As Gulf states accelerate investments in artificial intelligence and next-generation connectivity, questions about technological power, security, and economic leverage are moving from research labs to the center of national strategy.

In an interview with Professor Merouane Debbah, the Director of Khalifa University’s 6G Research Center, he discussed how artificial intelligence and wireless technologies, along with the concept of “sovereign AI,” are transforming global innovation. He also highlighted why the Middle East is positioning itself as a key player in this transformation.

Debbah was recently named by Google’s Gemini among five Arab figures who helped shape the global technology landscape in 2025, recognition he attributes largely to his leadership of TelecomGPT, the first large language model designed specifically for the telecommunications industry.

Unlike general-purpose AI models built for conversation or content generation, TelecomGPT was engineered to understand how telecom networks operate — and to manage them autonomously. The goal, Debbah said, was to move networks from reactive systems toward intelligent infrastructure capable of diagnosing problems, anticipating failures, and optimizing performance without human intervention.

“We didn’t just train a model,” he said. “We engineered a framework for AI-native telecoms, built around the real constraints and protocols operators deal with every day.”

That shift mirrors a broader transformation that defined 2025, when artificial intelligence began moving from an overlay on existing systems to becoming the logic governing them. Across sectors such as energy, mobility, and communications, vertical foundation models emerged to manage complex infrastructure rather than simply respond to user queries.

AI systems, Debbah said, are now making decisions, coordinating actions across physical and digital environments, and blurring the line between computing and cognition. The next phase, he added, will be marked by the convergence of AI with scientific computing and real-time control, allowing models to evolve continuously inside live systems.

This vision underpins what Debbah describes as “intelligent infrastructure” — systems designed to behave less like machines and more like biological organisms. In practical terms, that means decentralized, self-healing networks where intelligence is distributed across components rather than centralized.

In telecom networks, this translates into AI agents embedded at the edge, enabling cell towers, routers, and base stations to sense disruptions, reroute traffic, adjust parameters, and retrain themselves in real time. TelecomGPT, he said, is among the first attempts to give infrastructure reasoning capabilities — allowing networks to interpret and respond to their environment.

The implications extend beyond telecommunications. Debbah’s concept of “cognitive connectivity” envisions intelligence diffused throughout everyday environments — from streets and vehicles to hospitals and buildings — creating what he describes as a distributed brain.

For individuals, that could mean fewer apps and more seamless interaction. Vehicles would adapt routes based on intent rather than instructions, hospitals would adjust rooms automatically based on patient data, and networks would anticipate demand surges before large events. Achieving that, he said, requires tight integration between advanced AI models and next-generation connectivity, including 6G.

These developments intersect directly with the Gulf’s push for sovereign AI, as countries such as the UAE and Saudi Arabia channel billions of dollars into domestic AI capabilities. Sovereignty, Debbah said, is not about isolation but about maintaining end-to-end control over the AI stack — from data and training to deployment and governance.

In practice, that means training models on localized, domain-specific data, hosting them on national infrastructure, ensuring transparency and auditability, and aligning AI systems with local legal and cultural frameworks. TelecomGPT, trained on private, multilingual telecom data including Arabic, was designed to allow operators to retain ownership of the intelligence running their networks.

Despite this momentum, regional models such as Falcon and Jais have yet to achieve the widespread adoption of global platforms like ChatGPT or Gemini. Debbah said the barriers are largely structural rather than technical, pointing to ecosystem inertia, limited developer tooling, and a lingering bias toward global brands.

That dynamic is beginning to shift in sensitive sectors such as telecom, defense, and health care, where trust, explainability, and alignment increasingly outweigh raw performance metrics. At Khalifa University, teams are now focused on turning research models into deployable, plug-and-play systems rather than academic prototypes.

Looking ahead, Debbah sees 2026 as a potential tipping point for another transformation: the decline of app-based interfaces in favor of agentic AI, where users delegate tasks to autonomous agents instead of navigating menus and screens.

For that transition to scale, he said, systems must solve challenges related to cross-domain orchestration, persistent memory, and governance. In that landscape, the winners will be those who control orchestration layers — including telecom operators, cloud providers, and sovereign platforms — while closed ecosystems risk losing relevance.

Despite rapid advances, some of the most fundamental questions in AI remain unresolved. Debbah pointed to gaps in understanding how intelligence emerges from structure, how multiple models can collaborate effectively, and how AI can reliably design physical systems such as materials or wireless protocols.

Future breakthroughs, he suggested, may come not from larger models but from new architectures inspired by biology and advanced mathematics — an area where he believes the Arab world has a distinct opportunity to contribute by combining scientific heritage with growing technological ambition.

“That,” he said, “is where the next revolution may begin.”