just moves the same problem somewhere else.
A poorly understood workload does not become a good workload because it is running on different infrastructure.
How FirstNet approaches workload placement
Because I work across private cloud, colocation, backup, disaster recovery and public-cloud opportunities, I get to see the same customer problem from multiple angles. The answer is rarely as simple as “move it to cloud” or “move it back”. Most of the time, the right answer sits somewhere in the middle: some workloads belong in public cloud, some are better suited to private cloud and others still need dedicated hardware in colocation. In many cases, the right answer is a combination.
Our job is not to force every workload onto the same platform. It is to understand the business objective, ask the right questions and design an operating model that the customer can afford, support and recover. We bring the hosting platform, connectivity, data protection and recovery requirements into one conversation, so customers understand where FirstNet’s responsibilities begin and end and what remains under their control.
Start with the objective
Customers often think they are making a cloud decision. Most of the time, they are making a business decision about cost, performance, recovery, control and accountability. That is why I would not start with the platform. I would start with the business objective and work backwards.
Once the objective is clear, the right hosting model normally becomes easier to identify. It may be public cloud, private cloud, colocation or a combination of all three. What matters is that each workload has a clear reason for running where it does.
For me, repatriation is not about proving that public cloud was wrong or private cloud was right. It is not about technology. It is about business objectives.
Related reading
Services: private cloud hosting and management, colocation, backup and disaster recovery, cloud migration and all cloud services.
Answers: Should we move to Azure or to FirstNet private cloud?, Private cloud vs public cloud, Are we locked in? and What’s the right RPO and RTO for us?
Sources
1. Flexera, 2026 State of the Cloud Report (March 2026), and Flexera press release, 18 March 2026.
2. Gartner, “Gartner Says 50% of Critical Enterprise Applications Will Reside Outside of Centralized Public Cloud Locations Through 2027” (30 October 2023).
3. Customer observations and workload-placement experience from cloud, backup and disaster recovery engagements conducted by the author.
About the author
Greg Bayliss-Lane is Cloud and Security Product Manager at FirstNet Technology Services. He works across private cloud, colocation, backup, disaster recovery and public-cloud opportunities, helping organisations align infrastructure decisions with business objectives.