Amazon Web Services confirmed at LEAP 2026 that its first Saudi cloud region remains on track for December, backed by more than $5.3bn. Alongside it, AWS and Humain committed a joint $5bn to an AWS Humain AI Zone, extending a collaboration that now spans region, zone and compute.
Amazon Web Services used LEAP 2026 to confirm that its first cloud region in Saudi Arabia remains on track to launch in December, backed by investment the company has put at more than $5.3bn. Alongside that confirmation, AWS and Humain committed a joint $5bn to an AWS Humain AI Zone, a dedicated block of artificial intelligence infrastructure inside the Kingdom.
A cloud region and an AI zone are different pieces of construction, and the difference matters to anyone selling into either.
A region is a set of availability zones — physically separate data centres, far enough apart to fail independently and close enough to run synchronous replication between them. That geometry sets the civil programme. It requires several sites rather than one, each with its own grid connection, its own cooling plant and its own security perimeter, linked by dedicated fibre on diverse routes. The buildings are conventional in construction terms; the electrical and mechanical content is not, and the fibre routing between them is a civil works package in its own right.
An AI zone is a different animal. Training and inference at scale concentrate power density in a way general-purpose cloud does not. Racks that would once have drawn a few kilowatts now draw tens of kilowatts, which moves the cooling problem from air to water and changes the structural loading of the floor beneath them. A campus built for that is closer to a light industrial facility than to a conventional data hall.
The pattern across Humain's announcements at the show is of capacity being contracted in layers, each with a different partner. Together with the US firm Together AI, the company set out 55MW of artificial intelligence capacity and a new data centre capable of scaling to 250MW, and launched a Saudi artificial intelligence cloud running on NVIDIA Blackwell Ultra hardware. It has separately been expanding capacity with AMD and Cisco.
Read as a construction pipeline rather than a technology roadmap, the sequence is consistent. Land and power first, then the shell, then the electrical and mechanical fit-out, then the hardware — and the hardware partner changes far more often than the building does. A campus specified for one generation of accelerator has to accept the next, which is why the flexible part of the design is the cooling distribution and the power train rather than the rack itself.
The December date on the region is the nearer test. Cloud regions slip regularly, and a launch date restated at a public event three months out is a firmer signal than an announcement made two years ahead of one. It also sets a marker other operators are scheduling against: a live region changes what latency-sensitive workloads can be hosted in the Kingdom rather than in Bahrain or the Emirates, and that in turn shapes where the next tranche of capacity gets built.
Neither AWS nor Humain has published site locations, contractor appointments or a phasing schedule for the AI Zone. Those are the disclosures that convert a headline figure into work for civil contractors, electrical subcontractors and cooling specialists, and none of them was made this week.