Leap’s first two days bring in a string of data center agreements

Saudi Arabia is building out the infrastructure needed to turn its AI ambitions into a functioning industrial stack. A string of agreements announced around this week’s Leap conference spans capacity, connectivity, computing, and AI applications, as the Kingdom moves to build the physical backbone for an AI economy.

The build starts with data center capacity: Humain and digital infrastructure firm DataVolt have started construction on 100 MW of an AI-ready data center at Oxagon, Neom’s industrial city on the Red Sea coast, according to a statement. The capacity forms part of DataVolt’s 360 MW first phase within a planned 1.5 GW AI data center campus, with the first 100 MW expected to come online in 2028.

There’s more: Al Moammar Information Systems said it will spend USD 1.2 bn expanding its data centers to 192 MW of total capacity; NHC Innovation will invest USD 880 mn in data centers at Khuzam Digital Valley, targeting 65 MW by 2033; and Zain KSA signed a non-binding MoU with DataVolt to explore building data centers in the Kingdom, capacity it would share with government, enterprise, and cloud providers.

The infrastructure to deliver it is coming next: Amazon Web Services (AWS) is targeting a December launch of its Saudi cloud region and has committed up to 50 MW to the Humain AI Zone by 2028, according to a statement. AWS announced back in 2024 that it woul launch a “hyper-scale” cloud region in Saudi Arabia by 2026 as part of a USD 5.3 bn investment ticket. Microsoft plans to open its own Saudi cloud region in November.

And the AI is starting to leave the data center: Humain is partnering with US autonomous-vehicle software company Applied Intuition to deploy thousands of self-driving trucks across Saudi logistics corridors by 2030, Bloomberg reports. Autonomous trucks can use those models and AI buildouts across logistics corridors, while the same infrastructure could eventually support ports, industrial sites, and other assets that need continuous AI processing.

Moving AI into freight exposes a different infrastructure gap: Scaling autonomous freight requires infrastructure well beyond the compute itself, Jesse Marks, founder and CEO of Rihla Research and Advisory tells EnterpriseAM, including consistently mapped roads, reliable connectivity along major corridors, and road networks with sufficiently predictable conditions for autonomous systems to operate safely, he adds.

But more infrastructure does not automatically mean more resilience. The Gulf got a warning in March, when outages at AWS infrastructure in Dubai spread beyond a single data-center failure while a separate power outage hit capacity in Bahrain, disrupting businesses and pushing some workloads onto infrastructure in other jurisdictions.

What’s next? That resilience depends on having “adequate power supplies, fiber routes, and operational systems to sustain the model. If dispersed across geography, this reduces the likelihood that a single disruption or even an Iranian strike could knock the entire system offline during a period of conflict. It, however, may require a more decentralized operating system of power, cables, and cooling to keep centers up and running. That could get expensive,” Marks notes.