Cloud vs on-premise vs hybrid: Which data storage is right?

The global computing ecosystem is continuing to expand and evolve at an unprecedented pace, forcing many organisations to reassess their information technology infrastructure requirements.

As organisations adopt AI tools and data-intensive systems, they are generating, analysing and retaining more information than ever before. At the same time, many are seeking greater control over how that data is stored, secured and governed.

There are also financial decisions to be made, shaped by the potentially high costs of storing data in the public cloud (one owned and managed by an external party, for example Google Cloud) and accessing it over time.

Storing digital assets off-site at third-party-owned and operated data centres has long been used as the go-to, default solution. However, increasingly, many organisations are looking beyond the public cloud.

Dr Nicola Sfondrini, partner, cloud infrastructure at PwC Italy, explains that when entities were deciding where to run applications and workloads 10 years ago, the first question was often whether they should use the cloud.

“Cloud economics has traditionally been built around a relatively simple assumption: that organisations should seek the best combination of performance, scalability and cost,” Sfondrini says. He points out that digital sovereignty — the ownership and control of digital infrastructure, data, technologies and IT systems — is now changing cloud economics from a narrow exercise in cost optimisation into a broader assessment of control, resilience and dependency risk.

“Digital sovereignty is making that equation significantly more complex,” he continues.

“Organisations are increasingly being asked to consider where data is stored, who operates the infrastructure, which jurisdiction applies, how dependent they are on individual providers, and whether critical workloads can realistically be moved if geopolitical, regulatory or commercial conditions change.”

The growing repatriation

Many organisations are increasingly repatriating data-heavy and legacy applications back from the public-cloud to on-premise environments. Yet data sovereignty is not the only driver.

A recent survey by US-based IT research group IDC found organisations using external cloud platform-as-a-service systems had repatriated around 18 per cent of their application workloads from public cloud to on-premise environments over the 2025–26 period. IDC said the leading drivers are workload/application performance (36 per cent) and security (30 per cent).

 

"In manufacturing or other operational settings where disruption could halt production, on-premise infrastructure remains important because latency, reliability and immediate access are critical."

 

IDC says the cloud repatriation “is a growing trend”, but adds that only 8 to 9 per cent of organisations plan a full exit from the public cloud. The remainder, it says, are organisations repatriating some workloads while keeping others in the public cloud and adopting a hybrid approach.

According to an August 2026 report by research publisher Programs, the worldwide demand for data storage centres is expected to nearly triple by 2030 to store the massive amounts of data that companies and consumers are generating and accessing. At least 225 new data centres are currently being proposed for development in Australia alone.

Thanks to the proliferation of AI applications, the volume of data being created, transferred and stored continues to rise — but that does not mean all of it belongs in the public cloud.

What is optimal?

Sfondrini believes the question is not whether companies should move from the public cloud to on-premise, but which workloads should sit where.

Sfondrini says infrastructure decisions are shifting from “cloud first” to a “what is optimal” mentality.

“Asking that question in 2019 was simple. The better question to ask now is: where will each individual workload operate at optimal efficiency based on economics, risk profile, performance needs and strategic value?” he says. This makes a hybrid cloud strategy an increasingly important consideration.

Dr Dimitrios Salampasis, associate dean and associate professor of emerging technologies and fintech at Swinburne University of Technology in Melbourne, agrees that organisations are becoming more selective about which workloads belong in the public cloud, private infrastructure, a sovereign cloud [owned by the organisation] or hybrid models.

“The public cloud remains attractive for variable demand, rapid experimentation, global deployment and managed services; while very large, stable, data-intensive or regulated workloads can justify private, sovereign or hybrid infrastructure,” he says. “Building a private cloud is very difficult.

So, I think the retainment of mainframes remains particularly relevant for banks, insurers and governments because of their high-volume transactional workloads.”

Rosalie Adriano FCPA, founder and managing director of advisory business RA Inspired, and a member of CPA Australia’s Digital Transformation Centre of Excellence, says decisions about cloud and on-premise infrastructure should be driven by business need rather than ideology.

She argues that there is no universal answer — the right model depends on the organisation’s operating environment, risk profile and tolerance for downtime.

“In manufacturing or other operational settings where disruption could halt production, on-premise infrastructure remains important because latency, reliability and immediate access are critical,” Adriano says. “By contrast, offices, distribution centres and enterprise platforms may be better suited to cloud or hybrid arrangements, particularly where flexibility and scalable capacity are valuable.”

Adriano cautions that cost comparisons can be misleading if they focus only on visible IT spend. Using the public cloud may reduce some upfront or dedicated capacity costs, but organisations must assess the end-to-end impact, including resilience, redundancy, operational continuity and downstream consequences.

Tallying the costs

The decision to repatriate data from the cloud to on-premise is more than a capital expenditure versus an operational expenditure decision, Salampasis says.

“The relevant measure is the risk-adjusted total cost per useful unit of computing, including utilisation, scalability, predictability, electricity, financing, staffing and training, latency, migration, hardware refreshes, cybersecurity and regulatory exposure.”

Transaction costs can include data egress charges [to move out of the public cloud], migration planning, application refactoring, integration work, testing, downtime risk, vendor contract constraints and the internal effort required to keep workloads portable.

Once those costs are included, a workload that appears cheaper in the public cloud on a unit-price basis may become more economical on-premise if it is large, stable, storage-heavy or subject to frequent data movement.

