Managing the costs of cloud computing

For more than a decade, cloud infrastructure sat comfortably in the domain of engineering. It was framed as a technical enabler, both elastic and scalable. Finance signed the invoices, engineering optimised when they could, and most companies accepted a degree of messiness as the price of speed.

 

We’re now seeing a decisive shift, and CFOs are stepping in. In Cloud Capital’s latest survey of 100 CFOs at US- and UK-based technology companies, cloud infrastructure now exceeds 10% of revenue and is the second-largest cost after headcount. At that scale, no CFO would tolerate loose governance, reactive controls or persistent forecast misses. Yet that is exactly how cloud has been managed for years.

 

 

Volatility is the forcing function

Our study also found that almost three-quarters of CFOs report monthly cloud forecast variances of at least 5%, with many seeing swings above 10%. By way of comparison, payroll typically stays within 2–3% of forecast and fixed costs rarely deviate by more than 2%. If payroll missed plan by 7% every month, boards would be asking hard questions. With cloud, this level of variance somehow became normalized. But that tolerance is evaporating fast and the financial consequence is becoming clear. Cloud has been the only major cost line where accountability was optional.

 

Eighty-nine percent of CFOs say rising cloud costs have put downward pressure on margins over the past year, making it a material financial risk.

 

AI has only accelerated this reckoning. AI and machine learning already account for 22% of total cloud spend in our data, introducing cost patterns that behave nothing like traditional SaaS infrastructure. Training workloads spike unpredictably, inference costs scale directly with usage and experimentation introduces intentional noise.

 

These non-linear dynamics break the forecasting assumptions finance teams have relied on for decades. The result is a widening control gap where CFOs can see the spend, but struggle to predict it. And as cloud’s share of revenue continues to grow (73% of CFOs expect it to increase again next year) variance on a growing base compounds into real margin risk.

 

 

Why CFOs are stepping in

Cloud Capital’s research shows that when Finance gets involved in cloud cost management, forecast predictability doubles. Teams with Finance involvement are twice as likely to achieve sub-5% monthly variance compared to engineering-owned models. Confidence in cloud COGS jumps by 50% and visibility improves materially.

 

The strongest outcomes emerge where ownership is shared. Joint Finance–Engineering teams consistently outperform every other model. They combine financial accountability with operational control. Finance sets the spending envelope and owns the variance story. Engineering operates within those constraints and pulls the technical levers. Together, they turn cloud from an unpredictable expense into a governable system.

 

This is a subtle but important distinction. Finance ownership does not mean Finance runs infrastructure. It does mean applying the same rigour to cloud that already exists for payroll, fixed costs and capital allocation, with clear budgets, explicit accountability and a regular forecasting cadence. CFOs are stepping in because cloud is now too large to sit outside the company’s normal operating system of control.

 

The organisations achieving this level of control share three traits.

 

First, visibility. Highly predictable teams can attribute cloud spend by product, team and customer in near real time. Without this, forecasting is guesswork.

 

Second, governance. Budget caps, approval workflows and policy enforcement are not bureaucracy; they are guardrails. Eighty-five percent of the most predictable organisations have fully implemented cloud governance, compared to barely half of the broader sample.

 

Third, cadence. Cloud does not behave like a quarterly expense. The most controlled teams forecast monthly, catching variance early before it compounds. None of the monthly forecasters in our data reported double-digit variance.

 

These systems reinforce each other. Visibility without governance creates insight with no control. Governance without cadence creates stale plans. Cadence without data creates noise. Together, they form the operating system of financial precision.

 

 

Precision matters

Organisations with highly predictable cloud forecasts improve gross margins nearly three times faster than those with persistent variance. This sort of precision enables proactive optimisation and more confident investment decisions. This is why we know the shift we’re seeing is not temporary. CFOs are stepping in because the economics demand it.

 

Cloud infrastructure has matured into one of the most powerful levers on the P&L. AI is amplifying that effect. The old model, where Finance reviewed spend after the fact and Engineering owned optimisation in isolation, cannot support an expense at this scale or volatility. The next competitive advantage will not be who negotiates the biggest discount. It will be who can make cloud measurable, forecastable, and accountable.

 

Now is the time to bring cloud infrastructure under tight financial governance, before variance quietly eats your margin.

 

---

 

Ed Barrow is CEO at Cloud Capital, leading the team to help turn cloud cost chaos into clarity for financial leaders

 

Main image courtesy of iStockPhoto.com and MF3d