Operation 300bn offers a useful blueprint for sovereign AI: lasting capability comes from building the system behind the technology, not simply acquiring the technology itself. A Sovereign AI Factory combines models, trusted data, people, platforms, governance and security into a repeatable environment for deploying AI at scale.

[For more news, click here]

When the UAE launched Operation 300bn in 2021, the headline goal was straightforward. The strategy would increase the industrial sector's contribution to GDP from AED133 billion to AED300 billion by 2031. But look past the number and a broader ambition emerges. Operation 300bn is also about building in-country value, developing national skills and making supply chains resilient enough to withstand a shock. Finished products can be bought. The capacity to create, adapt and improve what a country depends on is much harder to acquire, particularly when external supply is disrupted or national requirements change.

AI leaders now face a similar choice as artificial intelligence moves into government services and critical operations. In these high-stakes environments, acquiring a product is only the beginning. The harder task is building a capability that the institution can control, adapt and sustain as its needs evolve.

Access does not provide control

Generative AI has made sophisticated capabilities widely accessible. A general-purpose model can retrieve information, generate content and analyse data on demand, allowing organisations to experiment quickly. But that convenience can hide the dependencies underneath it, and those dependencies become serious when a government is involved.

A tax authority, defence organisation or space agency needs to know where its data goes, who controls the systems processing it and how an AI-assisted decision can be explained. The institution remains accountable for the outcome regardless of which vendor's model produced it. The technology therefore has to operate within its security and policy boundaries, with controls that the institution can inspect and enforce. A model may help it finish a task today, but sovereign capability demands lasting control over the data, decisions and accountability surrounding that task.

A model is not a factory

Manufacturing offers a useful analogy because the pitfall is familiar. A company can buy sophisticated machinery without building a resilient industrial capability. That machinery becomes productive only inside a wider system with reliable inputs, skilled people, established processes, standards and quality control. AI works the same way. A model is an important component of sovereign AI, but it is not the capability itself. Locating the infrastructure within a national or organisational boundary does not create one either.

A Sovereign AI Factory brings those elements together. It is a controlled environment where teams can build, deploy and govern AI according to the organisation's requirements. Models and data are the raw inputs. Platforms, people and processes turn them into useful services, while security and governance keep those services within defined boundaries. The payoff is repeatability. Instead of running disconnected pilots with bespoke controls, the organisation gains a consistent route from experimentation to production and a foundation it can reuse without starting the governance work again.

Governance belongs on the production line

A Sovereign AI Factory does more than produce models faster. It creates capabilities an institution can trust in a defined setting, such as a knowledge assistant grounded in approved information or an agent supporting a document-heavy government process. A citizen-facing service might require a different level of oversight from an internal research tool. In both cases, success depends on more than model performance. The system has to draw on the right information, fit the process it supports and follow organisational policy. Explainable decisions, audit trails and meaningful human oversight become part of the product rather than paperwork added later.

Manufacturers learned long ago to build quality control into the production line instead of inspecting for defects at the end. AI needs the same discipline. When governance, security and compliance shape testing and deployment from the outset, teams can move faster because those controls do not have to be reinvented for every use case.

Agentic government raises the stakes

The need for this foundation grows as AI starts to act rather than simply advise. The UAE's agentic AI framework aims to transform 50% of government sectors, services and operations within two years. It envisages systems that monitor changes, offer recommendations, manage operations and carry out sequences of actions. Once an agent joins an operational process, the organisation has to define what it can access, which actions it may take and when a person must step in. An unclear boundary can now affect a live service rather than a private experiment. Those boundaries must remain visible and enforceable as the agent works across more data and systems.

Early public-sector initiatives suggest a practical starting point. The Brussels Tax Department has explored agents that retrieve information and consolidate data for staff. In Italy, an agent built for the Ministry of Cultural Heritage made millions of records searchable through plain-language queries. Both started with a bounded problem, trustworthy information and a clearly defined role for people. That narrow scope makes value easier to demonstrate and the risks easier to manage. A factory approach makes the same discipline repeatable rather than exceptional.

The factory is the enduring asset

Operation 300bn recognises that industrial strength comes from the productive system behind the finished product. Infrastructure, standards, skills and supply chains allow a country to manufacture repeatedly at the quality it needs, while international technology remains an important part of the system.

Sovereign AI follows the same logic. Governments do not need to build every model or component themselves. They need a repeatable way to turn models and trusted data into operational capabilities that stay within their boundaries, even as technologies and requirements change. As AI moves deeper into public services, that productive capacity will matter more than possession of any single model.

Contributed by: Soliman Ayman, Agentic AI & Enterprise Transformation Leader, MEA, DXC Technology
Ayman Soliman is an AI and enterprise transformation leader with more than 20 years of experience helping organisations across the Middle East modernise customer engagement through AI, automation and digital transformation. He currently serves as Agentic AI & Enterprise Transformation Leader for the Middle East and Africa at DXC Technology, where his work focuses on translating emerging AI capabilities into practical, enterprise-scale transformation

Related Articles

Exclusive: Who Do You Trust? Rethinking AI Sovereignty

Sovereign AI Is Moving From Government Policy to Enterprise Procurement and This Partnership Shows How

AHOY Is Tackling Two Challenges in the Next Phase of AI