
The software budget has developed a peculiar habit: it keeps growing even when nobody makes a decision to increase it.
An employee signs up for an AI assistant. Another department chooses a competing product. Both subscriptions renew automatically. Before long, the finance team is paying for tools that do much the same job, often without knowing how many people use them.
Across Europe, such spending habits could be contributing to a multibillion-euro problem.
According to Spendesk’s European Spend Report 2026, businesses with 50 to 250 employees could collectively save around €7bn a year if their software spending matched that of the most cost-conscious comparable companies.
The research, based on 12 months of payments data from around 2,500 European businesses, shows how widely software costs vary between similar companies. It also points to a growing problem for finance teams: while businesses can buy and use software more easily than ever, keeping track of the costs is becoming harder.
AI is adding to the pressure. New tools often enter businesses before finance teams have agreed how to manage them, while usage-based pricing makes future bills difficult to predict.
Axel Demazy, CEO of Spendesk, believes this is exposing the limits of traditional financial oversight.
“Much of this growth is driven by urgency rather than strategy,” he says.
This leaves CFOs with a difficult question: how much of the technology budget is delivering value, and how much is going to waste?
The hidden cost of software sprawl
Software purchasing has become remarkably easy. An employee can subscribe to a new service in minutes, often without involving procurement. For growing companies, this flexibility can help teams respond quickly to changing needs.
The trouble comes later, when subscriptions accumulate and responsibility for managing them becomes unclear.
Spendesk analysed payments for more than 35 widely used software products, comparing monthly spending per employee among businesses of similar sizes. Some companies were paying several times more than their more cost-conscious peers.
Monthly spending on Atlassian ranged from €2 to €12 per employee, while HubSpot costs varied from €6 to €24.
Different product packages and contract terms can explain some variation. But Demazy identifies a deeper problem: companies have little insight into what their peers actually pay.
Software vendors publish prices, but negotiated rates can vary substantially. Without reliable benchmarks, finance teams may struggle to establish whether their contracts represent good value.
Across the companies studied, Spendesk identified about €69m in potential annual savings from pricing differences. Extrapolated to European businesses with 50 to 250 employees, that estimate reaches €7bn.
The figure represents a modelled opportunity rather than confirmed waste. Yet it suggests that more disciplined purchasing could release substantial funds.
Duplication adds another burden. Spendesk estimates that eliminating overlapping software subscriptions could free up a further €3bn annually across the same European business segment.
Among companies purchasing AI tools, 72% pay for two or more competing products. The figure is 64% for design software.
Demazy attributes much of this duplication to teams solving similar problems independently. A company card is often all an employee needs to acquire another tool. By the time finance becomes aware of the purchase, it may already be embedded in departmental workflows.
Some overlap is justified. Different tools may offer specialist capabilities that warrant keeping both. The difficulty is establishing where that is the case.
Without a shared view of purchases across departments, unnecessary subscriptions can survive repeated renewals without scrutiny. Finance teams may only discover the problem when they review spending months later. And as more software providers adopt usage-based pricing, even knowing what the next bill will be is becoming harder.
AI is making the software bill harder to predict
AI spending presents an even tougher problem for finance teams. Spendesk found that total AI expenditure increased by 340% over 12 months, even as the number of companies buying AI products changed relatively little. Existing users are spending far more as they expand their use of the technology.
Demazy argues that traditional budgeting is no longer suited to the way businesses use AI. “Stop budgeting AI like a subscription,” he advises.
Conventional software licences generally follow predictable payment schedules. Many AI products instead charge according to consumption, with costs increasing as employees make more requests or automated systems complete more tasks.
An AI agent running thousands of requests could generate a substantial bill without anyone approving an additional subscription.
Spendesk’s figures illustrate the resulting volatility. Among 1,347 companies with at least three consecutive months of AI expenditure, the typical company’s monthly bill varied by around 61%. Its highest monthly bill was approximately eight times its lowest.
Demazy believes finance teams need to respond to spending changes much sooner. “CFOs need real-time controls rather than an annual review: a spending limit with an alert layer that catches a spike the week it happens, not at month-end close.”
With earlier warning of rising costs, finance teams can intervene before an unexpected bill arrives.
But real-time spending controls are only part of the answer. A higher AI bill could reflect greater productivity, or it could mean that a business is paying more without seeing much benefit.
Finance teams therefore need to work more closely with the people using these tools. That means understanding what the business gets for its money, as well as monitoring how much it spends. It also requires more frequent decisions about where to invest, when to intervene and which costs are worth keeping.
Bringing company spending back into view
Tracking what the business has already spent is no longer enough. Finance teams need to see where money is going as it happens, so they can respond before costs get out of hand.
That requires more than a detailed budget or a monthly report. When purchases are scattered across corporate cards, invoices and departmental budgets, finance teams can struggle to see where costs overlap or why they are rising.
AI presents an opportunity to fundamentally rethink financial management. Rather than spending their time gathering data, processing transactions and reviewing costs after the fact, finance teams could increasingly rely on intelligent systems to automate routine work, identify emerging risks and surface the decisions that need real human attention. The result is a finance function that can anticipate problems rather than simply respond to them.
Spendesk is putting this approach into practice through its AI automation platform, which combines spending visibility with tools that reduce manual finance work.
Spendesk AI Connect, for example, allows finance teams to query live spending data through AI assistants such as Claude or Dust. Users can ask questions about company expenditure without preparing reports manually.
Its Bookkeeping Agent, currently in beta, helps classify payables using previous accounting decisions and flags transactions that need attention. Smart Invoice Routing, also in beta, directs incoming invoices to the appropriate company entity, leaving finance teams to deal with exceptions.
Demazy sees an opportunity to rethink how finance teams spend their time. AI could take on much of the work involved in collecting figures and processing transactions, leaving people to investigate problems and make decisions that require human judgement.
“AI handles the volume, finance decides what needs a second look,” he says.
That distinction matters as businesses rely on more software and AI becomes part of everyday work. Automated systems can help identify duplicate subscriptions, flag unusual spending and reduce routine processing. But finance teams must still decide whether those costs are justified and where investment will deliver the greatest value.
The €7bn savings opportunity shows how much businesses could gain from taking a closer look at software spending. Yet the wider challenge is not simply to cut costs. As AI changes how companies spend money, finance teams will need to move beyond reviewing what has already happened and take a more active role in deciding what should happen next.
Explore Spendesk’s European Spend Report 2026 for further insights into SaaS pricing, software duplication and the changing economics of AI expenditure across Europe.
The software budget has developed a peculiar habit: it keeps growing even when nobody makes a decision to increase it.
An employee signs up for an AI assistant. Another department chooses a competing product. Both subscriptions renew automatically. Before long, the finance team is paying for tools that do much the same job, often without knowing how many people use them.
Across Europe, such spending habits could be contributing to a multibillion-euro problem.