Forrester says AI costs are set to show up in software bills

Forrester is warning enterprise buyers to expect higher software bills next year as software and AI vendors raise prices and add usage charges to cover the cost of running AI services. The research firm based the forecast on a survey of more than 2,600 business and technology decision-makers, and said 80% expect data and software spending to increase.

The pressure point is familiar to anyone buying AI tooling at scale: vendors are moving away from predictable flat subscriptions for some products and toward usage pricing tied to consumption. Forrester cited recent moves by Anthropic, OpenAI and GitHub over the past six months as examples that have already created concern among customers. It also pointed to Microsoft’s new premium E7 license, which combines M365 Copilot, Agent 365 and security tools on top of E5.

Forrester did not quantify the expected price increases, and the survey does not say how much of the budget growth is tied to AI versus broader software inflation. The direction, however, is clear: AI infrastructure is expensive, and vendors are looking for ways to push more of that cost into customer contracts.

Bain & Company estimated last year that AI datacenter build costs could reach $2 trillion by 2030. That number has become part of the backdrop for enterprise software pricing. Model access, inference, storage, security controls and agentic features are now being packaged into suites and metered products that can expand bills as adoption rises.

Usage pricing moves the risk to buyers

For buyers, the change is less about whether AI tools are useful and more about whether costs can be forecast. Token-based and usage-driven products can make software spending harder to model than seat-based SaaS. Forrester said organizations should update FinOps practices for AI workloads, including runtime cost controls such as model routing, semantic caching and usage guardrails.

That recommendation lines up with a July finding from KPMG, which said nearly a third of corporate leaders reported difficulty understanding and controlling operating costs when rolling out business AI at scale. KPMG said many organizations are still building the ability to forecast, monitor and manage AI spend as usage-based pricing becomes more common.

Forrester’s chief research officer Sharyn Leaver said the organizations likely to perform best in 2027 will be those investing in “trusted data, strong governance, organizational readiness, and the ability to continuously adapt as technology and customer behavior evolve,” rather than those that spend the most on AI.

AI is not yet cutting IT staffing budgets

Forrester also pushed back on the idea that AI is already reducing personnel costs across IT. The firm said staffing accounted for 35% of IT budgets in 2025 and that IT staffing spend has not declined in recent years, despite layoff announcements at companies including Oracle, Microsoft and Meta.

Looking to 2027, 67% of technology decision-makers told Forrester they expect staffing budgets to rise, 23% expect them to remain flat and 10% expect a decline. Forrester said data and analytics roles are expected to grow, with 68% of data technology decision-makers expecting that budget line to increase.

The report’s practical message for technology leaders is that AI spending is unlikely to be offset automatically by headcount reductions. Buyers face a dual budget problem: rising software and infrastructure-linked charges on one side, and continued demand for people who can manage data, governance, security and operations on the other.

This story draws on original reporting from The Register.