For years, FinOps has been associated mainly with monitoring public cloud bills. However, this model is no longer sufficient. An increasing proportion of technology expenditure is now spent on SaaS applications, data platforms and AI services, which are often purchased and used outside the direct oversight of the IT department.
This shift is taking place at a significant juncture. Gartner forecasts that global spending on AI will rise by 47 per cent in 2026, to $2.59 trillion. At the same time, many organisations expect investments in artificial intelligence to be at least partly funded by savings in other areas of technology. FinOps is therefore no longer merely a method of reducing infrastructure costs. It is becoming a means of financing the next wave of digital investments.
The cost of technology has moved beyond IT
According to the State of FinOps 2026 report, 98 per cent of the FinOps teams surveyed are already managing AI expenditure or are preparing to do so. In the case of SaaS, this figure stands at 90 per cent. A year earlier, it was 65 per cent. The scope of responsibility is also expanding to include licences, private clouds and data centres.
The reason is simple. The IT budget is increasingly failing to reflect the full cost of technology. Tools are being purchased by sales, marketing, HR and product teams. Charges appear on company credit cards, in local contracts and as add-ons to systems already in use. As a result, a company may have good control over infrastructure costs, whilst at the same time losing control over its entire software portfolio.
Data from Zylo illustrates the scale of the problem. In a survey of 218 IT leaders, as many as 78 per cent of respondents had encountered unexpected charges related to AI or usage-based billing. Over 60 per cent had to scale back other projects as a result. At the same time, SaaS expenditure rose by 8 per cent over the course of the year, even though the number of applications in use remained virtually unchanged.
The licence is no longer the primary unit of cost
Traditional SaaS management has primarily involved counting users and removing unused accounts. This remains important, but is increasingly insufficient. Providers are moving towards billing models based on credits, tokens, the number of queries, data volume or operations performed.
In this model, the cost is not known at the time the contract is signed. It is incurred as the product is used. The same application can be inexpensive for one team and very expensive for another, depending on the configuration and intensity of use.
Therefore, organisations should measure not only the cost of the application, but also the cost of a specific outcome: a customer served, an analysis performed, a document processed, a query to the model, or a process carried out by an AI agent. The FinOps Foundation recommends combining financial data with information on usage and business value, as well as collaboration between finance, procurement, IT and product owners.
The greatest value comes from visibility
Market experience shows that the primary source of savings is not price negotiations, but rather identifying the true scale of technology usage. Adobe reports $60 million in savings and avoided costs thanks to streamlining its software management. AMN Healthcare states that, following the restructuring of its data environment, it reduced monthly costs from around $200,000 to $14,000, whilst storing more data.
These are case studies published by technology providers and should therefore not be treated as universal benchmarks. However, they demonstrate a repeatable pattern: costs fall when a company is able to align contracts, invoices, technical configuration and actual usage.
FinOps outside the cloud should not mean setting up yet another team to check invoices. What is needed is a shared model of accountability for technology expenditure. Every significant application, data platform and AI service should have a business owner, a usage metric and a clearly defined value.
The FOCUS standard, developed by the FinOps Foundation, can help by standardising cost data from the cloud, SaaS, AI, data platforms and data centres. It also covers virtual billing units such as tokens, credits and DBUs.
The cost of technology cannot be analysed only after an invoice has been received. It must be a factor in decisions regarding the choice of tool, its configuration and the scale of its use. FinOps is therefore becoming not so much a cost-saving system as a system for making better investment decisions.
