The UAE didn’t treat AI like another flashy IT project.
They made it plumbing.
New research from INSEAD and the tech group Yango spells it out. The Emirates have moved faster and harder than any comparable government at treating artificial intelligence as public infrastructure. Not a portfolio. A utility. The paper, AI as Public Infrastructure: Lessons from the UAE for GovernmentTransformation, suggests the rest of the world could learn something from this. It isn’t about the hardware. It is about how you rebuild an institution to hold it.
Why pilots always die
Most governments love launching AI pilots.
They launch them. Then they let them wither.
Few manage to turn these experiments into something durable. The research argues execution usually fails for three boring but lethal reasons: data is fragmented, the talent gap at the policy-tech intersection is massive, and governance frameworks are always chasing the deployment, never leading it. The UAE experience shows what happens when you actually fix those gaps. It turns out success is a design problem, not just a tech one.
Leadership. Redesign. Procurement.
The paper points to three institutional choices. Not tech advantages. Choices.
- Leadership commitment that doesn’t flicker.
- A domain-level redesign of public sector processes.
- Procurement used as a strategic lever.
The future of government AI will be determined less by access technology than by the ability redesign institutions around it
Abu Dhabi and Dubai played it differently but aligned with federal goals. Abu Dhabi went for the infrastructure play. Sovereign cloud. Shared platforms. A commitment of AED 13 billion. They treated AI like a foundation. You can’t rush a foundation. Dubai wanted speed. They set up task forces and structured pipelines to blast through from pilot to scale.
Look at the Dubai Centre for AI. 33 government entities came up with 183 ideas. Dozens of them got killed by technical reality and strategic misalignment. Only fifteen survived. These were the high-impact hits: mobility, health, logistics. Discipline wins.
Then there is TAMM in Abu Dhabi. It started as a platform but became an AI-enabled engine hosting over a thousand services. It reuses modular containers across different government channels. It proves AI can be shared public infrastructure. It doesn’t have to be isolated departmental tools hoarding data behind firewalls.
The structural roadblocks
The research lists five persistent barriers.
They exist everywhere.
- Data environments are broken.
- Cross-entity coordination is weak.
- We lack policy-technology translators.
- Procurement is rigid and hates iteration.
- There is no capacity to measure the actual risk or the unintended consequences
The UK. Singapore. The US. The EU. China. All have the tools. All have different results. Why? Because tools don’t govern people. Institutions do.
A vision built over decades
The UAE Government 4.0 strategy isn’t happening by accident. It is moving fast because it has had 25 years to build the digital ground floor.
They released the national AI strategy back in 2017. First to appoint a minister for it. They knew. AI was always going to be essential public infrastructure. Recently the goal has been set. Agentic AI in 50% of sectors, services, and operations. Within two years. That isn’t a timeline. That is a deadline.
Speed like this requires a sense of urgency. It requires radical change in how you actually govern. It works here because they stopped asking if AI is useful and started asking how the state runs without it.
So what do you do with that?
Or do you just wait for the next pilot to die?
