Spiralling cloud costs have become the second-largest and most unpredictable major expense for midsize IT firms, according to new research from Cloud Capital, with 89% of CFOs saying that the increase in cloud spend has squeezed gross margins over the last year.

Polling 100 senior finance leaders from growth-stage tech companies across the UK and US, the forecasting platform’s Cost of Compute study found that cloud spend as a percentage of revenue is increasing as AI requirements accelerate, with half (51%) of firms spending between 2% and 5% more than last year.

This means cloud spend has rapidly become one of the largest cost lines for midsize firms, but despite CFOs’ traditionally keeping a tight control over expenses at this scale, three-quarters (74%) report that cloud spend now varies between 5-10% or more each month.

At the root of the problem, unsurprisingly, is AI. The survey revealed that AI now consumes up to 22% of cloud budgets, with the speed of the technology creating ‘non-linear cost patterns’ that financial teams are struggling to build into forecasting models.

Firms heavily invested in AI are 2.6x more likely to see their margins slip than those with more moderate spend, leading the report to suggest that AI-heavy organisations are seeing compute costs begin to outpace revenue capture.

Attempting to rein in runaway costs and introduce structured controls, 72% of finance leaders have implemented guardrails for AI investment decisions. 

For example, when faced with an AI feature that promises to boost revenue by up to 5% but raises cloud costs by 4%, almost half (47%) of CFOs say they would choose to pilot first with a cost cap, while 27% would only approve deployment if they could cut costs elsewhere.

“The teams that stay ahead treat infrastructure and cost modelling as cross-functional responsibilities,” said Edward Barrow, Cloud Capital CEO.

“For CFOs, this means cloud and AI can’t be ‘someone else’s domain’ anymore. Predictability now depends on being at the table with engineering.”

According to the study, this collaborative attitude is key to cloud cost predictability. Cloud Capital found that when finance teams take an active role in cloud management, projects become far more predictable, with forecast accuracy nearly doubling compared to engineering‑led efforts.

Recommended reading

• Are Security Teams Losing Visibility in the Cloud?
• Only 8% of Organisations Encrypt The Majority of Their Cloud Data
• Comment | Navigating Today’s Cloud Security Challenges

The report found that 39% of joint finance-engineering teams reach highly predictable forecasts, but also deploy advanced optimisation strategies at nearly twice the rate of singularly owned projects, at 19% compared to just 10% for engineering-only teams. 

Cloud Capital argues that firms’ ultimate goal for the coming year should be to move cloud spend from an unpredictable, variable cost to a governable expense, one kept in check by diverging teams working in tandem.

That may prove a big advantage in an increasingly competitive landscape, with organisations that prioritise highly predictable forecasts improving gross margins at 2.8x the rate of their unpredictable peers, leading 44% of CFOs to cite improving forecast accuracy as one of their top priorities for 2026. 

“The findings are clear: who owns cloud costs determines whether Finance can forecast them, operational systems create the infrastructure for precision, and precision compounds into measurable margin advantage,” added Barrow.