Companies should brace for larger software bills next year, according to Forrester, as vendors raise prices and attach usage charges to artificial intelligence features.
The technology research firm's 2027 budget planning guides, drawn from a global survey of more than 2,600 business and technology decision-makers, found 82% of technology leaders expect their budgets to rise.
More than 80% of leaders across all functions expect an increase, and up to a quarter think it will be 10% or more.
The mechanism behind the rise is as important as the number.
Vendors are increasingly charging by consumption rather than folding costs into a flat per-user licence, which shifts the expense of running AI infrastructure from the supplier's balance sheet onto the customer's.
Under a licence, the vendor carries the risk that a product proves expensive to run.
Under usage-based pricing, the customer carries it, and the bill arrives after the spending decision has already been made.
Forrester's own advice suggests bigger budgets will not automatically buy better results.
The firm warned that increasing investment without modernising operating models and improving data foundations will simply accelerate duplicated work and technical debt.
Its recommendations for 2027 include building machine-readable knowledge layers that AI agents can actually navigate, and targeted data clean-up rather than blanket spending.
Buyer scepticism is running alongside the optimism.
"They guarantee nothing, and promise nothing," one commenter wrote on The Register's forum, summarising the mood among purchasers. "It is all on you to double check the 'intelligence' outputs."
That captures the underlying trade-off.
Vendors are monetising heavy AI infrastructure by exposing customers to variable bills, while declining to underwrite whether the output is any good.
