Saudi Arabia third worldwide in building major models: the software sovereignty equation imposes its choices regionally
Listen to this article
Read by Anchor
The Saudi Data and AI Authority (SDAIA) placed Saudi Arabia third worldwide in developing leading AI models, after producing eight major language models by 2025, five of which were rated among the most prominent internationally. This shift moves competition in Gulf and regional markets from merely consuming imported APIs to building a sovereign model layer capable of local operation and evaluation.
The national achievements announced went beyond foundational models such as the national model “Alam” and other specialized models produced by Saudi entities, extending to the development of 58 local AI applications by 2025. This coincided with the launch of the “BALSAM” index in partnership with more than ten institutions, a metric that evaluated 20 Arabic-language models, giving the ecosystem a unified measurement tool. These steps reflected in the kingdom’s positions in international reports, placing it fifteenth among 195 countries in the 2025 AI Government Readiness Index from Oxford Insights, and nineteenth globally in the 2025 Human-Centric AI Index from Stanford’s Institute for Human-Centric AI.
In contrast to this digital progress, SDAIA adopted a cautious strategic stance toward what is known as artificial general intelligence (AGI). The authority stated that the future of these systems, capable of learning, adapting, and performing cognitive tasks at a level comparable to human abilities, remains surrounded by mystery and differing expert opinions on its definition and feasibility. It warned that the rapid expectations sparked by generative AI do not constitute decisive evidence of the imminent arrival of that stage.Instead of tying preparations to a fixed timetable, Riyadh considered AGI a future scenario that requires prior preparation based on building organizational and technical capabilities.
This preparation relies on specific tracks that include developing national competencies, expanding international partnerships, and increasing the adoption rates of smart solutions across sectors, alongside improving data quality and governance. The kingdom adopted a phased governance approach that begins with establishing ethical principles and institutional frameworks before expanding to advanced risk management. This approach reflects a clear recognition of the dual nature of opportunities and risks: while advanced AI supports health, education, and scientific research, it simultaneously imposes cyber-security challenges, privacy, bias, information accuracy, and legal accountability for automated decisions.
This institutional building extends to academia through initiatives such as SAMAI and training programs, camps, and technical competitions. In the university field, the Faculty of Computing at Fahd bin Sultan University in Tabuk provides an operational model for this strategy, having introduced specialized programs in AI, machine learning, and data science.All undergraduate students, regardless of their majors, were required to study a general AI course.The faculty’s AIMD Lab offers an environment for testing cyber-security algorithms and training models, while the university provides a quantum computer for exploring complex computational challenges.
What does this mean for the region?
This transformation places technology managers in Gulf, Egyptian, and Levantine banks and companies before a new practical reality: reliance on Western model APIs is no longer the only option, and Arabic models evaluated by unified standards such as the BALSAM index offer an alternative that reduces data sovereignty risks and governance.
At the skill level, making AI teaching mandatory in universities and providing advanced computational infrastructure such as quantum computers raises regional hiring standards, meaning that superficial knowledge of tools will no longer suffice to compete in a job market that now demands competencies capable of building and governing models rather than merely consuming them.