Cloud was supposed to be cheaper. Now, many engineering teams are doing the math and finding out that infrastructure decisions made a few years ago no longer make sense. This does not mean that every cloud strategy is a failure; it’s just an indication that cloud platforms aren’t ideal for every organization and project.

Cloud migration meant moving applications and workloads to cloud infrastructure managed by public cloud providers like AWS, Azure, and Google. As cloud offerings matured, cloud migration expanded to include “cloud-to-cloud” migration, where the move is from one cloud provider to another, and now repatriation.

In this article, we will explore cloud repatriation, a move to owned infrastructure often viewed as a reverse cloud migration. 

What is Cloud Repatriation?

Cloud repatriation is the process of moving applications, data, and cloud workloads from public cloud infrastructure to other hosting environments. 

Repatriation involves moving data from the cloud to on-premises infrastructure, for example, on-premises data centers. The move can be a full repatriation, resulting in a fully on-premises infrastructure, or partial, resulting in a hybrid cloud infrastructure.

Colocation (colo) is also another repatriation option. Colocation is a good option if you don’t have your own data center and want to manage your own hardware. It involves renting space at a data center. You, however, need to buy, configure, and manage hardware. 

With cloud repatriation, organizations can build two private cloud computing models: an on-premises private cloud hosted at their own data center or a hosted private cloud at a colocation facility.

How Unexpected Costs Drive Cloud Repatriation

Cost remains the key driver behind cloud repatriation. However, there are other contributors like vendor lock-in, security and compliance requirements, and a need for more control of your data and infrastructure.

The promise of the cloud was that you could scale as needed. You didn’t have to manage infrastructure. You could have redundancy, enabling high availability. But that promise has come at a substantial cost.

There are many factors that may contribute to unsustainable cloud spending.

Surprise Cloud Bills

Cloud spending can leave a significant dent in your opEx. A quick Google search will reveal stories of surprise cloud bills affecting organizations of every size, from startups to Fortune 500 companies.

In most cases, higher-than-expected cloud bills aren’t caused by a single misstep. They are the cumulative result of architectural decisions, reliance on default configurations, and opaque pricing models. The key factors are misconfigured resources, overprovisioning, auto-scaling issues, zombie infrastructure, and high data egress fees obscured by complex billing.

How Much Cloud Spending is Too Much?

For some companies, it’s not the occasional surprise bill that gets to them. It’s the sheer amount of spending that goes to the cloud. 

A well-documented example of companies moving away from the cloud is 37Signals. Their CTO, David Heinemeier Hansson, highlighted that the company spent $3.2 million on AWS costs in 2022. And this was after doing everything they could to optimize their spending.

This kind of spending kick-started their data migration from AWS. The move back to on-premises is still ongoing for their BaseCamp and HEY products.

Cloud Sprawl

Cloud sprawl refers to the unmanaged growth of cloud resources within an organization. 

Creating cloud resources on most public cloud services is easy. It is, however, harder to track the tools and technologies that every individual on every team is using.

Over time, these tools may inflate the cloud bill, and visibility is limited. Cloud sprawl also results from misconfigured, overprovisioned, forgotten, or unused resources. 

Cloud Egress Fees

Cloud repatriation, at face value, may seem like the all-round cost-effective solution or a way to get rid of vendor lock-in forever. Cloud egress costs can be inhibiting, though.

In early 2023, Holori cofounder Antoine Jeol posted egress costs from several cloud providers, based on publicly available estimates.

Larger public cloud providers have higher egress costs. Shouldn’t hyperscalers benefit from economies of scale?

The biggest public cloud players waived egress fees in 2024. However, depending on factors like whether you are moving to another provider or on-premises, charges may still apply. 

Egress fees, like their cloud bill counterparts, are obfuscated. Microsoft, for example, only offers the first 100 GB per month free for the “Internet Egress” service routed via their Premium Global Network service. 

Microsoft’s Internet Egress service allows you to transfer data out of Azure to the public internet via Microsoft’s premium network infrastructure.

Lift and Shift

Many companies migrated to the cloud without refactoring their applications and workloads to be cloud native.

Refactoring is a cloud deployment strategy that means rethinking your architecture and redesigning it for cloud environments. You could re-architect from a monolithic architecture to a serverless one, for example. Refactoring is driven by a strong business need to improve performance and scale.

While refactoring could be the best cloud migration strategy, it can also be time-consuming and expensive in the short term. It’s also difficult to refactor while ensuring user uptime. 

