Skip to content
UNTIL 31 DECWhich kind of AI user are you?
Take the two-minute check
AIAC AI ASSURANCE COUNCIL™

AI hype must not cloud the geography of sovereignty

A regional AI hub is not a sovereign asset for every country it serves. Where workloads run, and under whose law, still decides who is in control.

Vinod Rao

Co-founder, lemme Z, Hyderabad

· 5 min read

§ 1Article

Artificial intelligence is increasingly discussed in terms of model capability, computing power and investment. Yet one of the most important questions is geographical. Who controls the infrastructure that serves a region, and what does that mean for the countries relying on it?

The UAE’s Stargate project makes this question particularly relevant. Announced in May 2025, Stargate UAE is a planned one-gigawatt AI computing cluster within a larger five-gigawatt UAE-US AI Campus in Abu Dhabi. The partnership brings together G42, OpenAI, Oracle, NVIDIA, SoftBank and Cisco. The first 200 megawatts were expected to go live in 2026. These are substantial ambitions, but announced capacity should not be confused with capacity already operational.

The project’s significance extends beyond the UAE. OpenAI has stated that Stargate UAE has the potential to provide AI infrastructure and computing capacity within a 2,000-mile radius, which is about 3,200 kilometres, potentially reaching half the world’s population. This is a statement of intended reach, not proof that all countries within that radius will use the infrastructure. Nevertheless, it places the project in a distinctly regional context.

That is where the sovereignty question begins.

A regional AI hub is not automatically a regional sovereign asset

There is a compelling economic case for building large-scale AI infrastructure in the UAE. It can attract investment, develop local capabilities and give governments and businesses access to advanced AI without requiring them to build every component themselves. Regional proximity may also improve latency and service delivery.

But infrastructure serving several countries creates relationships more complicated than a map of data centres suggests. Where does a workload actually run? Which entity operates the infrastructure? Which laws govern the service? Who controls the models, software updates and access to computing capacity? What happens if commercial terms change, an export restriction is imposed, or a geopolitical dispute affects technology supply?

These are not arguments against Stargate UAE. They are questions that should accompany the enthusiasm surrounding it.

A country may benefit enormously from AI services delivered from a neighbouring state. But access to those services is not the same as sovereign control over them. Nor does a regional AI hub automatically become the sovereign AI infrastructure of every country within its reach.

The distinction matters because AI is becoming an input into government operations, financial services, healthcare, energy and other critical sectors. Dependence on an external service can become a strategic vulnerability when the service is difficult to replace or when the terms of access are determined elsewhere.

Recent events have made the physical dimension of this issue particularly clear. In September 2026, Reuters reported that Amazon Web Services could not restore access to resources and data held exclusively in a damaged availability zone in the UAE, while its Bahrain region had also suffered extensive damage during the conflict with Iran. The incidents demonstrate why geographical concentration, redundancy and recovery planning matter to the continuity of digital services. They do not establish that Stargate UAE itself has suffered similar damage, but they make the wider resilience question difficult to ignore.

G42 makes the geography more interesting

G42 is central to this discussion. It is a key partner in Stargate UAE, with responsibility for building the one-gigawatt cluster, while OpenAI and Oracle are to operate it. NVIDIA, Cisco and SoftBank also have defined roles in the wider collaboration.

Separately, G42 is working with India’s Centre for Development of Advanced Computing (C-DAC) and Cerebras on a national-scale AI supercomputer to be hosted in India. The Condor Galaxy India initiative involves an eight-exaflop cluster comprising 64 Cerebras CS-3 systems. G42 and C-DAC are involved in its installation, deployment, operations and maintenance.

These are distinct initiatives, not evidence that India is relying on Stargate UAE. Their juxtaposition nevertheless highlights an important regional reality. The same technology groups can participate in infrastructure projects across national boundaries, while each country seeks to build its own AI capabilities.

This is not inherently contradictory. International partnerships can accelerate domestic capacity, and no country can realistically develop every component of the AI stack independently. But the location of a computing facility is only one part of the sovereignty equation. Hardware, model development, software, technical expertise and operational responsibilities may involve organisations from several jurisdictions.

India’s ability to govern workloads on infrastructure hosted within India may differ from another country’s ability to govern workloads served from a regional hub in Abu Dhabi. The relevant question is not simply who owns the servers. It is how much practical control each country retains over the capabilities it needs.

Geography must remain part of the debate

The language of AI leadership can make these distinctions easy to overlook. Gigawatts of computing capacity, frontier models and ambitious deployment timelines dominate announcements. But the political geography of AI does not disappear because the technology is presented as borderless.

For governments considering regional AI infrastructure, five questions deserve clear answers:

  • Where will their workloads run?
  • Which jurisdiction governs the service?
  • How dependent will they be on external operators and technology suppliers?
  • Can critical services continue during a disruption?
  • Can they migrate elsewhere without unacceptable cost or loss of capability?

These questions are especially important when AI becomes embedded in public services and critical national systems. They should be addressed before dependency becomes difficult to reverse, rather than after a dispute or disruption exposes it.

The answer is not technological isolation. Cross-border cooperation can create substantial value, and regional infrastructure may be more practical than attempting complete self-sufficiency. The objective should be to understand dependencies, negotiate meaningful safeguards and preserve credible alternatives.

Stargate UAE could strengthen the UAE’s position as a regional AI hub while expanding access to advanced computing across neighbouring markets. G42’s separate work in India illustrates how international partnerships can also support domestic AI infrastructure.

Neither development settles the question of sovereignty. They make it more urgent.

AI may be borderless in its ambition. The infrastructure, laws and political relationships that make it work are not.

Sources

About the author

Vinod Rao

Vinod Rao

Co-founder, lemme Z, Hyderabad

Vinod Rao is Co-founder of lemme Z, where he leads AI strategy and partnerships for hearable devices. He previously headed AI sales at ParadigmIT, working on AI and digital governance solutions for public sector clients. He speaks and writes on responsible AI across infrastructure, platforms and applications.

LinkedIn

Published under the Council's contributor guidelines and article format. A byline is not an endorsement, and confers no advantage in any examination or certification decision.

§ 2Related guides

Continue reading