In an exclusive conversation with Tech Revolt, Deloitte Partner Yousef Barkawie discusses where AI adoption stands across the region, what sovereign AI means in practice, the rise of agentic systems, infrastructure constraints, and the Middle East's opportunity to shape the global AI landscape. 

[For more news, click here]

The conversation around artificial intelligence in the Middle East has moved well beyond whether businesses should experiment with the technology. The focus is shifting toward what happens when those experiments are expected to deliver measurable business value and what organizations need to change to make that possible. 

Yousef Barkawie, Partner at Deloitte Middle East, works at the intersection of technology, business strategy, and enterprise AI adoption. He leads the organizations growth platforms and technology alliance relationships across the Middle East and is responsible for its AI strategic growth offering, working with clients on the adoption of artificial intelligence. 

Rapid advances in the technology have changed the equation over the past year. Access to computing infrastructure has expanded, while language models, foundation models, and reasoning capabilities have become increasingly sophisticated. The bigger challenge, however, lies in how organizations adapt their businesses around these capabilities. Barkawie believes the companies that understand this transformation will be better positioned to turn AI investments into meaningful results. The questions facing the region now extend beyond adoption. Organizations must consider how they control their AI infrastructure and data, where they can innovate and create rather than simply consume technology, and how increasingly autonomous systems will change the way enterprises operate. 

In this exclusive interview with Tech Revolt, Barkawie shares his perspective on those developments, along with the role of talent, power, governance, and business transformation in determining how the region's AI ambitions translate into results. 

Regional Expansion & the Middle East’s AI Opportunity 

Q: Where regional AI adoption actually stands now versus a year ago: What separates the organizations running AI in production from those still cycling through pilots? 

A: Things have changed quite a bit over the past year, and we've seen this trend not just locally in the Middle East, but around the world. First, access to compute and infrastructure has increased dramatically. The technology itself has advanced: the language models, the large reasoning models out there today are in much better shape than they were a year ago. That means the technology adoption cycle has improved and become faster. 

I think the biggest difference, especially for our clients in this region, isn't in what they're building. It's the realization that the business transformation needed to accompany AI adoption is a priority, and that's what will make or break the effectiveness of AI at scale. The organizations and clients who understand and execute on that business transformation agenda, who make AI more meaningful for themselves, are the ones winning in the market today. 

Q: Sovereignty: What does "sovereign AI" mean at an architecture level for a regional client, rather than as a procurement or policy position? 

A: AI sovereignty is a global topic; we're not the only region looking at it through this lens. What's happening right now is that AI sovereignty isn't just about who I buy from, though that's an important topic. It's also about where my infrastructure sits, physically and contractually. That's the first dimension. 

The second is data: data sovereignty, data protection, and data residency laws and regulations have been around in the Middle East for a while, but they've taken on a much bigger role when it comes to AI. 

AI governance at any organization adopting AI at scale is where sovereignty is actually protected and maintained across layers, including how you preserve institutional knowledge and experience embedded in your knowledge bases and databases, wherever they're physically located.  

At a national level though, how long is a piece of string?  Sovereignty consideration may extend to zero-trust constructs, whether it is at the hardware design and manufacturing level, at the cyber security solutions that protect critical information assets, even the base technology stack that runs systems, sovereignty policies need to address resilience and dependencies across the entire value chain.  

AI sovereignty has many dimensions. It's not only about procurement; it's a technology architecture consideration, governing policies, and it's a business practice and governance process. Getting all of these pieces talking to each other the right way is how you get AI sovereignty right. 

Q: Where does the Middle East have a genuine opportunity to lead globally rather than adopt what's built elsewhere, and where it realistically doesn't? 

A: I'll reflect a little. Four or five years ago, Deloitte was running a twenty-four-hour AI marathon connecting colleagues around the world, talking about AI adoption when Gen AI was still quite new. Even then, I said I believe AI is the great equalizer for a region like the Middle East, because it's no longer about having massive scale the way we'd see in, say, Southeast Asia, or needing manpower at the scale we used to. 