On the extreme end of refactoring is rehosting, also known as lift and shift. 

Lift and shift means simply moving from one hosting option to another without any changes to your core architecture. Cloud migration strategies like lift and shift ended up bloating cloud bills for many companies.

Other Cloud Repatriation Drivers

More cloud repatriation case studies show that this trend can be attributed to other factors beyond cost efficiency.

According to a Citrix survey, other contributors to cloud repatriation include unexpected security and compliance issues, or a failure to meet or set internal expectations. Performance issues, compatibility, and service downtimes were also cited as factors. 

When Should Business Leaders Consider Cloud Repatriation?

Some use cases are better suited for the cloud than others, and repatriation is not necessary in most cases, nor would it be a good choice for every industry.

While cost can be a catalyst, it’s necessary to examine other factors. A rule of thumb when thinking about cloud repatriation is the nature of your workloads: what works better on the cloud and what could benefit from alternative infrastructure. Other factors to consider include your level of scale, regulatory compliance, seasonality, and more.

Workload Characteristics

If your workloads need a lot of storage, then repatriation might make sense for you. Storage requirements like backup and redundancy also drive up cloud costs.

37Signals was spending around $1.5 million a year on Amazon S3, which was roughly half of their annual $3.2 million cloud bill. This shows how fast storage costs can add up.

Other workload characteristics to consider are compute and networking. If you have fairly predictable compute and networking needs, then on-premise or colocation solutions can be a great option. 

Level of Scale

Cloud computing makes more sense for small businesses that do not use much resources. Payment models like reserved instances and upfront payments are also more suitable for small businesses. 

Storage may be underutilized when the business is still small, and compute and network resources may still be bursty and unpredictable. It is therefore typically cheaper to work with a cloud provider.

Repatriation may not make sense until you reach a certain level of scale. You may end up spending more if you go with an option like two or more colocation sites as you also need network connectivity.

However, as your business grows, networking, compute, and storage resource use can increase exponentially, especially when coupled with cloud sprawl and architectures that are not optimized for the cloud. It may be cheaper to buy your own hardware or go with colocation options. 

You may need to hire people to help set up the infrastructure and maintain it. However, even after adding up these human resource costs, the initial investment can be worth it, as you can make significant cost savings in the long run.

Additionally, maintenance may not impact costs too much.

Compliance and Regulations

Some data protection laws require the storage of data within particular geographic areas. For example, if you are a health care provider in Florida, you need to ensure that all patient information is physically maintained in the United States or Canada.

If you plan to provide services for a highly regulated industry, you may need to move to an on-premise solution or a hybrid cloud.

How To Move From Cloud to On-Premises

For many companies, the move to the cloud in the 2010s was a head-first dive into uncharted waters. A prudent tech leader will want to take a step back to avoid making the same mistakes.

You may find that for your company, simply changing cloud providers or using a multicloud approach could be the best option, a cloud migration strategy known as “repurchasing.”

The cloud repatriation process involves careful planning and consideration of future scalability needs.

Step 1: Do the Math

The first step is to thoroughly examine your current cloud expenditure. Break it down by category and service: compute, network, and storage.

Think about whether your spending is fully optimized. There’s a high likelihood that you have some cloud sprawl, and you could further streamline your costs. Clean house.

If the costs still seem unacceptable and your FinOps agree, then it’s time to start considering other options. A cost-benefit analysis between cloud and on-premises solutions would help identify the more cost-effective option.

At this point you are closer to starting your migration. However, you need to calculate other costs like public cloud egress fees and hardware and staffing costs for on-premises systems or colocation.

You might also need to upskill your team or hire new talent for your on-premises environment options. Make sure you know what you’re up against.

Step 2: Analyze Your Infrastructure

You don’t want to do another lift and shift. Check whether your current infrastructure is on-prem native.

Identify your applications and workloads. How much storage are you consuming for your data and backups? How much replication have you done? What about compute and networking resources? 

Make sure you understand every bit of your cloud infrastructure and how it impacts the costs. 

Step 3: Decide whether to go hybrid or full on-premises

If you decide to repatriate, you need to decide whether to go full on-premises or hybrid. 

There are workloads that make more sense in the cloud, for example, user-facing mission-critical applications and web servers.

These tend to serve users all around the world, and a public cloud may handle traffic spikes better. You could also take advantage of Content Delivery Networks (CDNs) for caching, to reduce latency.

You may decide to repatriate storage, backup, networking, and compute, especially where these are relatively predictable. The important thing is to ensure that critical workloads access the resources they need.

