An article by Abdullah Al Jaafari, CEO and Founder of Resquad AI

At the end of this month, Riyadh will host the fifth edition of LEAP. It is easy to forget how unlikely the first one felt back in 2022. A brand-new technology conference drew more than 100,000 registered attendees from over 80 countries to Saudi Arabia.

Nobody is asking whether LEAP can attract an audience now. Last year’s edition drew more than 200,000 attendees and opened with $14.9 billion in AI investment announcements on its first day. Those announcements pushed the total value of technology-related infrastructure investment announced in Saudi Arabia since LEAP debuted past $42.4 billion. The event has settled one argument: Saudi Arabia, and increasingly the wider region, can attract capital, companies, and attention on an enormous scale.

That is exactly why I will be watching this year’s edition through a different lens. The interesting question in Riyadh is no longer how much is being invested. It is what that investment actually produces: technology that gets deployed, products genuinely built in the region and, above all, people capable of building and sustaining them.

Every announcement from the main stage can be held against that filter. Four areas will tell us the most: infrastructure, agentic AI, Arabic-first products and talent.

The first is infrastructure, although I suspect the headline numbers will tell us less than they used to. I expect more data centres, more cloud capacity, more chip deals and another chapter in the HUMAIN story – fitting in a year Saudi Arabia has formally designated its Year of Artificial Intelligence. HUMAIN itself was launched in 2025 to operate across the AI stack, from data centres and cloud infrastructure to models and applications.

The progress is real. In only a few years, the Gulf has moved from being primarily a buyer of global technology to building a serious part of the infrastructure behind it.

But buying compute is something the Gulf has already proved it can do. The more revealing question is what happens after the infrastructure switches on.

Who is building products on the capacity announced a year ago? Who deploys them inside companies and government agencies? Who is still there six months later, addressing the less glamorous problems that never appear in a press release?

Gigawatts are an input. This year, I want to hear about outputs.

The second is agentic AI, because agents are where deployment either happens or it does not. At recent events, agents mostly lived in demonstrations and keynote videos. The conversations I am hearing now are different. Companies are asking about integration schedules, costs and ownership. Government organisations are asking where agents can run inside real services rather than controlled experiments.

The UAE has gone furthest. Since January, its National Artificial Intelligence System has served as an advisory member of the Cabinet, the Ministerial Development Council and the boards of federal entities and government companies. In June, the UAE went further, establishing an Artificial Intelligence and Data Authority whose mandate explicitly includes a unified digital government system using agentic AI.

So at DeepFest, which runs alongside LEAP, I will be listening for fewer presentations about what agents may eventually do and more examples of where they already run, what they cost and whether they work.

Deployments are evidence. Demos are not.

The third is Arabic-first technology, because it separates products built for this region from products merely sold to it.

For a long time, Arabic support appeared somewhere near the end of a product presentation. It was treated as a localisation task: build the product in English, translate the interface and declare the region covered. The results were often technically functional but culturally weak.

That is becoming harder to defend. The models have improved, governments are asking for technology designed around local needs, and a language used daily by more than 400 million people cannot seriously be treated as a niche.

If this is the year when products designed in Arabic, for Arabic-speaking users, reach the main stages rather than a corner of the exhibition hall, it will signal something larger than better language support. It will mean the region is starting to produce more of its own technology, not simply procure it.

It will also change who can take part in building here.

Which brings me to the fourth and hardest area, and the one I care about most: talent.

Talent will be discussed constantly at LEAP. It always is, and the initiatives behind the discussion are serious. Saudi Arabia’s national data and AI strategy aims to train and qualify more than 20,000 data and AI specialists by 2030. More recent figures from SDAIA put the number of specialists and experts reached through its training programmes at 14,495.

Its SAMAI initiative passed one million citizens trained in AI last year, and the second phase has expanded into 11 government ministries, with an emphasis on practical adoption in the workplace. The UAE, meanwhile, launched its National Programme for Coders with a scheme to grant golden visas to 100,000 coders, including people living outside the country.

Yet the global competition for those skills remains sobering. IDC estimated that by 2026 more than 90 per cent of organisations worldwide would feel the effects of the IT skills crisis, with $5.5 trillion in losses caused by product delays, impaired competitiveness and lost business. AI skills were identified as the most in-demand among enterprises surveyed.

In practice, the situation means the Gulf is competing for the same scarce technical talent as the United States, Europe, India and every company that has decided it needs an AI team.

Attracting people to the region is necessary, particularly now. But it cannot be the only plan. If sovereign AI depends indefinitely on imported expertise, then an important part of that sovereignty remains outside the country.

The frustrating part is that no keynote can fix this issue. A government can fund a data centre or license a model. But you cannot sign an MoU that brings an experienced engineer into existence. Ten years of production experience cannot be procured in time for the next conference.

Something has changed over the past 18 months, though, and I think it deserves much more attention than it is getting.

The entry point into software has moved

A founder in Jeddah who understands logistics, or a doctor in Dubai who knows exactly where clinics waste time, can now describe an idea in ordinary language and have a working version of it by the end of the day.

Andrej Karpathy coined the unfortunate phrase “vibe coding” in early 2025 for one version of this way of building. I do not particularly like the phrase, but the change behind it is real.

Within weeks, Y Combinator partners said that for roughly a quarter of the companies in its Winter 2025 batch, about 95 per cent of the code was AI-generated. Importantly, YC also stressed that these founders were highly technical; AI was changing how they built, not eliminating the need to understand what they were building.

At the other end of the spectrum, Replit chief executive Amjad Masad said 75 per cent of the platform’s customers never write a single line of code. Software creation is increasingly expanding beyond the people who learned to build it in the traditional way.

Of course, generated code does not remove the need for engineers. Quite often, it proves why they are needed.

The first version works. Then the product meets real users, real data and real security requirements, and architecture, reliability and maintenance stop being abstract words. Even as Y Combinator highlighted the amount of AI-generated code in its latest companies, its partners were careful to make the same point: founders still need enough technical depth to understand the code and find the bugs.

I have watched people go through this process. They build something small, discover its limits and start asking much better questions. They learn because they finally have a real project in front of them, not because somebody gave them another introductory course.

Some bring senior engineers into the project. Some become capable technical leaders themselves. In both cases, the project becomes the beginning of their technical education.

This is why I think the shift belongs at the centre of the region’s talent conversation rather than at its margins.

The Gulf does not need every future founder to spend 10 years becoming a software engineer before testing an idea. It needs far more people with local knowledge to begin building and enough experienced engineers around them to turn the good ideas into dependable products.

Banks running innovation programmes, universities designing capstone projects, ministries procuring software and large companies retraining their staff can all treat first-time builders as the beginning of a talent pipeline rather than a curiosity.

The Gulf has spent the past few years securing access to compute. The next few will be judged on whether that access produces more builders, more products and more companies created here.

The number I would like to hear

LEAP 2026 will be impressive; that much is safe to say. The halls will be full, the schedule will be dense, and the announcements will once again be counted in billions.

But there is another number I would like to hear from the opening stage: How many people in this region built and released software for the first time during the past year?

Not how many attended an AI course or opened a coding assistant, but how many made something another person could actually use.

It is the one number that connects everything above.

It tells you whether the data centres have tenants, whether the agents have owners and whether Arabic-first products have makers.

People will once again describe chips as the new oil. Perhaps they are. But access to chips will not decide which countries build the strongest technology economies. The more important question is what people create with them.

When LEAP begins measuring new builders alongside deals and infrastructure, we will know that the region’s sovereign AI plans are becoming more than an investment story.

See you in Riyadh.