Forrester: AI Bills Will Drive Up Software Prices in 2027

Forrester has prepared an unpleasant forecast for IT departments for next year: software and data budgets will rise as technology vendors grow bolder about passing the cost of maintaining artificial intelligence on to customers. The company's research shows that the shift in billing models rolled out by the biggest AI firms in recent months is only just beginning to show up on corporate customers' bills.

The end of flat rates

For years, software vendors sold access through a subscription model: a fixed monthly or annual fee regardless of how intensively a company used the tool. Forrester points out that this model is cracking under the weight of AI costs. Providers of language models and development platforms are shifting to billing based on actual usage, most often measured in tokens or queries.

Anthropic, OpenAI and GitHub are the three names Forrester cites explicitly as companies that have changed their approach to parts of their services over the past six months. Instead of a single subscription rate, customers are increasingly paying for actual model usage, which in practice means bills that are harder to predict than under a classic subscription.

Microsoft raises the bar

Forrester also points to Microsoft's decision to introduce a premium E7 license. It adds access to M365 Copilot, the Agent 365 agent and additional security tools on top of the existing E5 package. It's another example of a mechanism the firm's analysts describe bluntly: vendors aren't cutting the base prices of their services, they're adding new, pricier layers with AI-based features.

For procurement departments, this means that simply sticking with an existing package no longer guarantees a stable cost. Software makers are structuring their offerings so that the most useful AI features land in the pricier license tiers.

Headcount budgets rise too

The pressure on spending isn't limited to software alone. The survey found that 67 percent of technology decision-makers plan to increase hiring budgets in 2027, 23 percent want to keep them at current levels, and only 10 percent expect cuts. Among analytics teams, as many as 68 percent of respondents anticipate growth in headcount budgets. In 2025, personnel costs accounted for 35 percent of total IT spending, so any shift in this area has a major impact on a company's overall bill.

What this means for companies in Poland

For Polish companies and integrators that buy Microsoft licenses, GitHub Copilot, or access to Anthropic and OpenAI models through local partners, Forrester's findings have a direct bearing on IT budget planning for 2027. It's worth checking now which contracts include usage-based billing clauses and reviewing the token consumption history of the tools the company relies on most heavily.

Forrester recommends that companies adapt their FinOps practices to a new reality in which costs depend on token consumption rather than a fixed, predetermined rate. Classic IT spend-control tools, designed for flat subscriptions, do a poor job forecasting bills that grow with the scale of model usage.

Organizations that perform best in 2027 will not be the ones that spent the most on AI. They will be the ones that invested in the fundamentals: trusted data, strong governance, organizational readiness, and the ability to continuously adapt to changing technology and customer behavior - Sharyn Leaver, Chief Research Officer, Forrester

The investment scale behind it

The rising bills customers face stem from the scale of infrastructure investment on the vendor side. Bain & Company estimates that the cost of building data centers to support artificial intelligence will reach $2 trillion by 2030. A KPMG survey cited in connection with similar analyses shows that nearly a third of business leaders struggle to understand and control operating costs when deploying AI at scale, which helps explain why billing has climbed to the top of IT departments' list of concerns.

Forrester's findings don't mean companies should abandon AI-based tools. They suggest, rather, that buying such services without clear usage-monitoring rules and negotiated cost caps is becoming increasingly risky financially. According to the forecast, 2027 will be the first year these costs fully hit corporate customers' budgets.