Step 4: Replatform to Open source

Consider what parts of your IT infrastructure could benefit from open-source solutions. While open source is not a repatriation strategy by itself, the tech stack that you use could make a huge difference in your architectural costs.

You could decide to work with a stack like Linux + Kubernetes + OpenStack and reap the benefits of a robust ecosystem and large open-source community. Some industry-standard tools are native to these ecosystems.

While there might be a learning curve, you would be building truly open private cloud solutions without worrying about vendor lock-in. You can select a variety of vendors that work with open source tech, avoiding proprietary solutions.

Step 5: Create Your Migration Plan

Begin by buying and configuring your on-premises hardware. Then, talk with the cloud provider about how much data you can migrate from the public cloud environment per hour, per day, or even per month. Depending on the amount of data you are migrating, egress could take months. Set milestones and get ready for the big move.

Future Outlook: Cloud Repatriation Trends in 2025 and Beyond

Cloud repatriation is still primarily a response to optimizing costs. According to the CIO Cloud Trends Survey and Report by Azul, CIOs are taking a multi-pronged approach to managing cloud spending. 

Cost optimization efforts are still the preferred option for many CIOs, ranging from taking advantage of built-in cloud provider cost management tools, monitoring, leveraging public cloud providers’ discount programs, FinOps, newer technologies (ARM chips and high-performance runtimes) to payment plans like reserved instances and savings plans.

Repatriation is also gaining momentum as a cost-saving measure, with 22% of CIOs stating that they plan to repatriate some workloads. Additionally, 17% plan to cut or delay projects to cut expenditures.

The Flexera State of the Cloud Report showed that 21% of cloud workloads and data have been repatriated thus far.

The future of the cloud will likely be hybrid. Hardware is now far more capable than 10-15 years ago. Data centers and colocation facilities can capitalize on faster network infrastructure and server CPUs with up to 128 x86 cores to run multiple workloads across virtualized and containerized architectures. 

Businesses can also rent racks in colocation centers, so there is no need to buy every single piece of hardware for your on-premises infrastructure.

Final Thoughts

If the rush to the cloud has taught us anything, it is that there is price to pay for the widespread shiny object syndrome in the IT industry. Economic factors were instrumental in driving cloud adoption. The numbers did add up at some point, when the ubiquitous message was “move to the cloud.”

Today, economic factors drive the conversation around repatriation. As startups become scaleups and enterprises, they start considering alternatives to streamline operations and avoid lock-in. Organizations look for solutions that align with their business objectives. For many companies, the cloud made a lot of sense ten years ago, as they were experiencing exponential growth. However, for mature organizations, cost efficiency is the primary consideration.

If you choose to repatriate, we hope this article serves as a reference point and guide to help you do the proverbial look before you leap.

If you’re considering repatriation but lack the in-house capacity to execute, BairesDev can help. Our engineers bring deep infrastructure experience and can integrate quickly with your internal teams. Book a call today.

Frequently Asked Questions

When should you repatriate from the cloud?

Cloud repatriation makes sense when public cloud costs become unsustainable, especially for predictable workloads that no longer benefit from elastic scaling. It is also worth considering when performance tuning, data control, or regulatory constraints require more ownership over infrastructure. Most companies do not exit the cloud entirely but selectively move specific workloads.

What workloads are best suited for cloud repatriation?

Workloads with stable usage patterns, high data transfer volumes, or long-term storage requirements often benefit most from repatriation. These include backup systems, large-scale databases, and internal services that do not rely on cloud elasticity.

What are the biggest risks of cloud repatriation?

Repatriation introduces upfront capital expenditures, staffing demands, and increased operational complexity. If poorly scoped, it can lead to infrastructure sprawl or degraded performance. Without disciplined planning, companies risk trading one set of inefficiencies for another.

What role do hybrid architectures play in a repatriation strategy?

It offers a middle ground between public cloud and fully owned infrastructure. Hybrid infrastructures allow organizations to shift targeted workloads on-premises while continuing to leverage cloud services for customer-facing or burstable systems. This flexibility helps reduce risk and support phased migrations.

What compliance or regulatory issues can repatriation help address?

Repatriation can support compliance with data residency, sovereignty, and industry-specific regulations that require full control over physical infrastructure. Hosting sensitive workloads on dedicated systems allows for stricter access controls, auditable processes, and clearer enforcement of geographic data boundaries. For highly regulated sectors, this control can reduce legal risk and simplify certification efforts.