The Gulf's AI infrastructure race is often described through gigawatts. The financing structure may be more important. Saudi Arabia and the UAE are increasingly assembling projects in which sovereign capital, global cloud companies, specialist data-centre developers and chip suppliers take different pieces of the risk.
That is a more mature model than simply announcing a state-funded technology campus. It also makes the commercial test clearer: which partner provides capital, which builds the physical asset, which supplies compute and which brings customers?
Saudi Arabia is combining a national platform with external operators
Saudi Arabia's Public Investment Fund launched HUMAIN to operate across the AI stack, from data centres and cloud to models and applications. One day later, AWS and HUMAIN announced plans for more than $5 billion of joint investment in a Saudi AI Zone.
That commitment sits alongside AWS's separately announced $5.3 billion Saudi infrastructure region. The structures overlap strategically but should not be added together as if every dollar describes the same asset. One is a cloud-region programme, while the later partnership is aimed specifically at AI infrastructure and services.
The UAE model distributes capability across an international alliance
Stargate UAE uses a different structure. G42 sits at the centre of a partnership with OpenAI, Oracle, Nvidia, Cisco and SoftBank. Khazna is responsible for major elements of the physical data-centre build inside the wider UAE-US AI Campus.
The advantage is access to global technology and operating expertise. The dependency is that the project relies on several partners whose hardware, capital and strategic priorities are not controlled by one government or one company.
Specialist developers are becoming part of the capital stack
DataVolt's work at Oxagon shows another layer. The developer agreed an initial $5 billion first-phase investment with NEOM and later deepened its relationship with HUMAIN. Specialist developers can bring financing, construction and operating expertise that national AI platforms do not need to reproduce internally.
This matters because large AI campuses are infrastructure projects before they are software projects. Financing has to cover land, power, cooling, buildings and equipment long before the full revenue stream is visible.
What to watch next
The next stage of Gulf AI competition should be judged through financial close, construction progress, energisation and customer use. Sovereign backing can accelerate the first commitment, but durable economics still depend on utilisation. The region's strongest projects will be those that convert partnership announcements into financed and operating capacity.