Huumain’s calculations equate land and power in a plan to build 6 gigawatts of AI computing

Competition for AI leadership in the region is no longer measured by the number of memoranda of understanding, but by the actual ability to secure megawatts, guarantee electricity supply lines, and obtain the land needed to run massive data centres. This clear shift toward heavy-infrastructure standards is reflected in the details of Huumain’s plan, owned by the Public Investment Fund, which its CEO Tarek Amin announced at the LEAP 2026 conference in Riyadh, setting operational targets that place power and land at the core of the expansion strategy.

Huumain aims to develop infrastructure capacity of up to 6 gigawatts dedicated to AI within the Kingdom, after the company secured 14 gigawatts of grid capacity and acquired 211 land sites, including seven complexes fully specialised in processing AI data and algorithms. This scale of resources reflects the belief that the current global challenge is no longer limited to chip supply alone, but also to providing ready-made power and space capable of handling unprecedented computing loads.

Building computing capacity in parallel with securing commercial demand is the core pillar for preventing data centres from becoming idle assets.The company’s timeline maps a path that starts with 250 megawatts in operation by the end of 2026, then jumps to 1 gigawatt by 2027, based on four operational pillars covering power and data centres, computing and intelligence, and physical AI, a system established in May 2025 to move beyond a business model that offers a single isolated service.

The execution side of the plan is translated through a network of direct technology and industrial partnerships, led by the joint announcement with AMD and Cisco to launch infrastructure based on Instinct MI 355 X processors and EPYC processors alongside Cisco networking, with plans to roll out up to an additional 250 megawatts gradually from the second half of 2027 to reach 1 gigawatt by 2030. In addition, construction will begin on a 100-megawatt data centre for its first-phase in Oxagon, Neom, in cooperation with Data Vault; work with Amazon Web Services will be expanded to provide 50 megawatts by 2028; an agreement with Together AI will secure 55 megawatts and build a new 250-megawatt centre; and the company will host Reflection AI models locally and partner with Applied Intuition to operate thousands of autonomous trucks by 2030.

Localising these massive computing capacities practically changes the rules of the game for engineering teams and tech companies across the Gulf, Egypt and the region as a whole.Instead of relying on renting cross-border cloud processing capacity, which raises response-time costs and complicates compliance and data-sovereignty requirements, operating these megawatts locally provides a low-latency environment for training and running large enterprise models, giving financial institutions and logistics sectors in the region a direct ability to negotiate inference costs and build physical AI solutions supported by stable, regionally available computing power.