Your Software Bill Is Going Up. AI Is Why.
A major tech research firm warns that AI vendors are passing their enormous infrastructure costs on to business customers through higher prices and unpredictable usage fees.
Key points
- Forrester, a technology research company, surveyed more than 2,600 business decision-makers and found that 80 percent expect their data and software budgets to rise in the coming year.
- Anthropic, OpenAI, and GitHub have each shifted some services from flat monthly fees to pay-per-use billing in the last six months.
- Staffing still accounts for 35 percent of IT budgets in 2025, and 67 percent of tech decision-makers plan to increase their staffing spend by 2027.
- Consulting firm Bain and Company estimated in 2024 that building AI data centres globally will cost roughly 2 trillion dollars by 2030.
- KPMG found in July 2025 that nearly one in three corporate leaders struggles to understand and control costs when rolling out AI across their organisation.
If your company uses business software, expect a bigger invoice next year. That is the headline finding from Forrester, a technology research firm that tracks how companies spend on tech. Its new report, based on a survey of more than 2,600 business and technology leaders, says vendors are raising prices and adding usage fees specifically to recover what it costs them to run AI.
The shift is already visible. In the last six months, Anthropic and OpenAI, the two most prominent AI companies right now, plus GitHub, the coding tool owned by Microsoft, have all moved some of their products away from simple flat monthly subscriptions. Instead, customers now pay based on how much they actually use. That sounds fair in theory. In practice, it makes budgeting much harder.
Microsoft gets a mention too. The report flags the company's new premium E7 licence, a high-end software bundle that stacks its AI assistant M365 Copilot, a task-automation tool called Agent 365, and security features on top of its existing E5 package. More capability, higher price tag.
Should businesses be worried about runaway AI costs?
Yes, and many already are. KPMG, the global consultancy, first reported in July 2025 that almost a third of corporate leaders said they found it difficult to understand or control costs when rolling out AI across their business. The problem is the pricing model itself. Traditional software came with a fixed annual fee. AI tools often charge per "token," a small unit of text the AI processes, which means costs can spike without warning if employees use the tools heavily.
Forrester recommends that companies build new financial controls specifically for AI spending. Think of it like putting a meter on a tap. Techniques such as directing requests to cheaper AI models when a powerful one is not needed, and setting hard spending limits per team, can stop costs from spiralling.
One finding cuts against a popular narrative. Despite headlines about tech layoffs at Oracle, Microsoft, and Meta being linked to AI making workers redundant, the data does not support that story. IT staffing costs have not fallen. They made up 35 percent of technology budgets in 2025, and 67 percent of tech decision-makers plan to spend more on staff by 2027, not less. Forrester calls the framing of these cuts as AI-driven "AI washing of layoffs," meaning companies use AI as a convenient explanation for job cuts that are really about finances or restructuring.
Forrester's chief research officer Sharyn Leaver put it plainly: the organisations that do best by 2027 will not be the ones that spend the most on AI. They will be the ones that get their data, processes, and teams ready to use it well.
