The Gulf Model: Capital, Power, and a Trust Deficit

The Gulf states, principally Saudi Arabia and the United Arab Emirates, have mounted the most capital-intensive and headline-grabbing sovereign compute push on Earth, and their programs are the clearest illustration of both the promise and the limits of the site-layer sovereignty this course keeps describing.

Start with the endowment. The Gulf's advantages are almost perfectly matched to what large-scale AI compute requires. It has sovereign wealth on a scale few institutions anywhere can match, funneled through vehicles like Saudi Arabia's Public Investment Fund and Abu Dhabi's sprawling investment entities. It has cheap and abundant energy, the single largest operating cost of a data center. It has ample land and a state capable of moving fast when the leadership decides to. And it has an explicit strategic motive: diversifying economies away from hydrocarbons by capturing a share of the AI value chain, and buying geopolitical relevance in a technology the whole world now takes seriously.

Out of this came two anchor efforts. In Saudi Arabia, the state launched a dedicated AI company — created under the Public Investment Fund and publicly unveiled in 2025 — to build data centers, develop Arabic-language models, and act as the national champion for AI infrastructure. In the UAE, the Abu Dhabi-based firm associated with the emirate's technology strategy became the vehicle for enormous data-center ambitions, including a headline plan for a vast multi-gigawatt AI campus in Abu Dhabi and partnerships reaching into American technology companies. The scale of announced capacity in both cases is staggering by any historical measure of data-center construction.

But recall the four layers. The Gulf owns the site layer outright — land, power, buildings, capital. What it does not own is the silicon or the deepest parts of the stack, and it cannot manufacture either. That single fact defines the entire strategic problem: the Gulf can build the most magnificent data centers in the world and still be unable to fill them without American chips that require American licensing. Capital cannot substitute for a chip you are not permitted to buy.

This is why the Gulf's real work over the past few years has been diplomatic as much as industrial. Sitting in the middle tier of the export-control regime, and carrying a specific trust deficit with Washington rooted in longstanding technology and security ties to China, the Gulf states had to actively restructure relationships to qualify for top-tier chip access at the volumes they wanted. That has meant distancing certain Chinese vendors from sensitive infrastructure, adopting security arrangements acceptable to American regulators, and drawing major American technology firms into their projects as partners and stakeholders. In effect, the Gulf agreed to bring its compute buildout inside the American security perimeter in exchange for the chips to fill it.

Read through the course's framework, the Gulf model is negotiated dependency executed with maximum resources. The nations buy commanding control at the site layer and accept continued dependence at the silicon and stack layers, closing the trust deficit through concessions that reduce Washington's concern about diversion and Chinese access. It is a coherent and, on its own terms, largely successful strategy: it delivers real capacity, real economic diversification, and a genuine seat at the AI table. What it does not deliver is insulation from the chokepoint. If the terms of American access changed sharply, the Gulf's gleaming campuses would still depend on decisions made elsewhere.

There is a capital dimension the course flags as well. The sums involved are so large that the Gulf has become not just a builder of its own capacity but a major financier of the global AI infrastructure boom, taking stakes in foreign firms and projects and effectively exporting capital into the compute economy. This makes the Gulf a swing source of financing for the whole sector, which in turn gives it leverage: capital that large is courted, and being courted is itself a form of soft power that partially offsets the hard dependency on foreign chips. The Gulf cannot make an accelerator, but it can make itself indispensable to those who can, and that is a real, if different, kind of strategic position.

The lesson to carry forward is that the Gulf demonstrates both the reach and the ceiling of money. With enough capital and power you can own everything about a compute facility except the two things that matter most for true sovereignty — the silicon and the stack — and closing even the access gap requires political concessions, not just checks. The Gulf bought the most sovereignty money can buy, and its programs map precisely where that limit sits.