When NVIDIA began championing sovereign AI on the global stage, it underscored a broader shift in the market: the next phase of AI will not be shaped only by models, platforms, and chips. It will also be shaped by who can provide the infrastructure, operating environment, and trust required to run AI at scale within national and regulatory requirements.

This is one reason telecommunications carriers are becoming increasingly important enablers in the sovereign AI landscape.

For many years, telcos were viewed primarily through the lens of connectivity. That view is now evolving. In many markets, carriers are also well positioned to support the infrastructure layer of sovereign AI because they bring together a set of assets that are difficult to replicate quickly: land, power, fibre, and trust.

Sovereign AI is not just about access to compute. It is also about where AI runs, how it is governed, and whether it can meet national, regulatory, and sector-specific requirements.

That changes the conversation.

The first requirement is physical infrastructure. AI-ready environments need suitable data centre sites, permits, and supporting facilities. Many national carriers have long operated strategic network assets and infrastructure footprints that can be expanded or adapted for new AI workloads.

The second is power. High-density GPU environments require significantly more energy and cooling than traditional enterprise workloads. As a result, power availability, energy strategy, and AI-ready data centre design are becoming central to competitiveness.

The third is connectivity. AI training, inferencing, and distributed workloads all depend on high-capacity, low-latency networks. Fibre remains foundational, and this is an area where carriers already play a central role in national digital infrastructure.

The fourth is trust. Sovereign AI is not only an infrastructure issue. It is also a matter of governance, resilience, compliance, and operational control. Public-sector and regulated-industry use cases require confidence that critical workloads can be supported in ways that align with national priorities and sector requirements.

Taken together, these factors help explain why telcos are gaining relevance in sovereign AI.

No. The stronger model is complementary, not substitutive.

Hyperscalers bring software platforms, developer ecosystems, and global scale. Telcos bring local infrastructure, fibre depth, power, and facility access, as well as sovereign operating relevance. The strongest sovereign AI outcomes are likely to come from combining these capabilities effectively.

This is what makes Southeast Asia especially interesting.

Across the region, governments and enterprises are taking a more deliberate approach to AI capability-building, with increasing emphasis on resilience, regulatory alignment, and long-term competitiveness. As a result, the discussion is moving beyond AI experimentation towards a more practical question: where will sovereign, enterprise-grade AI run, and under what operating model?

For organisations evaluating partners in this space, several capabilities stand out. These include the ability to support high-density GPU workloads at production scale, credible access to power and advanced cooling, regional reach combined with local compliance readiness, and an ecosystem that extends beyond infrastructure into orchestration, deployment, and enterprise enablement.

That is likely to become one of the defining questions in the next phase of AI adoption across ASEAN.

In this context, Singtel reflects broader capabilities that are becoming increasingly relevant in sovereign AI.

Across its digital infrastructure portfolio, the company has been building in areas that align with this shift, including AI-ready data centre development through Nxera, compute and cloud capabilities through RE:AI, connectivity through fibre and 5G, and orchestration capabilities through Paragon.

The broader point is less about any single initiative and more about what sovereign AI is likely to require overall: the ability to connect infrastructure, software, connectivity, and operations in ways that support enterprise and public-sector needs.

As AI moves further into national and enterprise deployments, the market will increasingly reward organisations that can operationalise it securely, efficiently, and at scale.

In that environment, telcos are not the only players that matter. But in many markets, they are likely to become an important part of the infrastructure foundation on which sovereign AI will run.

• Sovereign AI is about more than models, platforms, and chips.
• Telcos are becoming more relevant because they bring land, power, fibre, and trust.
• The strongest model is likely to be complementary, combining telco infrastructure with hyperscaler platforms.
• In Southeast Asia, sovereign AI readiness will increasingly depend on infrastructure, compliance, and operating model.
• The broader direction of travel points to integrated infrastructure becoming more important in enterprise and public-sector AI adoption.