A few years ago, “move to the cloud” was the answer to almost every IT question. Speed, scalability, cost savings: the promise was compelling, and the race was on. Most organisations embraced a cloud-first mindset, migrating workloads as fast as they could. Fast forward to today, and many of those same organisations are staring at monthly cloud bills that feel completely out of control.
If that sounds familiar, you’re not alone. An estimated 21% of enterprise cloud spend (roughly $44.5 billion globally in 2025) is simply wasted on underutilised resources. In a recent survey, 31% of IT leaders admitted to wasting half of their cloud budget. Something clearly needs to change.
The Problem with Cloud First
Cloud-first made sense as a rallying cry. It pushed organisations past the inertia of legacy infrastructure and accelerated digital transformation. But as a long-term strategy, it has a fundamental flaw: it treats the cloud as the answer before you’ve properly defined the question.
Moving everything to the cloud regardless of workload type, cost profile, or data sensitivity creates fragmented, expensive environments. Security becomes harder to govern. Performance suffers when the wrong workloads end up in the wrong place. And the bills just keep climbing.
The backlash is real. A 2024 Barclays CIO Survey found that 86% of CIOs planned to move at least some workloads back out of public cloud, the highest rate ever recorded. That’s not a rejection of cloud; it’s a sign that the industry is finally getting smarter about how it’s used.
What Cloud Smart Actually Means
Being cloud-smart doesn’t mean being anti-cloud. It means making deliberate decisions about which workloads belong where, based on cost, performance, security, and business need.
Some workloads thrive in the public cloud: anything that needs to scale rapidly, reach global users, or spin up quickly. Others are better kept on-premises or in a private cloud, particularly those with predictable demand, sensitive data, or strict regulatory requirements.
P&G is a good example. After years of cloud-first migration, they now describe their strategy as “a living framework”, continually reassessing which hyperscaler is right for which workload, embedding financial governance, and using hybrid architectures where it makes sense. The goal, as their CTO puts it, is to keep technology aligned with business outcomes, not the other way around.
AI Has Made This More Urgent
If cloud costs were already a concern, the arrival of AI workloads has made strategic cloud planning essential. GPU compute is expensive. Training large models in the public cloud can be ruinously costly compared to owning the hardware outright. Meanwhile, sensitive customer data used in AI training raises serious governance and compliance questions.
Cisco, for instance, moved AI training and data-intensive inference workloads into private cloud environments, keeping only orchestration and non-sensitive services in the public cloud. The result: measurable double-digit reductions in cloud spend, better data governance, and improved model performance.
The lesson is that AI doesn’t just add another workload to your cloud estate. It forces you to think clearly about data gravity, privacy boundaries, and cost. In other words, it demands a cloud-smart approach.
A Practical Way Forward
If you’re wondering whether your cloud strategy needs a rethink, a few questions are worth asking:
- Do you know the cost profile of each major workload in your cloud environment?
- Are there workloads running in the public cloud that have predictable, steady demand and might be cheaper elsewhere?
- Is your data classification robust enough to support AI initiatives without exposing sensitive information?
- Do you have a FinOps practice in place to track, optimise, and govern cloud spend?
You don’t have to answer all of these perfectly to move forward. But asking them regularly is what separates cloud-smart organisations from those still flying blind.
The Cloud Is Still the Right Answer. Just Not for Everything.
We’re strong believers in the cloud at Idiro, and we’ve helped clients unlock real value from cloud platforms. But we’ve also seen the cost of an undifferentiated cloud strategy: bloated bills, governance headaches, and AI projects that stall because the data infrastructure wasn’t ready.
The shift from cloud-first to cloud-smart isn’t a retreat. It’s a sign of maturity. And in our experience, the organisations getting the most value from cloud are the ones asking harder questions about why, not just how.
If you’re looking at your cloud estate and wondering whether you’re getting what you should from it, we’d love to have that conversation.
