Rising Data Centre Costs in Saudi Arabia: How They Affect the Future of AI in Your Company

Building a data centre in Saudi Arabia costs about $10.80 per watt, roughly SAR 40.5, according to Turner & Townsend's Data Centre Construction Cost Index published on 26 October 2026. The Kingdom ranks 40th of 62 markets: more expensive than Qatar and the UAE, but far cheaper than Tokyo, which tops the table at $18.28. Costs are rising worldwide: 52% of the more than 400 experts surveyed saw increases above 6% in 2026, and AI facilities cost more than conventional ones. Your company does not build data centres, but it ends up paying for them in its cloud and AI bill. What you control is how you consume and what your contract says.

Where does Saudi Arabia stand in the index?

The Turner & Townsend index measures the cost of building a typical 60 MW data centre in 62 markets, based on more than 500 live or recently completed projects. This is where the Kingdom sits:

A figure in "dollars per watt" is the cost of building the centre divided by its power capacity, so it lets you compare centres of different sizes in different countries. A facility built for AI draws more power in the same space, so it needs stronger cooling and connections, and that is what pushes its cost up.

By our calculation, a 60 MW centre at $10.80 per watt costs about $648 million to build, roughly SAR 2.43 billion, before power and operations. The index gives no year-on-year rise for Saudi Arabia, but it says the average cost in North American and European markets rose 13% in a year (index page).

Why are costs rising?

• AI demand. The report says demand for computing power has outpaced the supply of powered land and skilled labour, and 52% of respondents expect most of their 2027 projects to be data centres for AI workloads.
• AI facilities cost more. In the United States they cost about 10% more than conventional ones because of rack density, liquid cooling and compressed schedules, and the most advanced reach between $14 and $16 per watt (index page).
• Skilled labour shortages. 81% of respondents said skilled labour in their region is moderately or highly constrained.
• Power. 56% named power as a major obstacle, 52% said utility connections typically take one to two years, and 32% said they take over two years.

In Saudi Arabia demand is growing: HUMAIN alone plans 3 gigawatts of data centre capacity by 2030, according to PIF's release, as we explained in our article on Aramco's investment in HUMAIN. The report itself expects inflation to ease in 2027, with 28% of respondents expecting rises of only 1% to 5%.

How does this affect your company?

The index measures construction cost, not the prices you pay your cloud provider, and we found no official figure on cloud price changes in Saudi Arabia. But the link is clear: whoever builds the centres recovers the cost from customers over years. That has three practical effects:

• AI compute is the most expensive part. Facilities built for it cost the most to build, so running large models will stay the fastest-growing line on your bill.
• Hosting inside the Kingdom has value and a price. If you need to keep data in the Kingdom, read what a local cloud region means for your business and compare prices between the local region and others before deciding.
• Available capacity is expanding. The new centres we covered in our AI infrastructure article mean more options over time, but they take years to build, and power is the slowest link.

5 steps to protect your company's AI budget

1. Measure the cost of each use separately. Know what you pay each month for each tool or use case, not just the total bill, and ask the provider for monthly reports broken down by project or department.
2. Match model size to the task. Simple tasks such as classification and short summaries do not need the largest and most expensive model.
3. Fix the price in the contract. Ask the provider in writing how long the price is fixed and on what terms it can rise at renewal, and avoid long commitments at a price that can change without a cap.
4. Reserve capacity only for steady workloads. Committing in advance lowers the price when usage is stable, and our guide to controlling your cloud bill covers other ways to cut it.
5. Keep the freedom to move. Require data export and the right to switch providers, so you are not captive to a provider that raises its price.

At Origami we review what companies pay for cloud and AI and redesign their usage to fit the budget. Contact us if you would like us to review your bill.