 

"For organisations with considerable scale, the cost–benefit equation can favour owning and operating infrastructure internally rather than paying for infrastructure-as-a-service through providers such as Amazon, Oracle Cloud, Google Cloud or Azure."

 

Salampasis, who is also a member of CPA Australia’s Digital Transformation Centre of Excellence, says transaction costs can materially change the answer because organisations need to understand each workload, the strength of the business case and how easily applications can be replaced or moved. Any decision, he says, must be grounded in financial rationality.

“Cloud can be economically strong where uncertainty and flexibility matter, but private infrastructure becomes more attractive when demand is continuous, predictable, storage-intensive or latency-sensitive.”

Talent, cybersecurity, compliance and maintenance costs can move the equation in either direction. On-premise environments may require specialist engineers, security monitoring, patching, hardware refreshes and audit controls, while cloud can reduce some maintenance burdens but introduce new requirements around vendor governance, identity management, configuration oversight, compliance evidence and cost-control skills.

Sfondrini says that digital sovereignty requirements often create additional costs through local providers, duplicated capabilities, stronger encryption, dedicated infrastructure, portability investments and tested exit strategies.

“Data residency can restrict where workloads run. Provider requirements can reduce access to global economies of scale. Portability can require architectural investments that would not otherwise be necessary. Exit strategies can introduce duplication and migration-readiness costs.”

The challenge for CFOs

Sfondrini says all of these factors create a new question for financial operations areas, because the objective can no longer be limited to determining the cheapest technically viable architecture.

“FinOps increasingly need to understand which portion of cloud spending represents consumption, which represents inefficiency and which represents the deliberate cost of maintaining control. That distinction matters. Without it, optimisation initiatives can unintentionally remove capabilities that exist specifically to reduce strategic risk.”

For CFOs, infrastructure is now a capital allocation challenge rather than a simple IT procurement decision. Cloud infrastructure investments need to be evaluated over the full life of the workload. 

Salampasis says the comparison should not stop at whether cloud spending sits in operating expenditure while on-premise infrastructure requires capital expenditure. Rather, a robust model should compare depreciation, financing costs, utilisation rates, committed cloud spend, price escalation, contractual flexibility, residual asset value and the cost of underused capacity, then test those assumptions under different growth, latency and regulatory scenarios.

“CFOs should model cloud, private and hybrid options over time, asking what to rent, own or keep portable, what capabilities and dependencies are needed, how much contractual lock-in is acceptable, and what risk appetite the organisation has for sovereignty, cybersecurity and regulation,” Salampasis says. “The best answer is to match workloads to infrastructure and economics.”

Guy Ioppolo FCPA, managing partner of business software solutions consultancy Ioppolo & Associates, says many large organisations remain reluctant to migrate heavily customised enterprise resource planning software and operating systems to only public-cloud platforms because the move can involve a very high cost, major disruption and substantial business risk.

“Data security and privacy concerns, particularly for government departments and other organisations such as financial institutions where this is a primary focus, make on-premise or hybrid arrangements attractive,” Ioppolo says.

Cloud costs tend to become less predictable when organisations move beyond sporadic or experimental workloads into large-scale, always-on environments with heavy storage, high data transfer, complex licensing, multiple regions and rapidly growing AI demand. At that scale, small changes in usage patterns, data movement or service configuration can materially change the monthly bill.

Ioppolo, who is also chair of CPA Australia’s Digital Transformation Centre of Excellence, says that “for organisations with considerable scale, the cost–benefit equation can favour owning and operating infrastructure internally rather than paying for infrastructure-as-aservice through providers such as Amazon, Oracle Cloud, Google Cloud or Azure.”

With the growing relevance of AI, he notes that locating AI engines close to organizational data can improve responsiveness and control while reducing concerns about what data leaves the organisation.

“On-premise AI has become more accessible, especially through open-source models such as Meta, Mistral and DeepSeek, and vendors are increasingly packaging hardware with preloaded AI and enterprise applications,” he says.

However, Ioppolo draws a clear distinction between large organisations and small-to-medium enterprises, including most accounting firms.

“For SMEs, cloud is more likely to make sense because it offers greater flexibility, lower ongoing infrastructure burden and avoids the need for specialist staff to maintain on-premise environments,” he says. “Overall, on-premise, cloud or hybrid choices should be made according to business size, cost, information security, disruption and operational needs.”

A 2025 report by Deloitte, The AI infrastructure reckoning: Optimising compute strategy in the age of inference economics, found that organisations are quickly discovering their existing infrastructure may not be designed for AI’s demands.

The report states that AI workloads are driving the emergence of “AI factories”: integrated infrastructure ecosystems specifically designed for AI processing that bring multiple specialised components into a single solution.

“These AI factories can also offer excess computing capacity through service models, allowing organisations to monetise unused processing power while maintaining strategic control over critical workloads.”

Data centres will need to continue to evolve to accommodate increasingly specialised tools for specific applications, it continues. “The organisations that successfully navigate this infrastructure transformation are likely to gain sustainable competitive advantages in AI deployment and operation.

Those that fail to adapt are likely to face escalating costs, performance limitations and strategic vulnerabilities as AI becomes increasingly central to business operations.”

“Just as cloud computing reshaped IT strategy over the past decade,” the report notes, “hybrid AI infrastructure will probably define technology decision-making for the decade ahead.”

 

_{Production credit
Banner image Olga Zarytska/Moment via Getty Images}