With the promise of the cloud turning 20 in 2026, many corporate leaders are facing a financial reckoning as costs spiral beyond initial projections. To navigate this complex landscape, we sat down with Dominic Jainy, an expert in enterprise IT strategy and cloud cost management. He offers a pragmatic perspective on why the cloud’s initial promise of being cheaper hasn’t always materialized and how executives can regain control. Our conversation explores the hidden costs driving budget overruns, the strategic value of on-premise infrastructure, the power of financial transparency in curbing waste, and the critical “speed premium” that keeps businesses invested in the cloud despite the expense.
Many organizations find their cloud infrastructure costs exceed initial budgets by 30% or more. What are the primary hidden or unpredictable costs that cause these overruns, and what financial models should leaders use for a more realistic long-term forecast?
The core challenge isn’t that cloud is inherently deceptive; it’s that its costs are not predictable. This is the fundamental reason it often fails to be “cheaper” in the way leaders expect. Unlike a capital expenditure for an on-premise data center, where you have a fixed cost, cloud spend is operational and can fluctuate wildly. The survey data is stark: 83% of CIOs are spending an average of 30% more than they planned. To get a real grip on this, you have to move beyond a simple one-year projection. I advise my teams to build out models comparing three distinct options: a full build-out in a new location, a pure hyperscaler environment, and a hybrid approach. When you project those costs over three, five, and even seven years, the picture becomes much clearer, and you often find that the pure cloud option remains surprisingly expensive over the long term.
You’ve found that for some workloads, a colocation facility can be more cost-effective than a public cloud. Could you walk us through the criteria you use to decide which applications stay on-prem, and what specific financial or performance metrics drive that choice?
Absolutely. There’s this tendency to think of cloud adoption as an all-or-nothing proposition, but that’s a trap. We maintain a hybrid environment precisely because, for many of our core compute workloads, it costs us less to run them in our own colocation facility. The decision isn’t based on a gut feeling; it’s a strategic calculation. The key factors we evaluate are the organization’s maturity, its specific regulatory and compliance needs, and, critically, the requirement to scale services up or down quickly. If a workload is stable and predictable, the financial argument for self-hosting becomes very compelling. The primary metric driving this is that long-term cost analysis I mentioned. When we map out the total cost of ownership over five or seven years, the numbers often show that running our own equipment in a colo, complete with a necessary disaster recovery site, is the more financially sound decision.
Improving cost transparency for department leaders has been shown to reduce waste. Can you describe the steps you took to make this data accessible and the key metrics you provide that have proven most effective in changing behavior and curbing over-provisioning?
This is where you can make a huge impact without repatriating a single workload. The problem is often not malice, but a lack of diligence because the costs are invisible to the people spinning up resources. The first step is to make that spend painfully transparent. We piped detailed cost and usage data directly to the CIOs in our various departments. We didn’t just show them a single number; we broke it down into its core components so they could see exactly where the money was going. We also deployed a technology business management platform to give us a unified view across the enterprise. The most effective metrics are the ones that show direct action. We track the number of decommissioned resources and, especially, the reduction in software licenses. When you can show a department head they’ve curtailed 6,000 or 7,000 licenses by actively managing their environment, the behavior changes almost overnight. They become partners in cost control.
Despite rising expenses, leaders often stay with the cloud for its “speed premium.” Could you share an example where the speed of cloud deployment delivered a strategic advantage that justified its higher cost, and how do you quantify this benefit when building a business case?
The “speed premium” is very real, and it’s the main reason we don’t just pull everything back on-premise. My gut tells me that if we tried to do everything in-house, we simply couldn’t afford to pursue the kinds of initiatives we are today. The strategic advantage is time-to-market. When a department needs to launch a new service, getting access to compute and storage in a cloud environment is just exponentially faster than the old way. The business case isn’t built on a dollar-for-dollar cost comparison but on opportunity cost. We quantify it by asking: “What is the cost of delay?” If it takes six months to procure, install, and provision hardware for a project on-premise, that’s six months of lost revenue or competitive advantage. That speed is a benefit that far outweighs the higher operational cost for certain strategic projects.
We’re seeing a “pendulum swing” with some workloads returning from the cloud, yet on-prem costs for licensing are also rising sharply. How do you navigate this complex trade-off, and what factors determine if repatriation is a financially viable long-term strategy?
You’ve hit on the central tension facing CIOs today. There isn’t a one-size-fits-all answer. We are absolutely seeing that pendulum shift back as organizations re-evaluate workloads that could be self-hosted more cheaply. However, at the same time, the on-prem world is getting more expensive. We still have a significant on-prem footprint—maybe 25% to 30%—and licensing costs there have increased tremendously, in some cases even outpacing the rise in cloud services. Navigating this requires a balanced, workload-by-workload analysis. The decision to repatriate must be a long-term strategic one, based on whether the application’s demands are predictable and if the organization has the in-house skills to manage it efficiently. It’s a constant balancing act between the operational expense of the cloud and the rising capital and licensing costs of maintaining a modern, secure on-premise environment.
What is your forecast for cloud cost management over the next three to five years?
I believe we’re moving past the “cloud-first” mantra and into an era of “cloud-appropriate.” The financial reckoning is forcing a level of sophistication and diligence that was missing in the early days of mass adoption. Over the next three to five years, I forecast that hybrid and multi-cloud strategies will become the default, not the exception. The focus will shift from migration to optimization. We’ll see a surge in the adoption of sophisticated FinOps tools and practices, embedding cost accountability directly into engineering teams. The pendulum won’t swing all the way back to on-premise, but it will settle in a more balanced middle ground where workloads are placed based on a rigorous analysis of cost, performance, and strategic value, rather than a blanket ideology. The wild, unpredictable spending of the last decade will give way to a much more deliberate, business-driven approach to cloud investment.