MENA’s US$1 Trillion AI Opportunity Has a Grid and Water Problem

The Middle East and North Africa may have one of the world’s biggest AI growth opportunities.

It also has one of the hardest environments in which to cool the machines that opportunity depends on.

The World Economic Forum says AI could unlock more than US$1 trillion in economic and sustainability value across MENA by 2030.

A companion infrastructure report carries the less glamorous message: compute does not scale by itself.

The US$1 trillion figure is not a forecast of AI company revenue

WEF’s estimate combines several categories of potential value.

It includes up to US$200 billion in sustainable financing, US$370 billion in sustainable investment flows enabled by the public sector, US$250 billion in avoided long-term asset losses, US$90 billion in improved energy-sector valuation and US$100 billion in sustainable projects launched through cross-sector partnerships.

Those are scenario-based value opportunities.

They are not a prediction that MENA AI companies will generate US$1 trillion in sales.

The infrastructure advantage is real

MENA may be able to connect new data centres to the grid in roughly two to five years, according to WEF.

In some mature markets, equivalent connections can take more than a decade.

That speed can become a competitive advantage for countries trying to attract AI infrastructure.

The UAE already has the capital, network investment and policy ambition to take advantage of it.

But the region’s climate makes cooling expensive

Cooling can account for more than 40% of total electricity consumption in data centres globally.

In the Gulf, extreme heat raises the difficulty.

Some conventional evaporative cooling systems can consume around 1.5 to 2 litres of water per kWh of cooling.

Water itself may require energy-intensive desalination, creating a loop where more compute increases cooling demand, cooling increases water demand and water production increases electricity demand.

This is why data centres are becoming an energy-policy story

A hyperscale AI facility is not a normal commercial building.

It can concentrate very high and sometimes volatile electricity demand in one place.

AI training and inference workloads can shift quickly.

Grid operators therefore need to plan for both capacity and demand variability.

The WEF report argues that data centres should be treated as critical, dynamic loads rather than ordinary customers.

The UAE is already building the compute side

Robius recently examined Mistral AI’s interest in UAE data-centre investment and the broader move toward sovereign compute.

The UAE also has large-scale expansion under way through G42, Microsoft and Khazna Data Centers.

The strategic logic is clear: if the country wants to host more AI models and services locally, it needs more compute under local control.

The WEF analysis adds the missing question: what powers it, and at what resource cost?

Network capacity is another part of the equation

Compute has to connect to users, cloud regions and international markets.

Etisalat plans to increase UAE international connectivity from 20 Tbps to more than 500 Tbps by 2030. Robius explained why that target belongs in the AI infrastructure story.

The physical AI stack therefore has at least four constrained layers:

• compute;
• electricity;
• cooling and water;
• network connectivity.

Local control does not solve resource constraints

Sovereign AI discussions often focus on where data and models are hosted.

That is important.

But a locally controlled data centre still depends on stable power, water or cooling infrastructure, network capacity and hardware supply chains.

Robius saw the same principle at a smaller operational level in e& UAE’s locally controlled hosting stack. Control of the service environment matters, but it does not remove the physical infrastructure underneath it.

Better cooling could become a regional technology market of its own

The constraints also create opportunities.

Closed-loop liquid cooling, dry cooling, treated wastewater, heat reuse, workload scheduling and AI-based energy optimisation can all reduce resource pressure.

That means the next wave of regional AI infrastructure investment may not only benefit GPU vendors and data-centre operators.

It may create demand for power electronics, cooling technology, water treatment, grid software, energy storage and infrastructure analytics.

The grid connection race can move the bottleneck

If MENA can connect data centres faster than mature markets, developers will move more quickly.

But fast connection does not guarantee long-term system resilience.

Adding large loads without coordinated generation, transmission, storage and demand management can simply move the bottleneck from permitting to operations.

That is why the WEF report frames AI infrastructure as a systems problem.

What investors should ask before celebrating the US$1 trillion

The useful questions are operational:

• How much new grid capacity is actually committed?
• What is the expected power demand per facility?
• How is cooling designed for Gulf temperatures?
• What water source is being used?
• Can workloads shift during grid stress?
• What network capacity connects the site?
• Which resource costs are included in the economics?

The Robius Read

MENA’s AI advantage is not only money, ambition or access to energy.

It is the ability to build the entire physical system faster than markets where infrastructure is already constrained.

But speed creates its own risk.

If compute expands faster than grids, cooling systems and water planning, the infrastructure advantage can erode.

The US$1 trillion opportunity will therefore depend on something much less glamorous than the next frontier model: power cables, substations, cooling loops, water systems and the ability to operate all of them as one coordinated stack.

Sources

• World Economic Forum: AI could unlock US$1 trillion for MENA, 14 October 2026
• World Economic Forum: Powering AI at Scale, 14 October 2026
• World Economic Forum: Unlocking Value, 14 October 2026

Checked 16 October 2026. The US$1 trillion figure is an estimated opportunity across several economic and sustainability categories, not a forecast of AI-sector revenue.

Robius.news | Dubai, UAE | 2026 | Built to be first. Built to be trusted.