Operations | Monitoring | ITSM | DevOps | Cloud

2025 is winding down, but cloud and AI spending is still heating up. September brought some big news that could impact your bottom line. Here are the can’t-miss cloud cost updates from last month: Enjoy!

Many SaaS businesses prioritize customer acquisition and retention over increasing gross margins, particularly during the startup and scale-up stages. It makes sense. A company can accelerate its revenue growth by acquiring and retaining more new customers, rather than simply selling more to existing ones. Yet, here’s the thing. Revenue growth measures the increase in the amount of money a business earns from sales.

As the Senior Manager of Finance & Accounting here at CloudZero, and with a career in FP&A that includes tenures at large public companies, I’ve spent a significant amount of time observing the interactions between the folks who plan the company’s budget and those who spend it. While engineering and operations teams focus on the execution side, my team ensures that the company has the resources required to make each endeavor a success.

Cloud cost intelligence has moved beyond simple cost-cutting. Now, it’s about creating value. Cloud bills continue to rise, and workloads are becoming increasingly complex. Teams also need to understand what they’re spending, why, and how that spend ties to business results. FinOps has become the framework for bringing finance and engineering together. It’s helping teams manage costs, improve margins, and plan with confidence. But challenges remain.

My wife and I spent hours this summer at home digging in the dirt. We planted new shrubs and perennials and created a small vegetable garden. We spread many square yards of fresh topsoil and grass seed over areas of lawn that needed rejuvenation. It turns out, I should have done all that landscaping with a FinOps leader’s mindset — before my water bill tripled when I wasn’t looking.

A SaaS team once spotted their first breach not in a SIEM dashboard, but in their AWS bill. Their compute costs spiked by 400% overnight. Turns out, an attacker had spun up dozens of high-powered instances for crypto mining. Logs eventually confirmed the intrusion, but the cost anomaly was the first signal that something was wrong. This incident isn’t unusual. Cloud costs often reflect consumption, but they can also reflect compromise.

Cloud costs are no longer just an engineering problem or a finance problem. It’s now an everyone problem. That was the central message from the FinOps Training At Scale webinar that took place on Sept. 25, 2025, where CloudZero’s Larry Advey (a.k.a. “Fred FinOps”) and Director of Tech Enablement Umesh Rao walked through the realities of building FinOps practices that work in the real world.

The world of cloud costs is always evolving, and AI spend is quickly becoming one of the most unpredictable and confusing cost drivers. As more organizations integrate generative AI into their products, FinOps teams are struggling to account for — and control — these new, often mind-boggling cost streams. In fact, 44% of engineering professionals say improving AI explainability is a top priority in AI budgeting, according to CloudZero’s State Of AI Costs In 2025 report.

Cloud costs don’t wait for your finance team to catch up. They spike on product launches, dip when usage slows, and sometimes blow past forecasts overnight. Every container spun up, every gigabyte stored, and every terabyte transferred adds to the tab. The main culprit is often variable costs. In this guide, we’ll break down how variable costs affect budgeting and the strategies you can use to turn cost variability into a competitive advantage.

AI adoption is reshaping how organizations innovate. It’s also driving cloud costs higher. CloudZero’s State Of AI Costs In 2025 report finds that for mature FinOps and engineering leaders, visibility into AI costs is a critical first step, but it’s not enough. To enable fast, responsible AI and machine learning innovation at scale, teams need pragmatic, flexible guardrails. They don’t need rigid budgets or knee-jerk shutdowns that slow progress or push teams into shadow ML.