The Gulf is betting billions on becoming the world’s AI engine

AI economics are beginning to look like those of the energy sector. Both require massive infrastructure, reliable supply, and long-term investments. Countries that control these resources can influence far beyond their own borders.

Saudi Arabia and the UAE are investing in data centers, power, and connectivity to support domestic AI adoption and serve international customers. The strategy draws on familiar strengths: access to energy, sovereign capital, and the ability to coordinate large infrastructure projects.

But computing capacity behaves differently from an oil reserve. New generations of processors can change the economics of existing hardware. Buildings and power connections may remain useful for decades. The equipment inside them needs a much shorter investment cycle.

For the Gulf, the challenge is to turn a construction boom into a durable business.

COMPUTE AS A PUBLIC UTILITY

Andrew Jackson, Chief AI Officer at B Capital, believes the Gulf will eventually have to reach the point where it guarantees public access to compute, as it does for essential utilities. The Gulf has already taken that step, in fact, having invested sovereign capital in computing at gigawatt scale. “Compute built at gigawatt scale with sovereign capital is a national capability rather than a product and national capabilities have an obligation of access.”

He expects this to result in guaranteed baseline capacity for public services and education, preferential or guided terms for domestic businesses, and market rates for international customers. The actual framework will take time to work out. The fundamental principle Jackson advocates is that “the benefits of intelligence as a utility must go to the population.”

Abdulaziz Omar Almubarak, VP of Business Development at DataVolt, takes a rather different view. He definitely sees access to compute as a key driver of economic competitiveness. He says governments should focus on creating the conditions that allow broad access to compute rather than directly regulating access to it. That would mean supporting R&D and local innovation, as well as investing in the infrastructure AI needs, such as power, land, connectivity, and faster permitting.

“The countries best placed for long-term success will be those whose governments are in a position to support both the local innovation ecosystem and the fundamental core infrastructure,” he says.

In many places, power availability is a bigger challenge than chip shortages. Data center projects are delayed for years waiting for grid connections, even though the necessary processors are already available.

The Gulf is in a better position because it’s building power generation in parallel with the data centers and can draw on solar, nuclear, and gas. That doesn’t mean power is a solved problem, though. Cooling in the region’s hot climate adds to costs, water remains scarce, and demand for electricity is growing faster than the power sector has historically been able to keep up.

“Chip supply is the constraint that changes year to year. Energy is the constraint that changes decade to decade,” Jackson says. He thinks the region is wise to invest in power infrastructure before demand rises, since these projects take a long time.

Almubarak also believes reliable power is key to expanding AI infrastructure and thinks attention has been overly focused on chips. In Saudi Arabia, he doesn’t expect power availability to be a fundamental constraint. The kingdom’s long-term energy strategy, investments in power infrastructure, and renewable energy expansion should enable it to provide ample power at a competitive cost.

“Power, connectivity and compute must be planned together from day one,” Almubarak says, adding that all three need to develop in parallel to build a successful AI ecosystem.

WHO GETS ACCESS 

AI sovereignty does not require a country to build every component of the technology stack domestically. Rather, it comes down to controlling where data is stored, which laws govern its processing, who has access to it, and whether sufficient computing capacity is available to serve its citizens when needed. By these measures, Jackson says, the Gulf is making tangible progress.

He views the region’s approach as forming partnerships to access advanced chips and leading models, while also building indigenous capability in sectors like energy, investment and finance, and healthcare. Scaling up deployments and building local capability over time also provide an edge.

“Dependency that is diversified and contractually secured is not vulnerability,” Jackson says. The most capable AI systems are being developed through international partnerships, he adds, and the Gulf is emerging as a trusted partner in those.

Almubarak approaches sovereignty from the infrastructure layer. For him, the practical questions are where data is stored, who operates the facilities, and which country’s laws apply.

Genuine AI and data sovereignty means having foundational infrastructure inside the country, protected by national laws, and operated by its people. This involves building power systems, cooling, data centers, and a skilled local workforce to support AI.

Almubarak points to Saudi Arabia’s creation of Humain as part of a broader effort to develop sovereign AI capabilities across the technology stack. DataVolt’s contribution is the AI-ready infrastructure it is developing at Oxagon in NEOM, in partnership with Humain. The data centers are one part of the kingdom’s wider effort to establish greater control over its AI capabilities.

“No country owns the full stack,” Jackson says. The Gulf is building its sovereignty by deciding where the data is stored, under what laws it is processed, and who can access it.

‘True AI and data sovereignty simply won’t happen unless the foundational infrastructure is in the country, protected by the country’s laws and operated by the country’s own people,’ Almubarak adds.

WHO WILL BUY THE CAPACITY?

Jackson identifies three customer segments for the Gulf’s nascent AI infrastructure market. The first group is the frontier AI labs and hyperscalers who can’t build enough compute capacity at home and are turning to the Gulf.

The second group includes governments and corporates in the Middle East, Europe, Africa, and South Asia that want reliable, low-latency AI services with their data kept in the region. The third is the Gulf’s own economy, as more local businesses and public sector organizations start using AI.

Jackson considers the second group the most important because it makes AI infrastructure a driver of regional economic growth rather than just an export business. He does, however, warn against comparing AI too closely with oil.

“Compute is perishable; it loses value with every generation of hardware and it’s dependent on a very small number of large buyers,” he says. To mitigate those risks, the Gulf needs long-term contracts with key customers, a broad base of regional customers, and investment in AI applications where the value is more durable than in raw compute capacity.

For Almubarak, Saudi Arabia will focus first on its own market. The kingdom’s immediate priority is to support local AI adoption, innovation, and economic growth before targeting international demand from hyperscalers, cloud providers, enterprises, and AI-native companies.

He lists connectivity as one of Saudi Arabia’s key competitive advantages. The mix of subsea cable and terrestrial links between Europe and Asia provides alternative paths for international data traffic, enhancing the network’s resilience and reliability and potentially reducing latency.

The kingdom’s terrestrial cable linking Europe to East Asia is important because it provides an alternative to the Red Sea routes. This could benefit inference workloads, real-time applications, and cross-border cloud services, all of which rely on a reliable, low-latency connection.

Almubarak points out that, with its access to power and land, Saudi Arabia is well positioned to become the interconnector between Europe, Asia, and Africa.

DESIGNING FOR CHANGE

Almubarak says that to keep AI infrastructure relevant over time, it’s better to design for flexibility rather than commit to one generation of technology. Computing platforms will continue to evolve, but foundational infrastructure such as power, cooling and connectivity can be built to support different hardware over several decades.

Jackson believes infrastructure builders simply can’t reliably predict which technologies will lead in five or six years. So, infrastructure should be built to handle future hardware and compute requirements.

“Nobody can design a data center in 2026 and know what will be running in it in 2032, so you design for change rather than for a forecast,” he says.

Jackson points to modular data center designs, liquid cooling, higher rack densities, and investing in software and operations as key elements of this strategy. He also says it’s important to have a broad mix of customers from government, enterprise, research, and technology partners, not just one AI model or provider.

“AI infrastructure is a program, not a project,” says Jackson. The region’s long-term advantage will be its ability to keep investing and evolving as technology develops.