Menu
For a CFO monitoring company finances, asking the IT Director for the annual cloud budget can feel like asking "how long is a piece of string." Underutilised servers and carelessly stored data have made the cloud an expensive way to run IT infrastructure — here's why, and what to do about it.
An increasing number of enterprises are moving data and services from public cloud platforms back to on-premises infrastructure — a process commonly called cloud repatriation. This movement has gained real traction recently, as companies reassess their approach to cloud usage.
Unpredictable usage patterns and data egress fees can drive up cloud costs significantly — bringing workloads back in-house can restore cost control and stability.
Applications demanding minimal latency or high processing speed — especially those handling large data volumes — often perform better on dedicated on-premise hardware.
Strict data protection regulation requires increased oversight of how data is stored and handled — on-premise storage can streamline adherence to rules like GDPR.
Some organisations prefer the direct oversight on-premise infrastructure provides, reducing reliance on third-party cloud providers.
Relying solely on one cloud provider can restrict flexibility — repatriation gives businesses room to vary their infrastructure and avoid potential limitations.
"A 2024 Barclays CIO survey found 83% of enterprise CIOs planned to repatriate at least some workloads to on-premises or colocation data centres — up from 43% in 2020."Barclays CIO Survey, 2024 — widely corroborated, including by IDC and Citrix survey data on the same trend
Companies including Dropbox and Adobe have both undertaken cloud repatriation to gain better control over cost and performance.
Moving back to on-premises involves real financial commitment — hardware, facilities, and personnel with specialised skills.
Managing on-premise infrastructure requires strong internal IT capability and resource to handle the added operational complexity.
Most organisations land on a hybrid model, combining cloud and on-premise systems for a balance of flexibility and control.
Weighing up whether repatriation makes sense for your organisation? Talk to our team about the honest cost comparison.
Speak to our team →Predicting cloud costs can range from straightforward to highly complex, depending on the scope of your environment:
Single VMs or basic storage are easy to estimate. Complex multi-service deployments — containers, databases, AI services, serverless — are far harder.
Fixed-pattern workloads (always-on VMs, steady storage) are simple to manage. Dynamic workloads — serverless functions, auto-scaling, batch processing — are much harder to predict.
Providers price differently by region — UK costs can differ substantially from the US or Asia, complicating budget planning.
Minimal within one region, but can rise significantly for large outbound traffic volumes, inter-region transfers, or hybrid cloud environments.
Watch for API call charges on serverless/storage services, licensing fees for databases or operating systems, and monitoring/management tool costs.
On-demand pricing, reserved instances, spot pricing, savings plans, and hybrid benefit schemes all need factoring in.
Dev/test environments are simple to cost. Production environments — needing uptime, redundancy, backups, and DR — are far more complex.
Provider calculators like AWS Calculator or Azure Pricing Calculator help, but rarely account for networking limitations or scaling intricacies.
Cloud pricing models involve numerous SKUs per service, each priced differently. Add the potential for unexpected workload spikes, and usage-based billing that makes cost prediction genuinely difficult without precise usage patterns already in hand, and it's easy to see why "how long is a piece of string" is such an apt comparison.
Thoroughly evaluate your computing, storage, and data transfer requirements before committing to anything.
Take advantage of provider tools such as the Azure Pricing Calculator as a starting point.
Use monitoring and alert systems to track and anticipate ongoing expenses before they become surprises.
For intricate setups, professional input on cost optimisation and prediction can pay for itself quickly.
Begin with a modest deployment and adjust based on the costs and performance you actually observe.
Combining meticulous preparation, reliable tools, and consistent monitoring will greatly improve your ability to forecast cloud costs accurately.
Alongside repatriation of software and services, our article on data centre miniaturisation covers the falling price and shrinking footprint that's making it practical for businesses to bring some applications and storage back on-premise again.
Cloud computing still offers real benefits — but more companies are now reassessing their individual requirements, and in many cases, bringing certain workloads back on-site makes better financial and operational sense than staying all-in on cloud.
Tell us about your current cloud setup and we'll help you work out where repatriation or a hybrid approach might save real money.
Get in Touch