Experts warn software budgets could be set to soar as AI bills are on the rise

Experts warn software budgets could be set to soar as AI bills are on the rise

AI software vendors are increasingly charging per consumption rather than a flat per-seat rate, making expenses more unpredictable.

Software budgets are rising, analysts warn. Forrester analysts warn around four in five leaders and ITDMs envision having larger budgets in 2027. Consumption-based AI pricing is making it harder to predict outlay. Targeted investment to improve data quality is key.

Forrester is predicting software budgets could be set to rise, with more than four in five leaders expecting to increase overall budgets over the next 12 months and 82% of tech decision-makers expecting larger budgets. While some of the extra cash could come as a result of increased confidence and readiness to spend on tech, the company's analysts warn that a shift in pricing structures could also be forcing companies to fork out more.

This comes as software vendors shift from traditional per-seat licences to token or credit pricing, which introduces so many more variables including model selection, context size, output length and agent operating time, leading to far more unpredictable outgoings.

The real reason businesses are preparing to spend more on software

"Business leaders are no longer planning for a return to stability - they’re planning for a future where volatility is a constant," Chief Research Officer Sharyn Leaver noted.

Recent shifts from major AI providers all point toward this emerging pricing model becoming the norm, with GitHub moving its Copilot plans to usage-based billing in June and OpenAI adding pay-as-you-go Codex seats in April. Anthropic also recently removed Claude from its standard subscriptions and seat-based models over difficult-to-predict demand, but set out plans to reintroduce it where capacity permits.

Acknowledging these major shifts, Forrester's report reveals two areas where companies can increase their budgets for 2027:

• Building machine-readable context and enterprise knowledge
• Increasing brand visibility in answer engines

The report also hints at the major role AI can play in marketing and customer-facing experiences, and the potential use that synthetic data can provide subject to testing.

All in all, it's more about targeting investments rather than throwing cash at the problem, as Leaver concludes:

"The organizations that outperform in 2027 won’t be those that spend the most on AI. They’ll be the ones that invest in the foundations that make AI effective."