CFOs tighten control as cloud costs hit profits

CFOs tighten control as cloud costs hit profits

Cloud costs now average 10% of revenue. Start-ups and SaaS companies are putting finance teams in charge as spending volatility turns cloud into a major profit risk. New research by Cloud Capital shows 89% of CFOs report margin erosion and 97% have formalised cloud governance.


Cloud infrastructure has become the second-largest expense after salaries for many start-ups, according to new research from Cloud Capital, with CFOs increasingly assuming control over its governance and financial planning.

The study — The Cost of Compute: What 100 CFOs Reveal About Cloud Infrastructure’s Impact on the P&L — surveyed financial leaders from 100 SaaS and technology businesses of up to 1,000 employees across the US and UK. It found cloud spend now averages 10% of revenue, rising as high as 20% for nearly a quarter of respondents.

The research highlights a structural shift over the past 18 months, with cloud cost management migrating from engineering to finance. Among companies where finance teams take ownership, forecast predictability doubles, and cost-of-goods confidence improves by half.

Edward Barrow, CEO and Co-Founder at Cloud Capital, said: “CFOs report month-to-month variability of 5–10% as standard. Right now, cloud’s unpredictability is disproportionate to its size and completely out of line with what CFOs expect from any other major expense. That’s the financial tension driving this shift toward tighter governance and Finance ownership.”

Cloud Capital’s analysis shows that AI and machine learning workloads already account for 22% of total cloud spend — a figure expected to climb rapidly in 2026. Despite the cost pressures, 72% of CFOs say they would accept short-term margin compression in exchange for AI-driven product gains, signalling a deliberate trade-off between profitability and growth.

Casey Woo, Co-Founder and CEO at Operators Guild, commented: “Cloud has moved into the top tier of operating costs. AI workloads already account for nearly a quarter of that spend. Forecast variance is hitting ranges that would be unthinkable for any other major cost centre. And margin performance tracks directly with how well teams can see, model, and govern this spend.”

The findings also reveal that 97% of surveyed CFOs have formalised cloud governance policies, with 62% fully implemented. However, only a quarter describe their cloud cost forecasts as “highly predictable,” and 92% of start-ups lack a dedicated FinOps function.

Barrow added: “Cloud infrastructure is now central to business performance, but as costs rise so does the pressure on finance teams to predict, justify, and optimise spend. The findings underline a maturity gap between cloud adoption and financial control.”

The report concludes that cloud’s financial impact will continue to grow as AI usage expands. For many fast-growth technology companies, the next phase will involve establishing data-driven, agile financial frameworks to balance innovation with cost stability.



  • Lloyds appoints new London ambassador

    Lloyds appoints new London ambassador

    Lloyds Banking Group appoints Ajneet Jassey as its new London ambassador. The bank has named the senior legal executive to focus on housing delivery, business growth, and technology engagement across the capital.


  • AI application from Anthropic reaches EU data reserves

    AI application from Anthropic reaches EU data reserves

    Anthropic’s legal AI debut rattles European data and software stocks. Shares in major European data, publishing, and legal software companies fell sharply after Anthropic launched a legal productivity tool for in-house counsel, raising fresh concerns about AI-driven disruption to high-margin professional information businesses.


  • Beyond the AI bubble — from hype to lasting impact

    Beyond the AI bubble — from hype to lasting impact

    AI’s business promise is cooling as leaders confront implementation reality. Cian Clarke, Head of AI at Nearform, argues organisations must move beyond hype, rethink model strategy, and build durable systems grounded in real data and operational discipline.