Until now, the digital economy has largely been built around centralized architectures. Bigger clouds. Bigger platforms. Bigger concentrations of compute and power.
Telecom operators helped build that world by connecting nations and enterprises to hyperscalers and cloud platforms — the most powerful privately owned infrastructure systems ever constructed. But that raises a difficult question: should a corporation’s most valuable data travel across infrastructure owned by another corporation? One ultimately controlled by someone like Jeff Bezos? (Hawks, spits).
And should governments, utilities and critical national systems become dependent on those same platforms?
There is no simple answer. But increasingly, evolving definitions of sovereignty are being viewed as the way forward. AI has changed what’s at stake for companies, governments and carriers alike.
Intelligence is no longer confined to the cloud core. It is now moving directly into the physical economy: healthcare, manufacturing, finance, logistics, energy and government infrastructure. The more intelligence becomes woven into those systems, the more uncomfortable centralized dependency begins to feel.
Telecom operators helped build the networks underpinning the AI economy. But what happens when the hyperscaler platforms running on top of those networks become too powerful — and too deeply embedded — to leave? That question sits at the heart of the growing sovereignty debate now unfolding across the global communications industry.
Defining sovereignty
Logan Wolfe, global head of sovereign cloud at Kyndryl, describes sovereignty as operating across three distinct but interrelated domains.
“We tend to look at sovereignty across three major domains — data, operations, and technology and supply chains,” Wolfe told FNTV. “Data underpins everything. Operations are the ability to continue independently of foreign undue influence. Technology and supply chains are about being able to use technology that is impervious, to a reasonable extent, to surveillance or disruption from unpalatable parties.”
Historically, the concept of data sovereignty was largely tied to national borders. Data was expected to remain inside secure, compliant infrastructure within a single country. If organizations expanded into another jurisdiction, they typically built another sovereign cloud environment.
But modern AI systems increasingly depend on the movement of data, workloads and inference across distributed infrastructure. At the same time, compute, networks, cloud platforms, semiconductors and energy systems are converging into a single strategic infrastructure layer spanning multiple countries and continents.
This creates an uncomfortable tension. The AI era demands global scale and distributed intelligence. Sovereignty, meanwhile, often pushes infrastructure toward localization, segmentation and tighter control.
Trucking through the tension
Shazia Zeb Sobani, VP of National Fiber Networks at TELUS, said the industry is now trying to reconcile those conflicting forces.
“Historically, when you think about network architectures, the networks were optimized around global efficiency,” Sobani explained. “They centralized operations and economies of scale. And sovereignty sometimes challenges all those three pillars, right? Because it basically means you'll probably move from centralized to distributed architecture if you really want data sovereignty.”
One obvious response would be to sever ties with hyperscalers altogether. That is fantasy.
The innovation gap is simply too large. No carrier can independently recreate modern AI infrastructure at hyperscale. Nvidia GPU ecosystems, hyperscaler AI tooling and cloud-native orchestration are now deeply embedded across the industry.
So the real question is no longer whether operators should depend on hyperscalers. It is how to balance dependency with vulnerability. Sobani believes the answer may lie in selective dependence. “Operators leverage hyperscaler infrastructure and AI tooling,” she said. “We are using hyperscaler infrastructure, but we have created our own AI tooling.”
That may become the compromise model of the AI era: carriers retaining control over essential strategic layers while consuming hyperscaler innovation where necessary. Technological purity may be impossible. But total dependency does not have to be the alternative.
In practice, this increasingly means sovereign systems connected through tightly controlled policy corridors — what some in the industry are beginning to describe as sovereignty tunnels. These are architectures where data remains sovereign while intelligence still moves across borders. Because enterprises increasingly want two things simultaneously: global AI innovation and local control.
Cashing in on trust
Underneath all of this sits a deeper concern.
For decades, carriers controlled the network. But in the AI era, true control increasingly belongs to whoever manages the intelligence layer above it. Carriers who merely own the pipes risk becoming the data plumbers of the digital age. Yet telecom operators still possess one enormous strategic advantage: trust.
Companies and governments expect critical systems to operate securely and continuously — and expect the information moving across those systems to remain confidential. Historically, that has been telecom territory.
The hyperscaler sector, meanwhile, faces a more complicated trust problem.
Virtually every major U.S. hyperscaler has faced lawsuits, investigations, settlements or regulatory penalties related to customer-data handling, privacy practices, surveillance, AI training or the monetization of user information. Amazon, Google, Meta, Microsoft and OpenAI have all faced varying degrees of scrutiny over how data gathered from users, enterprises or public sources has been collected, processed, transferred or repurposed.
The pattern is revealing. As hyperscalers have grown more powerful, data has increasingly ceased to be treated merely as customer information and instead become the raw material of a new industrial economy — one built on surveillance, prediction, advertising, automation and AI training at planetary scale.
Sobani said confidentiality remains central to carrier thinking. “The biggest thing was confidentiality of the data,” she explained. “How do we create a walled garden where our data stays with us? There are privacy requirements. We use sensitive customer data and we want to respect that for our customers.”
Once AI becomes embedded into critical infrastructure, questions about confidentiality and resilience stop being policy discussions. They become infrastructure requirements.
Cost and regulation
There is another reality here: sovereignty is expensive.
AI infrastructure demands enormous amounts of capital, power, cooling and specialized hardware. Rebuilding every layer locally is unrealistic for most carriers. Wolfe acknowledged the tradeoff directly.
“You can go open source. You can go to in-country providers,” he said. “But now the question becomes: are you sovereign, but losing out on innovation? Maybe some pieces absolutely remain sovereign — data is often one of them — but maybe other pieces don't need to. Maybe you mitigate the risks without going fully sovereign.”
That balancing act is now spreading across the industry. Not total dependence. Not total independence. Strategic optionality.
The broader implication is that AI infrastructure is rapidly becoming national infrastructure. Which means this debate is now bigger than telecom itself. Sovereignty increasingly becomes a question of governance, coordination and long-term planning.
As Blair Levin, architect of the U.S. National Broadband Plan, recently put it: “What networks do we really need to fully utilize AI? What do we do about universal service? What do we do about cybersecurity? What do we do to position the United States better for what is coming in terms of the AI economy?”
The next great infrastructure race will not be fought merely over roads, railways or broadband. It will be fought over intelligence itself.
Clear sovereignty strategies, combined with exceptionally high levels of trust, may become telecom carriers’ greatest strategic advantage in that contest.
For more on this shift — including full interviews with Shazia Zeb Sobani, Logan Wolfe and Blair Levin — watch Carrier 2.0 on FNTV, where we explore how AI, sovereignty, infrastructure and geopolitics are reshaping the future of communications and the wider digital economy.
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud, and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not necessarily represent the opinions of Fierce Network.