Sovereign artificial intelligence (AI) is being reframed at AI Everything Abu Dhabi, not as isolation, but as a foundation for global collaboration. During the event, Talal Al Kaissi, CEO of Core42, explored how countries can balance control over critical data and AI infrastructure while participating in a globally connected digital economy.
The discussion comes as governments worldwide accelerate investments in AI infrastructure, national AI strategies and regulatory frameworks amid growing concerns over data sovereignty, geopolitical risk and access to compute resources.
Serfaty argued that sovereign AI is often misunderstood as a purely domestic undertaking focused on building and controlling technology within national borders. “In actuality, it’s not about that, and it’s not going to be about that, nor should it be about that,” said Al Kaissi. “Governments and policymakers are wrestling with how to build and infuse the data and all of the things that they do well across borders from an AI sense.”
For Al Kaissi, sovereignty ultimately comes down to control and the ability of countries to determine their own risk appetite when deploying AI systems and managing sensitive data.
“Sovereignty does have multiple definitions, and each country tries to define it with their own imperatives in mind,” he said. “Most of what you need to do when you take a step back and think about what it means to be sovereign is an element of control.”
Al Kaissi argued that the UAE’s progress in sovereign cloud and AI infrastructure has been enabled by what he described as a regulatory environment designed to encourage innovation while maintaining oversight of critical workloads.
“All of the success that we’ve had at either G42 or Core42 has been the product of visionary leadership that created a regulatory permissive environment and this open lab we call the UAE,” he said.
According to him, sovereign AI requirements no longer apply only to governments and highly regulated sectors. As AI becomes embedded across industries, enterprises are increasingly seeking greater control over how their data is processed, stored and governed.
“Predominantly, the customer persona we serve is the public sector and regulated industries,” he said. “But increasingly, the same sovereign imperatives are going to be important even for large enterprises that are not regulated.”
A key challenge for governments, said Al Kaissi, is determining where sovereignty requirements begin and end, particularly when many cloud platforms and AI technologies originate from a small number of global providers.
“What you’re solving for predominantly is extraterritorial jurisdiction that can be applied by foreign governments over companies that come from those countries,” he added.
Rather than rejecting global cloud providers, Core42 has focused on creating technical and governance models that allow organisations to use hyperscale infrastructure while retaining sovereignty controls. Al Kaissi pointed to the company’s partnership with Microsoft as an example.
“What we call a sovereign public cloud enables sovereignty over a public cloud backbone,” he said. “We did that with Microsoft by building a sovereign control plane on top of it that gives you the technical and policy controls.”
Levels of sovereignty
The model provides different levels of sovereignty depending on workload requirements, ranging from fully isolated environments to sovereign public cloud deployments.
“You have everything from top secret all the way to open data,” said Al Kaissi. “You need to have environments that are suitable for each one of those types of data sets.”
The discussion also highlighted the growing importance of what he described as “strategic autonomy” and “strategic interdependence” as countries seek to balance national control with international cooperation.
“I don’t think there’s going to be a situation globally where there’s one country that can do everything on their own,” said Al Kaissi. “Strategic interdependence is going to be a continuum of that.”
One emerging concept discussed was the idea of digital embassies, where countries can securely host and replicate critical data outside their borders while maintaining sovereignty protections through technical, legal and policy mechanisms. He said such arrangements could help governments improve resilience while preserving control over sensitive information.
“Through a digital embassy construct, you can still preserve sovereign rights in a very similar and analogous way to a traditional embassy,” said Al Kaissi.
Strategic asset
The scale of investment reflects the growing recognition that compute infrastructure is becoming a strategic national asset. Al Kaissi noted that building AI infrastructure at gigawatt scale requires investments that can reach tens of billions of dollars, putting such projects beyond the reach of many countries.
“Not everyone can put the capital expenditure and economic resources required to build something that, on a gigawatt level, costs around 50bn USD,” he said.
As a result, Al Kaissi expects countries to increasingly rely on trusted international partnerships to access AI infrastructure and compute resources rather than attempting to build everything domestically.
For nations beginning their sovereign AI journeys, he recommended benchmarking successful initiatives and learning from countries that have already developed AI-native strategies.
“We benchmarked other countries when we started thinking about some of what we’ve done,” said Al Kaissi. “Now those same countries are benchmarking us.”
Looking ahead, he believes access to compute will remain one of the defining factors shaping national AI competitiveness. “Compute is going to be key, at least access to compute,” said Al Kaissi. “Whether it’s in the country or not, having that strategic access matters.”
