Washington’s recent easing of restrictions on advanced AI-chip exports to the United Arab Emirates points to a larger economic question. As artificial intelligence becomes more power-intensive, the geography of energy is becoming part of the geography of computing. That creates an opportunity for the Gulf — and a strategic choice for the United States.

For decades, the Gulf’s economic challenge has been framed as diversification away from energy. AI suggests another path: not abandoning the region’s comparative advantage, but extracting more value from it before the energy leaves the economy.

Cloud computing made the digital economy appear almost weightless. AI is exposing its physical layer again. Advanced computing requires chips, data centers, cooling and vast amounts of electricity. McKinsey expects AI-related data-center demand to rise sharply through 2030, making power increasingly important to where capacity is built.

For Gulf economies, this creates a new choice: export energy as a commodity, or use it domestically to produce electricity, then computing power, and ultimately a digital service sold abroad.

The energy stays; the output travels.

A recent World Bank paper examined whether energy-rich economies could turn cheap electricity into exports of AI computing power. Its warning matters: cheap energy alone is not enough. Hardware costs, regulation, trust and trade barriers can erase an electricity advantage. McKinsey adds that connectivity, advanced chips, customers and execution also matter.

Even if the Gulf attracts massive AI infrastructure, another question remains: How much of the resulting value stays in the Gulf?

If foreign companies supply the chips, own the models, operate the cloud and sell the applications, while Gulf economies provide land, capital and electricity, the investment can create jobs and deepen markets. Yet it may also resemble a sophisticated new form of energy export. The commodity no longer leaves in a tanker; much of the value created from it leaves through fiber.

Nor does moving up the value chain make every compute project worthwhile. Gulf governments should compare the domestic value created by energy used for computing with its best alternative use — including export — after accounting for capital costs, grid demands and public priorities.

That opportunity cost also has an international dimension. Gulf oil exports flow largely to Asia, especially China and India. Where additional compute power is generated from exportable hydrocarbons and production remains fixed, more domestic consumption means less energy available elsewhere. Compute can therefore change not only what the Gulf exports, but where part of its energy creates value.

Who supplies the technology then matters economically, not just politically. If Gulf compute runs mainly on Chinese technology, a greater share of the technological and ecosystem value associated with that energy could accrue to Chinese firms. If it runs on American technology, more of that energy advantage would be converted into digital output through an American-centered ecosystem — without the United States importing the energy itself.

AI investments should therefore be judged by what remains locally: skilled employment, a larger private sector, local suppliers, public revenue, technological capability and returns on capital. Citizens and essential public services should retain priority.

The useful analogy is not that “AI is the new oil.” It is downstream development.

Gulf producers learned to capture more value by turning crude into refined products and petrochemicals before export. AI offers another chain: energy becomes electricity; electricity becomes computing power; computing power becomes services, applications and new industries. Diversification need not mean abandoning an existing advantage; it can mean moving further up its value chain.

This is where Washington and the Gulf have complementary interests. The United States has leading AI technology. Gulf states have abundant energy, capital and infrastructure capacity. Geopolitics complicates that logic. Washington reasonably worries that advanced technology could strengthen strategic competitors, particularly China. Security concerns should shape safeguards, but not obscure the economics: demand for power-intensive computing exists independently of U.S. policy.

American leadership gives Washington influence over global AI infrastructure, but not permanent exclusivity. If trusted Gulf partners cannot find a workable route into the American ecosystem, their incentive to build capacity will not disappear. As alternatives improve, restrictions may increasingly determine who supplies the infrastructure rather than whether it gets built.

A better policy would recognize the complementarity directly. Washington should preserve a workable path to advanced American technology for trusted Gulf partners, with appropriate safeguards. Gulf governments, in return, should judge AI investments not by data centers announced but by how much local value they create above the electricity layer — and whether that value exceeds the opportunity cost of the energy consumed.

For the Gulf, success should not be measured in megawatts consumed. For Washington, it should not be measured merely in chips withheld. The better measure is how much shared economic value both sides can create before alternative partnerships become more attractive.