For years, the cloud has been touted as a way to achieve greater flexibility, faster service deployment and more cost-effective IT infrastructure spending. Today, that promise remains, but it is becoming less self-evident: more and more companies are discovering that the ‘pay-as-you-go’ model works best when an organisation truly knows what it is using, who is responsible for it, and what business value it creates on the other side of the invoice.
This is not a story about the end of the cloud or about disillusionment with technology. Rather, it is about the market maturing. The cloud is ceasing to be a simple symbol of modernisation and is becoming one of the most demanding areas of cost management within a company. This is particularly true where infrastructure is growing faster than control processes, and successive teams are provisioning resources faster than finance can link them to a specific product, customer or business outcome.
The scale of this phenomenon clearly illustrates the market’s growth rate. According to Omdia data, global spending on cloud infrastructure services in the fourth quarter of 2025 rose to $110.9 billion, up 29 per cent year-on-year, whilst full-year expenditure in 2025 stood at $399.6 billion. One of the main drivers of this growth was the transition of AI projects from the experimental phase to production deployments, which increases demand not only for computing power but also for storage and networking.
For the CFO, the problem begins when the cost is visible but its source is no longer so obvious. A cloud invoice shows the amount, but does not always answer the question of whether the company is paying for business growth, the development of new services, inefficient architecture, test environments left over from completed projects, or AI experiments whose impact on the bottom line is yet to be assessed.
In the traditional model, infrastructure expenditure was more periodic and predictable. The purchase of servers, licences or hardware required a decision, approval and a budget. The cloud has shifted some of this decision-making closer to the technology and product teams. This has given companies speed, but at the same time introduced a new level of volatility. Costs can arise immediately, build up gradually and, over a long period, appear to be a natural consequence of scaling, although in practice they are often a mixture of valuable growth and simple waste.
It is precisely here that the flexibility of the cloud reveals its other side. Resources are easy to provision, but it is harder to manage them consistently. Redundant instances, unused databases, excessive computing resources, old backups or development environments running longer than they should do not necessarily stand out as a single cost item. On an organisation-wide scale, however, they can create a constant, hard-to-pin-down financial burden.
Artificial intelligence has become an additional catalyst. In many companies, AI has triggered a new wave of cloud adoption, as projects based on data, models and automation require infrastructure whose cost is not always easy to predict at the pilot stage. An experiment that initially appears minor can, once scaled up to include more users, data or processes, significantly alter the structure of expenditure. This is particularly important because AI shifts the conversation about the cloud from the IT level alone to the level of business strategy, productivity, risk and profit margins.
This is why FinOps is becoming increasingly important, understood not as yet another cost-cutting measure, but as a common language for finance, technology and business. In industry literature, FinOps is described as the practice of maximising the business value of the cloud through shared responsibility amongst engineering, finance and business teams. One of the challenges, however, remains that billing data comes from multiple sources and has varying formats and taxonomies, which makes it difficult to translate this data quickly into management decisions.
In practice, therefore, the aim is not for the CFO to delve into the technical details of cloud configuration, nor for the CIO to reduce innovation to mere cost-cutting. A more mature dialogue begins when both parties share the same view: which resources support growth, which sustain critical processes, which are the cost of experimentation, and which simply no longer have a justification.
This also changes the way the success of cloud projects is assessed. Migration to the cloud alone is no longer sufficient proof of modernity. The ability to allocate costs to products, processes and outcomes is becoming increasingly important. For some organisations, a practical starting point will be better resource tagging; for others, more accurate consumption forecasting; and for others still, a discussion about which workloads should actually run in the public cloud and which function better in a hybrid model.
The most important thing, however, is to shift one’s perspective. The cloud is neither cheap nor expensive in and of itself. It is only as effective as the management model a company builds around its use. A bill that takes the CFO by surprise is not always a sign that the technology has failed. It is often the first clear sign that an organisation is adopting the cloud faster than it is developing financial transparency, accountability and the ability to measure value.
In this sense, controlling cloud costs does not mean stifling innovation. On the contrary, it can become a prerequisite for its further scaling. Because the greater the cloud’s significance to the business, the less sufficient it is to simply ask how much it costs. The question of what actually works better thanks to it is becoming increasingly important.
