Jasim Al Awadi highlights du Tech’s focus on sovereign cloud, GPU-as-a-service and agentic AI
du’s chief ICT officer sets out how the UAE can adopt AI at scale while keeping data, models and critical workloads firmly under national control
22 September, 2026
TT
16
For years, digital sovereignty was largely a question of geography: where a country’s data physically sat. As artificial intelligence moves into government services and critical infrastructure, that definition no longer stretches far enough. The harder questions now are who can access data, where it is processed, how it flows between systems, and what happens when it is used to train and run AI models.
That shift is central to du Tech’s strategy, and to the conversations that took place at GISEC Global 2026. The company has been building the infrastructure to let UAE organisations tap advanced computing and AI while keeping sensitive workloads inside the country, from its National Hypercloud and GPU-as-a-service to a hybrid model designed to balance innovation with security. Increasingly, it is also grappling with agentic AI, a new generation of systems that do not just answer questions but take actions, and the safeguards that must come with them.
Jasim Al Awadi, chief ICT officer at du, spoke to Gulf Business about moving the sovereignty debate beyond data residency, making expensive AI compute accessible without heavy upfront investment, and what it will take for governments to trust autonomous AI agents with real systems and citizen data.
The conversation around digital sovereignty has traditionally focused on where data is stored. How is du Tech trying to move that discussion towards how data, and increasingly AI, is actually used and controlled?
Digital sovereignty is as important an imperative as data residency. Knowing where data is stored remains crucial, but organisations also need visibility over who can access it, where it is processed, how it moves between systems and what happens when it is used to train or run AI models. These questions become even more important because sensitive information can move across applications, models and automated workflows.
For du Tech, our focus is on giving organisations control across the full lifecycle of their data and AI workloads. du Tech’s National Hypercloud is central to this approach. Built on Oracle Alloy and hosted within du Tech’s liquid-cooled data centres in the UAE, it provides a locally operated sovereign cloud environment with access to over 200 Oracle Cloud Infrastructure services, including embedded AI capabilities, advanced hardware and GPUs. This gives government entities the scale and agility to develop and deploy AI services while keeping sensitive data and workloads within the UAE and maintaining the security, governance and operational control their requirements demand.
AI infrastructure is becoming enormously expensive. How can models such as GPU-as-a-Service help UAE organisations access advanced computing power without having to make huge upfront investments?
Not every organisation needs to own and manage expensive GPU infrastructure to benefit from AI. GPU-as-a-Service changes the model by giving government entities and large enterprises access to high-performance computing capacity as needed, allowing them to scale with their workloads without committing significant capital upfront or taking on the complexity of managing the underlying infrastructure.
du Tech’s GPU-as-a-Service provides access to next-generation GPUs for high-intensity computing and AI workloads, including generative AI, large language models and deep learning. Scaling GPU capacity also lets organisations work with larger datasets and complex models, while faster processing can shorten model training and deployment cycles.
This capability is supported by du’s wider digital infrastructure, spanning AI data centres, fixed networks and an expansive 5G network, creating the connectivity and computing foundation needed to develop and scale AI workloads in the UAE. It gives organisations a more flexible and cost-efficient path to move AI from experimentation into production without having to build the entire compute environment themselves.
As government and critical infrastructure organisations adopt more AI, why is a hybrid approach becoming so important, particularly when it comes to balancing innovation, security and data sovereignty?
Different workloads have different requirements, so there is no single environment that fits everything. This is the principle behind du Tech’s National Hybrid AI, which gives government entities and enterprises the flexibility to develop AI applications in their cloud of choice and deploy them on du Tech’s sovereign infrastructure. Organisations can access advanced cloud and AI capabilities while ensuring sensitive data and critical workloads remain within environments that meet their security, governance and sovereignty requirements.
du Tech‘s National Hybrid AI brings together a sovereign-secure AI production environment with du Tech’s National Hypercloud, advanced 5G and fibre networks, hyperscaler partnerships and an ecosystem of system integrators. This integrated environment allows organisations to develop, deploy, manage and scale AI applications securely, while retaining greater control where applications or data require it. For government and critical infrastructure organisations, that balance becomes paramount as AI moves into core operations
Agentic AI raises a very different cybersecurity challenge because these systems can take actions rather than simply generate answers. What safeguards need to be in place before organisations can trust autonomous AI agents with real-world systems and citizen data?
Agentic AI changes the risk because an AI system is no longer only producing an answer; it may be able to access systems, use data and take actions on an organisation’s behalf. That makes clear boundaries an absolute priority. Every agent should have a defined identity, limited permissions and access only to the data and systems required for its role. Higher-risk actions should require human approval, especially where citizen data, financial transactions or critical systems are involved.
Organisations also need continuous monitoring, audit trails and the ability to trace what an agent did, what information it used and why an action was taken. There must also be clear mechanisms to intervene or stop an agent if it behaves unexpectedly. Human accountability remains critical – autonomy should never equal removing oversight.
At du Tech, we are working on the infrastructure needed to deploy Agentic AI securely. Our collaboration announced earlier this year in June with Open Innovation AI brings together du Tech’s National Hypercloud and Open Innovation AI’s AI infrastructure orchestration capabilities to explore secure, in-country deployment of Agentic AI workloads.
This includes GPU orchestration and resource management, as well as the development of sovereign AI capabilities that can enable organisations to deploy and scale autonomous workloads while keeping data, models and infrastructure within UAE jurisdiction.
The collaboration was orchestrated with the UAE Cyber Security Council, with security, regulatory compliance and data sovereignty embedded at the infrastructure level.
Read: CEO Fahad Al Hassawi on du’s H1 numbers, AI ambition and its next chapter

















