Cloud Waste Is Rising Again, and AI Is the Reason

After several years of steady improvement, enterprises are losing ground on cloud cost again, and artificial intelligence is the reason. Flexera’s 2026 State of the Cloud research put wasted cloud spend at around 29 percent, up from the previous year and the first increase after roughly five years of decline, with analysts attributing the reversal largely to AI workloads (reporting). For CIOs and CFOs who had finally started to feel in control of cloud spending, AI has reopened a problem they thought was closing. The discipline that fixes it, FinOps, matters more now than it has in years.

Why AI reversed the trend

For most of the past five years, organisations slowly got better at cloud cost. They tagged resources, switched off what they were not using, negotiated committed pricing, and built the habits that bring variable spending under control. Waste fell year on year.

AI has undone a chunk of that progress in a short time, for three connected reasons. First, AI workloads are simply expensive, consuming premium compute at scale. Second, they are provisioned fast, often by teams racing to ship AI capability, faster than cost controls can keep up with. Third, and most importantly, AI spend is unusually hard to tie to a specific outcome. An organisation can see clearly that AI is costing a great deal and still struggle to say what value each dollar is actually producing.

That last point is the real problem. Cost without a clear link to value is where waste hides. When you cannot connect spend to an outcome, you cannot tell efficient AI investment from expensive experimentation that is going nowhere, and the bill climbs regardless.

FinOps is no longer just about cutting costs

The reassuring news is that the discipline to manage this already exists and is maturing fast. FinOps, the practice of bringing financial accountability to cloud spending, has gone from niche to near universal, with the FinOps Foundation reporting that almost all organisations now actively manage AI spend, up sharply from a small minority two years ago (reporting). What has changed is the emphasis. In 2026, FinOps is less about simply spending less and more about spending accountably, tying cost to value so the organisation invests confidently rather than cutting blindly.

That shift matters because the goal with AI is not to minimise spend. It is to ensure the spend is producing value, and to stop the spend that is not. A pure cost cutting mindset risks starving the AI initiatives that are working. A FinOps mindset funds those and shuts down the ones that are quietly burning money without return.

The question is not why is our cloud bill so high. It is which of this spend is producing value, and which is just producing a bill. Most organisations cannot yet answer that for their AI.

Visibility is the foundation

Everything in FinOps starts with visibility, and visibility is precisely where most organisations are weakest on AI. Flexera’s research found that a large share of organisations lack a clean way to tie AI spend to a specific product or outcome. You cannot govern what you cannot see, and you cannot judge value you cannot attribute.

Getting this right is not glamorous, but it is decisive. It means tagging and allocating every significant cloud and AI cost to an owning team, and connecting that cost to the outcome it is meant to produce. Once spend is visible and owned, the rest of FinOps follows naturally: teams right size their resources, switch off what is idle, use committed and reserved pricing for predictable workloads, and review spend against value on a regular cadence rather than reacting to an alarming invoice after the fact.

What technology and finance leaders should do

1. Make AI and cloud spend visible and owned. Tag and allocate every significant cost to a team, and tie it to the outcome it supports. This is the single highest value step.
2. Judge spend by value, not just size. Identify which AI and cloud spend is producing measurable value and which is not, and act on both.
3. Apply the basics relentlessly. Right size, switch off the unused, and use committed pricing for predictable workloads. These still account for most recoverable waste.
4. Govern continuously. Treat FinOps as an ongoing operating discipline across engineering, finance and the business, not a quarterly cost cutting exercise.
5. Bring AI into the same discipline as the rest of cloud, rather than letting it sit outside normal cost governance because it is new and exciting.

Done this way, AI stops being the thing that blew up your cloud bill and becomes a managed, accountable investment like any other.

Where Jaykan fits

Jaykan helps enterprises and government organisations bring cloud and AI spend under control through FinOps discipline, as part of our cloud practice: establishing visibility and ownership, tying spend to value, and optimising continuously rather than reacting to the bill. We help clients fund the AI and cloud investment that works and stop the spend that does not.

If your cloud or AI costs are climbing faster than the value you can point to, a complimentary, senior led assessment will show you where the waste is and give you a practical plan to bring spend back under control.

Frequently asked questions

Why is cloud waste rising again?

After roughly five years of organisations slowly reducing wasted cloud spend, the trend has reversed, with wasted spend climbing again. Analysts attribute the reversal mainly to AI workloads, which are expensive, consumed unpredictably, and often provisioned faster than cost controls can keep up with.

What is FinOps?

FinOps is the practice of bringing financial accountability to variable cloud spending. It gives engineering, finance and business teams a shared, near real time view of what cloud is costing and why, with clear ownership of that spend, so cost decisions are made deliberately rather than discovered after the bill arrives.

Why is AI spend so hard to control?

Because it is variable and decentralised, teams can provision expensive AI resources quickly and easily, and because it is often difficult to tie a given amount of AI spend to a specific product or business outcome. Many organisations can see that AI is costing a lot but cannot clearly say what value each dollar is producing.

How do we get cloud and AI costs under control?

Start with visibility: tag and allocate every significant cost to an owning team, and link spend to outcomes. Then right size resources, switch off what is unused, use committed pricing for predictable workloads, and review spend against value regularly. The aim is accountability and value, not simply spending less.