It's become a level playing field for true adoption and true innovation. The ability to innovate at scale is no longer a monopoly held by a few regions; it's become open and fair. We're seeing so many entrepreneurs in the Middle East coming up with creative concepts, using technology in a way that reflects where the world is heading anyway: the globalization and scaling of technology. The Middle East has a lot to gain just by applying creativity and innovation to the way things get done. 

At the same time, there are massive opportunities and big plays happening here because we have the power, the energy, and the funding sources to compete at a much larger scale. We're seeing a lot of innovation, including what HUMAIN is doing, and hardware manufacturers becoming available in the region. I think over the next five years, we'll see this region become a producer of knowledge and technology, not just a consumer. 

The Shift From Passive AI to Agentic Systems 

Q: Agentic AI is increasingly positioned as the next major shift in enterprise technology. What is genuinely different about agentic systems, and where do you expect meaningful enterprise adoption first? 

A: Agentic AI is what's making AI adoption scale across organizations, because it's the difference between AI as a passive topic, opening a chatbot, asking a question, waiting for an answer, and that's it, versus an agent that runs continuously, makes decisions, and takes action. There's still a human in the loop and methods to govern and control it, but we've gone from a passive technology to an active one that does things without being prompted every time. That shift, and the value case behind it, is significant. Nobody thinking about AI adoption today is thinking about it without talking about agents and the agentic capabilities behind them, because the value of a passive interaction is minimal in an enterprise and business context. The value of automation at scale is really where agentic AI plays its biggest role. 

The difference between agentic AI and traditional automation matters too, and it's something we spend a lot of time working through with our clients. Agentic AI isn't macros or an RPA bot. It's about making informed decisions worked through with a probabilistic assessment and all the value that AI brings. Because autonomy is given to an agent, it can do far more than the automation we used to have. That obviously needs to come with the right governance layer, controls, processes, and interventions. The true value of AI in the enterprise is agentic, not passive. 

Q: Infrastructure and the physical layer: compute, power, talent. Which constraint bites first in this region over the next 24 months? 

A: I don't think it's compute anymore. Compute issues are being resolved, and it's never going to feel like enough because the world will always be hungry for compute, and therefore power, since the two go together. I'd add talent to that list, and governance, because that's where problems will surface if we don't solve them first. 

Q: Of the AI projects you saw begin across the region in 2025, what proportion, if any, are in production today, and what happened to the rest? 

A: A lot of the AI solutions in production today needed very little business transformation to maximize value; they're the no-regret type of solutions. Where clients are struggling or taking the most time to make AI really work, are the organizations completely rethinking their business operations and reimagining how they run their business functions. Those are the ones taking the longest, and the ones that haven't yet scaled out their AI use cases and value cases are the ones still planning. Those are the ones to watch: they're the ones that will get the most value out of their investments in the end. 

Barkawie's comments point to a practical challenge facing companies across the region; putting AI to work requires changes inside the business, not just new technology. Companies that are willing to rethink how their teams work and how decisions are made are likely to get more from their investments than those simply adding AI to existing processes. The Middle East has several factors working in its favor, including capital, energy, infrastructure, and a growing pool of technology entrepreneurs. Barkawie believes those conditions give the region room to build and experiment at scale and potentially develop products and ideas for a wider market rather than relying mainly on technology developed elsewhere. 

There are still difficult issues to address, from finding people with the right skills to establishing clear governance around AI and protecting data and institutional knowledge. Those challenges could prove just as important as the technology itself as organizations put more AI systems into day-to-day operations. The next few years will show how far the region can take its ambitions. If businesses can move from testing individual use cases to changing the way they operate, the Middle East could have a much larger role in the global technology market than it has had in the past. 

Related Articles 

Exclusive: Middle East Advances Arabic First Agentic and Edge AI Ecosystems  

Saudi Arabia Bets On Concrete to Win the AI Race, And a New Partnership Shows How  

Saudi Arabia’s AI Future Will Be Defined by Data, Governance, and Business Impact