ARTIFICIAL INTELLIGENCE | SAUDI ARABIA

From AI ambition to enterprise-scale execution

Saudi Arabia has moved decisively into the AI era. The next challenge is making AI work securely, consistently and economically across the enterprise.

Saudi Arabia no longer needs to prove its ambition in artificial intelligence.

Under Vision 2030 and the National Strategy for Data & AI, the Kingdom has built a clear national direction around data, AI, digital infrastructure and skills. Enterprise adoption is following that momentum.

According to Saudi Arabia’s General Authority for Statistics (GASTAT), 33.1% of establishments used AI technologies in 2025, up 20% from the previous year. Adoption is already considerably higher in some of the Kingdom’s most strategic sectors: 61.1% in information and communications, 52.9% in financial and insurance activities and 51% in education. [1]

The question facing Saudi organisations is therefore changing. It is no longer simply: “Where can we use AI?”

For many organisations, the answer is: partly. And that distinction matters.

AI adoption is accelerating. Enterprise readiness has to catch up.

Saudi business leaders are already highly confident about AI.

PwC’s 2026 Saudi Arabia CEO Survey found that 78% of CEOs believe their organisations have the right technology environment to integrate AI, while 61% say they have a clear roadmap for AI initiatives. Eight in ten also believe their organisational culture supports AI adoption. [2]

These are strong foundations. But another figure in the same research exposes the next challenge:

Only 14% of Saudi CEOs say their most-used AI tools can access all the relevant company data and documents they need. Globally, that figure is 29%. [2]

That gap is important because enterprise AI is ultimately constrained by what sits underneath it. An AI model may be powerful. But if customer data is fragmented across systems, business rules are buried inside legacy applications, integrations are undocumented or data lineage is unclear, the model cannot compensate for those limitations.

At scale, AI readiness becomes architecture readiness.

The model is rarely the hardest part

Much of the first phase of generative AI focused on models.

Which model should we use? Should it run in the cloud or privately? Which copilot should we deploy? What proof of concept should we build first? Those remain relevant questions, but enterprise adoption introduces a different set of problems.

Can an AI agent safely access a core business application? Can it retrieve the right information across different data domains? Can the organisation demonstrate where an answer came from? Can permissions follow the employee, customer and data involved? Can business logic that has existed inside an application for twenty years be exposed as a reusable service? Can the organisation measure the cost of AI consumption against the value created?

These are no longer model-selection questions.

They are questions of data architecture, application modernisation, integration, security, governance and FinOps.

And they become more urgent as AI moves closer to core business activity.

In Saudi Arabia, PwC already finds substantial AI deployment across business functions: 44% of CEOs report AI applied to a large or very large extent in demand generation, 40% in support services and 37% in strategic direction-setting. [2].The closer AI gets to operational and strategic decisions, the less tolerance there is for fragmented foundations.

What does an AI-ready enterprise actually look like?

AI readiness does not mean replacing an organisation’s entire technology estate.It means making the estate sufficiently connected, governed and adaptable for AI to operate across it safely.

Four areas become particularly important.

1. Data that AI can actually use

• Enterprise data must move beyond simply existing.
• It needs ownership, quality controls, metadata, lineage and appropriate access mechanisms.
• For AI, the difference between having data and having usable data is fundamental.
• An organisation may possess years of customer, operational and transaction history while still struggle to make that information safely available to an AI application.
• The 14% figure identified by PwC illustrates precisely this challenge in Saudi enterprises. [2]

2. Applications that can expose business capabilities

Legacy does not necessarily mean obsolete. Many long-running applications contain valuable business logic accumulated over years or decades.

The problem arises when that logic can only be accessed through the application itself.

AI-driven organisations increasingly need applications to expose capabilities through APIs, services and well-defined interfaces. This changes the objective of application modernisation. It is also about making the business knowledge inside that application accessible to the wider enterprise — including AI.

3. Governance built into the architecture

As AI moves from experimentation to production, governance cannot remain a separate compliance exercise. Identity, permissions, privacy, auditability, model controls, data sovereignty and responsible AI need to form part of the architecture itself. This direction is already visible at national level.

SDAIA reports 32 regulatory instruments related to data and 13 related to AI, alongside a national AI adoption framework and AI ethics principles designed to support responsible adoption across sectors.
[3] . For Saudi enterprises, governance therefore becomes not an obstacle to AI adoption, but one of the conditions required to scale it confidently.

4. AI economics that the business can defend

A proof of concept is relatively easy to finance. Hundreds of AI-enabled workflows are different.

Model inference, compute, data processing, storage, cloud services, observability and integration all contribute to operating costs.

As adoption grows, organisations need to know more than whether an AI solution works.

They need to understand: What does this process cost? What business value does it create? And how does that change as usage scales? This makes FinOps and cost governance increasingly relevant to AI strategy.

Saudi Arabia is already expanding the infrastructure required to support this future. SDAIA reports that national data-centre capacity increased by 42.4% between 2023 and 2024, reflecting the scale of infrastructure development supporting the Kingdom’s digital and AI ambitions. [3] .The next step is to make sure enterprises can turn that infrastructure into sustainable business value.

From AI projects to an AI operating model

There is a broader shift behind all of this. The first phase of enterprise AI was largely about use cases.

The next will be about the operating model.

Instead of asking whether marketing, finance, operations or customer service can each implement an AI solution, organisations need to consider what common foundations allow dozens or hundreds of AI-enabled processes to coexist. That means treating data as an enterprise asset, architecture as an enabler, modernisation as a continuous capability, governance as part of delivery, and cost optimisation as part of AI design. Saudi Arabia is particularly well positioned for this next phase.

The national ambition is clear. Investment is substantial. Infrastructure is expanding. Businesses are already adopting AI. The challenge is now deeper inside the enterprise. And this may become one of the most important distinctions between organisations over the next few years. The leaders will not necessarily be those with the most AI pilots. They will be those that have built an enterprise capable of turning AI from an experiment into a governed, scalable and economically sustainable part of how the business operates.

Key numbers

78% → 61% → 14% tells the core story: technology confidence is high, strategic intent is strong, but enterprise data readiness remains a critical execution gap.

Sources

[1] General Authority for Statistics (GASTAT) data, reported by Arab News:
“Saudi AI adoption among businesses rises to 33% in 2025.”

[2] PwC, 29th CEO Survey — Saudi Arabia findings, 2026.
View source

[3] Saudi Data & AI Authority (SDAIA), AI Year / national data and AI indicators.
View source