The regional war with Iran has exposed structural vulnerabilities in the GCC’s AI infrastructure ambitions. The two core vulnerabilities: The physical exposure of data centers to regional conflict, and a less visible dependency on Saudi petrochemicals for the materials that go into computing hardware itself.

Neither vulnerability has been enough to break the core investment thesis — global investors, hyperscalers, and frontier model makers are still in on regional data centers worth more than USD 80 bn — but the case for the Gulf is no longer just about cheap power. (Some estimates have that figure as high as USD 97 bn.)

The cat among the pigeons: Iran’s drone strikes in February on Amazon Web Services (AWS) data centers in the UAE and Bahrain knocked out a spate of banking, transport, and food delivery apps and websites across the globe — including our own at EnterpriseAM. Beyond the immediate impact, these attacks also showed that the same geographic concentration that makes the Gulf cheap and fast to build in also makes it a single point of failure. Two and a half months later, the UAE data center is still down.

A less visible disruption made the same point from a different angle. A wartime shock at Saudi Arabia’s Jubail petrochemical complex tightened global supply of the high-purity resin used to manufacture printed circuit boards (PCBs) — the layered substrates that connect components inside servers, telecom systems, and computing hardware. Copper foil prices climbed roughly 30% and total input costs in some PCB segments rose as much as 40%.

Investors aren’t running for the hills: The UAE’s USD 46.1 bn data center pipeline and Saudi Arabia’s USD 35.3 bn pipeline are still moving forward, with co-investments from Microsoft, AWS, OpenAI, and xAI. Nothing else on the global AI map currently offers what the Gulf does — cheap power, sovereign capital at scale, and a regulatory environment that lets foreign firms enter while keeping their data local. The question is whether that bundle is enough to absorb the new risk layers the war has surfaced.

The structural drivers haven’t changed

A basket of advantages is bringing data centers to our part of the world. The big one is capital — Gulf sovereigns are investing heavily not just the big Western AI labs, but in domestic ecosystems centered around players like Humain (KSA) and G42 (UAE) as well as advanced research and development outfits like TII (UAE), which recently sold advanced cryptographic tech to Opaque that has found its way into the tech stacks of global majors including Anthropic.

Money aside, Gulf governments just … make it easy. It’s easier to get power, that power is cheaper, our grid infrastructure is reliable, buildout times are incredibly fast, and there’s no not-in-my-back-yard movement of the type that’s bedevilling facilities in the United States, Canada, and elsewhere. Contrast that with the Western world’s aging grid and expensive power, where the AI boom is colliding with a stagnant utility sector that cannot keep up with the projected 30 GWh of new load required by 2030.

Builders and regulators may be willing to move quickly here, but the latter are also starting to feel confident enough that they’re “layering in” — to use a now-popular AI-ism — sovereign data protections that appeal to high-security outfits including big financial players.

“Good news for data centers is that they don’t depend on global shipping,” David Bellman (LinkedIn), Vulcan Product Lead at SynMax, tells EnterpriseAM. “In fact, this issue makes power even cheaper inside the Gulf as [energy] products can’t get out,” effectively increasing domestic supply.

There’s been lots of foreign interest: “The US and Europe are looking to move into the region,” as the market matures from an emerging space, Mehmet Gonullu, managing partner at Dubai-based B2B tech advisory Yassi Ventures tells EnterpriseAM. “I can tell you, we’re seeing major startups and scale-ups in the US and European market, and they want to be here.”

With massive sovereign investment in local AI infrastructure combined with data sovereignty guarantees, foreign firms benefit from the buildout while maintaining full control of their data. That combination makes market entry “much, much easier,” Gonullu says.

Still, new fault lines emerged

#1- Physical concentration: The February drone strikes demonstrated that AI infrastructure clustered in a small number of locations is exposed in ways distributed cloud regions are not. The outages cascaded globally — a clear sign that the region’s geographic compactness, long sold as an efficiency advantage, has a downside when the geography is contested.

#2- The chemical supply chain that underpins computing hardware: This second fault line is less visible but potentially more structural. The Jubail disruption reduced output of high-purity resin at exactly the layer that enables advanced electronics manufacturing. With Sabic dominating global production, substitution is limited.

The pre-war market was already tightening thanks to surging AI demand. High-performance copper-clad laminates (CCLs), used in 5G infrastructure and AI servers, depend on the same resin systems, copper foil, and glass fiber inputs now under pressure. PCBs sit at the intersection of all of these, meaning the resin shock is just one expression of a broader structural squeeze. PCB costs have risen sharply this year, copper foil is up roughly 30%, and total input costs in some segments are up as much as 40%.

Taken together, the two fault lines threaten to change the equation. The case for the Gulf used to be about power, capital, and regulation. It now has to absorb two additional questions: How exposed is a given data center to a regional conflict, and how exposed is the broader hardware supply chain to disruptions in a single petrochemical corridor?

Investors are recalibrating, not retreating

The investment thesis that has brought OpenAi, Microsoft, and others to the region is largely unchanged, but if the war lingers, hyperscalers will start exploring other geographies offering stable grids, cheap power, and predictable regulatory environments.

“Unseasoned investors could be in panic mode,” Gonullu says. “I think this is just a temporary moment, and I’m sure the region will bounce back stronger after [the tension winds down], and investor confidence — even if it’s a little bit shaken — is going to return very quickly.”

A slowdown in data center buildouts would have ripple effects across the wider AI ecosystem. The GCC’s infrastructure pipeline has become a key driver of startup relocation from the US and Europe, where firms are drawn by capital, compute access, and sovereign data guarantees. Breaking that pipeline would mean losing momentum that took years to build.

Where each country stands

The competition inside the region is increasingly defined by infrastructure rather than models. As Gonullu puts it, the race is no longer about LLMs but about GPUs, data centers, and the power systems behind them. The UAE, Saudi Arabia, and Qatar are converging on the same objective — control of the AI infrastructure stack. Each has different levels of exposure to the risks now giving some pause.

The UAE remains in the lead. Abu Dhabi’s strategy is anchored in sovereign capital and hardware access: MGX is consolidating the nation’s AI investments, G42 has secured access to Nvidia Blackwell chips, Khazna is expanding large-scale capacity, and new liquid-cooled superclusters are emerging in Dubai Silicon Oasis.

Saudi Arabia is the hyper-scale challenger — leveraging energy dominance and US alignment. Humain is building national computing capacity, new semiconductor agreements have expanded chip access, and industrial localization is strengthening cooling and power systems manufacturing.

Qatar is probably a sovereign niche player. The Qatar Investment Authority, it is focusing on targeted investments in semiconductors and AI infrastructure, including partnerships with Brookfield and large-scale data center expansion via Ooredoo.

The system is moving from competition toward consolidation — but at a slower pace than capital deployment would suggest. “We need to start becoming the innovator, rather than the first adopter,” Gonullu says. “Right now we are bringing something to the market, localizing it, and customising it to regional specs — but there is an evolution toward deeper innovation.”