Cloud deployments, in all parts of the telecom ecosystem, aren’t broken. Yet, in boardrooms from west to east, the decisions that underpinned them are being urgently revisited.
Why fix what still works? For many, cloud’s initial promise of infinite elasticity and cost-efficiency has hit a wall, and distribution requirements have outgrown the original promise.
In my conversations with C-suite leaders globally, there has been a sentiment shift. The "sugar rush" of public cloud adoption is fading.
The old buying criteria (i.e., FOMO, capacity and agility) are being replaced by a critical examination of unit economics and strategic control. The TCO has simply not stood the test of time, eroded by unpredictable data egress fees and the "Hotel California" reality of vendor lock-in: You can check in anytime you’d like but it’s prohibitively expensive to leave.
From renting infrastructure to owning the operating model
“Cloud is no longer just infrastructure. It is moving back to the center of the executive agenda, not as a utility, but as the primary operating foundation of the business. Today, it is the blueprint for how teams design, deliver and run the business.”
In truly cloud-native environments, the velocity of the design-to-market cycle is not a forced choice versus uptime. It is about uniform execution across fragmented geographies. Automation can no longer be an afterthought layered on top; it must be the metabolic rate of the organization.
In a word, this is about control.
I don’t mean control in the legacy IT governance sense. I mean control over how outcomes are produced. Companies are realizing they over-indexed on offloading technical decisions to hyperscalers in exchange for less friction. But as they seek to move faster, those "convenient" constraints have become architectural road bumps.
We are witnessing a trend of strategic repatriation. Operators are reclaiming the reins via insourcing and capability building. We recently saw a tier-one European operator invest in hiring and reskilling thousands of engineers to own their destiny rather than renting it. They realized that to differentiate, they couldn't just use the same tools in the same way as their competitors.
Taking ownership of central design doesn’t mean pushing partners aside. It means the nature of partnership changes. When you own your architecture, expertise in execution becomes your most valuable currency.
The new buying behavior: Co-creation, not consumption
Telecom leaders are moving away from simply layering new cloud tech on top of legacy spaghetti. They are making deliberate, structural changes.
In practice, this looks like radical consolidation. We are working with customers who are actively replacing two or three fragmented on-prem, private and public cloud solutions with a single, standardized global platform. They aren't doing this for remediation but to reduce the cognitive load on their operations teams.
Hybrid cloud is finally becoming intentional. Instead of an accidental mix of public and private resources, we are seeing environments designed with centralized orchestration. In some advanced models, the cloud software and management layers reside in the public cloud for agility, while the heavy lifting—the underlying infrastructure—remains owned and operated by the telco. This preserves margin and control where it matters most.
This shift has turned buying decisions into co-creation. Operators are no longer looking for vendors to hand them a black box. They want partners who will challenge their assumptions, accept their constraints and work through trade-offs.
“In recent engagements, solutions didn't start as predefined architectures. They emerged as a shared operating model that both sides were prepared to run.”
Data and AI: The sovereignty catalyst
If TCO was the spark for this recalibration, AI is the accelerant.
Operators aren't asking for AI features to simply "supercharge" existing clouds. They are terrified of the latency and data sovereignty implications of getting it wrong. They recognize that you cannot strip data considerations from cloud decisions.
It’s not a question of can AI run in the public cloud, but should it?
For massive datasets, "data gravity" is real. Moving petabytes of training data to the public cloud is often economically unfeasible and strategically risky. Operators are choosing to keep training and often inferencing in-house or at the edge.
Regulation, GDPR and sovereignty laws are drivers but this is really about the realization that telco data is a sovereign asset. Operators are refusing to feed their unique value into a generic public model. They want tighter control over the intelligence they generate.
The verdict: Cloud choices as a signal of intent
The cloud decisions operators make in the next 12 to 24 months will say more about their future than any 5G marketing campaign.
These decisions are a litmus test. They signal how much control an organization is prepared to take over its operating model, how confident it is in its ability to execute at scale, and whether it is content to be a utility provider or determined to be a technology powerhouse.
For telcos navigating this next phase, the question is no longer "Which cloud provider do I pick?"
The question is: "Is my current operating model fit for the future I want to build?"
Vivek Chadha is Rakuten Cloud SVP and global telco head. Mention him in the comments to start a conversation.
.svg)
