Data governance and system fragmentation force a hybrid AI architecture, moving beyond the model bottleneck in Saudi Arabia

According to the “Saudi Internet Report” issued by the Communications and Space Technology Authority, the adoption rate of AI tools among internet users in the Kingdom jumped to 45.2 % in 2025 from about 21.5 % in 2024. This trend coincided with figures from the General Authority for Statistics showing that 33.1 % of establishments adopted AI technologies in 2025, with an annual growth of 20 %. However, this rapid expansion in adoption shifts organizations to an operational challenge that is entirely different from merely providing chips or buying inference capacity for language models, as projects encounter the problem of data fragmentation across legacy systems, cloud environments, and separate department repositories, which hinders achieving the desired economic return.

During the participation of OpenText executives at the LEAP 2026 exhibition in Riyadh, the company’s Vice President for the Middle East and Turkey, George Chamber, said that information fragmentation is the most common obstacle to moving AI projects in Saudi Arabia from the experimental stage to full enterprise production. He explained that the issue does not lie in a lack of data, but in the difficulty of collecting it and verifying its quality and governance so that it becomes ready for automated consumption. Harold Adams, the company’s President for Japan, Asia Pacific, the Middle East and Africa, affirmed that corporate dialogue has moved beyond testing algorithms’ summarisation capabilities to concerns about safe deployment and permission controls, noting that models are the engine while trustworthy data is the real fuel; AI defines what is technically possible, and governance draws the permissible boundaries within the organisation.

Introducing AI agents into operations raises the level of operational risk; an error is no longer just inaccurate information but a chain of automated actions executed within sensitive systems.OpenText believes that granting agents execution autonomy requires three core pillars: trustworthy data, strict governance, and clear responsibility delineation, while keeping human oversight over workflows and enabling decision audits. The highest levels of maturity currently appear in repetitive tasks that handle high document volume such as customer service and knowledge archiving, whereas top-level strategic decisions remain constrained by human judgment and accountability, especially given that about 40 % of Saudi organisations are classified as AI leaders amid expectations that the market will exceed $4.37 billion by 2034.

Data-sovereignty requirements are no longer a later planning stage; they have become the standard that defines a project’s technical architecture from the initial inception.With the company’s regional headquarters opening in Riyadh to provide technical consulting and apply governance frameworks, the Saudi business sector is moving toward adopting hybrid and private AI models, led by the government, financial services, energy and health sectors. These sectors do not handle data with a uniform approach; instead they offload workloads to public cloud for non-sensitive tasks while keeping financial, medical records and citizen data within private, local architectures that ensure traceability and legal compliance.

This shift forces technology leaders in the region to reorder priorities, as betting on importing the latest models is no longer sufficient to justify spending unless it is preceded by a restructuring of customer records and operational documents. Your success in AI investment will not be measured by model size or the number of experiments you run, but by your ability to build governed data pipelines and connect hybrid systems in a way that protects data sovereignty and reduces operational risk, thereby moving the technology’s true value into daily decision-making rather than isolated technical showcases.