A quiet shift is underway across the Gulf. What began as ambitious national AI strategies has evolved into full-scale execution with billions in capital, hyperscale datacenters, and a race to build the infrastructure that will define the region's economic future. By 2030, AI is expected to inject USD 320 billion into the Middle East economy, with Saudi Arabia and the UAE alone responsible for more than USD 200 billion of that figure.

By Sami Alfaraj - MEA Head of Technology, Submer

[For more news, click here]

The region's datacenter footprint is expanding just as fast, projected to exceed USD 9 billion as hyperscale facilities rise across Saudi Arabia, the UAE, Qatar, Oman, and Bahrain.

But scale alone won't solve tomorrow's AI challenges. As demand accelerates, the limits of centralized cloud computing are becoming clear and the region's next infrastructure frontier is emerging at the edge, where central offices and base stations are being reimagined as compute hubs.

Significance of edge AI infrastructure

Processing data closer to the source dramatically reduces round-trip latency, eliminates backhaul congestion, and keeps sensitive data inside the jurisdiction where it is generated. While that advantage is valuable in almost any digital environment, for some applications, low latency is not simply beneficial; it is essential.

Cloud gaming is one example of why edge infrastructure matters. Because gameplay is streamed from remote GPU servers, even milliseconds of delay can impact the user experience, making low-latency GPU compute essential. But meeting that latency budget means putting dense GPU clusters in places that were never designed to host them — and that is where the conversation shifts from network design to physical infrastructure design.

But the importance of edge AI extends far beyond gaming. As AI adoption accelerates, demand for localized GPU infrastructure is growing across industries, from smart cities and industrial automation to healthcare and financial services, where real-time inference is becoming increasingly critical. For the GCC, where mobile penetration exceeds 100% in many markets, 5G adoption ranks among the highest globally, and the gaming market alone is projected to reach USD 9 billion by 2031, localized GPU compute is rapidly becoming a strategic foundation for the region’s digital economy.

Edge AI across industries

Distributed edge AI infrastructure allows organizations to process and analyze data closer to where it is generated, enabling faster insights and responses. Edge AI is already reshaping industries across the region.

• Automotive systems using AI for real-time navigation and driver monitoring • Industrial IoT enabling predictive maintenance and production quality control • Healthcare wearables providing immediate analysis of patient data • Smart cities using connected sensors to manage traffic and security • Retail and agriculture applications driven by real-time data from sensors and devices

Each of these workloads carries a thermal and energy footprint that scales with adoption. In a region where ambient temperatures already stress conventional cooling for most of the year, the decision of how to cool edge compute is not a back-office detail — it is a strategic choice that determines whether deployments are economically and environmentally sustainable at scale.

Core-to-edge AI architecture

When it comes to AI infrastructure, the most effective model is emerging as a core-to-edge architecture, with large cloud datacenters supported by networks of edge compute nodes.

But this architecture is only as effective as the physical infrastructure underneath it. Modern AI racks now routinely exceed 50 kW, with next-generation GPU platforms pushing past 130 kW per rack — densities that traditional air-cooled facilities, and most existing telco real estate, were never engineered to support. Getting core-to-edge right is therefore as much a thermal, mechanical, and site-design challenge as it is a compute or networking one.

These edge nodes provide a huge opportunity for telecom companies to expand their offering, moving beyond connecting users to cloud compute and instead providing the compute itself. Large-scale AI datacenters provide centralized compute power, while distributed edge nodes deliver real-time inference and localized processing. This shift creates a major opportunity for telecom operators.

A new opportunity for telecom operators

5G adoption across the GCC is among the fastest in the world, with the UAE and Saudi Arabia consistently ranked in the global top five for mobile network performance. Telcos can implement edge compute nodes throughout their networks, delivering localised GPU-as-a-Service solutions that provide low-latency data processing to local users. That GPU bandwidth can be utilized and monetized as required, whether that be for cloud gaming subscriptions or AI inferencing workloads. But that localized GPU compute also opens the door to a very important strategic opportunity – AI sovereignty.

Countries and territories all over the globe are beginning to worry about their reliance on foreign entities to provide the AI and cloud infrastructure that their citizens need. Ensuring that the full AI stack that a country or territory relies on is wholly owned and operated within that territory is key to AI resilience. AI sovereignty is no longer only about data residency.

It is increasingly about securing long-term control over the infrastructure, compute capacity, and AI services that will underpin national economies and critical industries. For the GCC, building localized AI infrastructure is becoming both a technological and strategic priority. But that’s only half of the issue. Data regulations like the UAE's Personal Data Protection Law (PDPL) and Saudi Arabia's PDPL insist that personal data is stored and processed locally, with tight controls on access. Middle Eastern countries that build out full-stack AI solutions, however, can ensure AI resilience, while also avoiding any conflicts of data regulation that could arise from engaging with foreign cloud providers.

Telecom companies can deploy edge compute nodes at scale, providing a credible solution to the sovereign AI challenge, while also laying the foundation for valuable new revenue models.

Designing the edge: from advisory to deployment

The gap between recognising the edge opportunity and operationalising it is wider than most operators expect. Retrofitting a central office or a base-station shelter for 50–130 kW GPU racks

is not a procurement exercise; it is an engineering exercise that touches power distribution, cooling topology, structural loading, water and heat-rejection strategy, and operational safety. Getting any one of these wrong compromises performance, sustainability targets, and the economics of the entire deployment.

This is where specialist advisory and design partnership becomes decisive. At Submer we work alongside operators and developers from the earliest stages — site assessment, thermal modelling, density planning, liquid-cooling architecture, and total-cost-of-ownership analysis — well before a single rack is ordered. By treating the edge as a design problem first and a deployment problem second, operators avoid the costly cycle of building, discovering thermal or power limits, and rebuilding.

The age of AI is accelerating, and the need to own and control that AI infrastructure has never been more important. Owning it begins with designing it properly.

Related Articles:

Infobip's Startup Tribe Turns Five, Freeing Founders From a Hidden Cost of Growth

AHOY Selects Montreal as its Second Global HQ, Deepens Mila Partnership, Commits to 500 New Jobs, And Launches World's First Physical AI Lab

Tech Revolt Magazine - June 2026