When Iran struck two Amazon Web Services (AWS) data centers in the United Arab Emirates and damaged a third in Bahrain in early March, Washington quickly deemed the Gulf’s artificial intelligence ambitions a catastrophic strategic miscalculation. Gulf states had made a multihundred-billion-dollar bet on AI infrastructure in a volatile, risky region.
The proposed solution was to relocate AI workloads that operate out of the Gulf to safer locations. This rested on two beliefs: The Gulf is too unstable for critical infrastructure, and there is no playbook for building AI infrastructure in such contested territory.
Both assumptions deserve examination, and the second is not true. A playbook already exists, written under far heavier fire, in Europe. Since the Russian invasion of Ukraine, Kyiv has kept its digital infrastructure running through sustained bombardment. It moved terabytes of government data across borders in a matter of weeks to keep government services functioning and the state digitally intact, rebuilt its power grid and broader architecture around the assumption of continued Russian aerial attacks, and treated survivability as the core design principle of its critical infrastructure and, by extension, of its own survival and ultimate victory in the war.
The comparison is imperfect, and the differences in geography and war intensity matter. But Ukraine remains the only state to have stress-tested its digital infrastructure in a full-scale war with a bigger enemy, and its lessons map onto the Gulf’s AI buildup after the Iran war. The question facing Gulf capitals is how they can adapt Kyiv’s wartime playbook to their own threat environment.
The case for relocating the Gulf’s compute ambitions rests on a fundamental misreading of why Gulf states are dedicating enormous resources to AI. The core belief in Abu Dhabi, Riyadh, Doha, and other Gulf capitals is clear: They cannot miss the next economic transformation that will reshape geopolitical powers and their countries’ place in the global order. They missed the Industrial Revolution, and when the post-World War II energy-dependent global order enriched them financially, the gains lacked strategic depth, leaving their political economy tied to the oil boom. AI is seen differently, almost existentially, for Gulf states.
It is therefore unsurprising that the Iranian attacks haven’t deter the Gulf’s compute buildout. Instead, the war has reframed the question entirely. It is no longer about whether to build—the Gulf will keep building compute, with no sign of changing course—but about how to build compute infrastructure in a contested environment prone to sharp shifts in security and military conditions. This turns on three factors.
First, given the political commitment to AI, the capital financing the Gulf’s AI buildout is not a case of venture capital chasing quarterly, or even yearly, returns. The UAE, Saudi Arabia, and Qatar are sovereign states financing their post-hydrocarbon future. They are not venture investors who flee when the war starts.
The Gulf might look into South Korea’s industrialization at breakneck speed, precisely when U.S. security guarantees appeared to waver in the early 1970s. Taiwan built the world’s most strategically indispensable semiconductor cluster under five decades of the threat of invasion. Strategic pressure concentrates national investment on priority fields, rather than leading to hasty withdrawals at the first sign of trouble.
Second, as the binding constraint on AI shifts to the energy needed to sustain the workload, the Gulf’s energy abundance and cheaper production costs make it a natural home for AI infrastructure. Frontier AI is, fundamentally, a function of electricity—reliable, affordable access to sufficient power underlies every state’s AI ambitions. To the point, the Gulf possesses the cheapest and most abundant power on the planet. Iranian attacks might disrupt external access to energy or raise costs, but they do not change the fundamental reality that the Gulf’s energy profile is unmatched. Beyond crude oil, the Gulf is undergoing a major pivot toward energy diversification: nuclear, renewables, and natural gas to provide the reliable, affordable power required for training and deploying large language models at scale. As AI economics shift toward intelligence per dollar, the Gulf’s cost advantage in energy production becomes critical. Inference, the compute-intensive but lower-margin phase of AI deployment, requires massive scale and cheap power, making the Gulf a natural hub for inference workloads.
Finally, geography adds a third factor. AI is evolving from training models to serving them at scale, with inference delivered to billions of users with minimal latency. Proximity becomes the product itself. A data center in Virginia or Frankfurt cannot serve South Asia, East Africa, or the Eastern Mediterranean without introducing delays that degrade user experience or introduce cross-border data complications. And maybe this doesn’t matter with text models, but it will matter most for physical AI. The Gulf sits within inference-grade latency of these markets. A latency advantage of this kind cannot be relocated.
The buildout in the Gulf will therefore continue. Gulf leaders have said as much, repeatedly and on the record. The honest task before Gulf policymakers and their U.S. hyperscaler partners is to ensure that AI infrastructure gets built and can survive as this “no war, no peace” situation drags on with no end in sight. In fact, even if it ends, the Iranian threat will always shadow the AI buildup. On that question, one country has more hard-won experience than any other, and that’s Ukraine.
When Russia invaded Ukraine in February 2022, the nation’s digital infrastructure lived where most states keep theirs: in large, centralized, on-premises facilities. This architecture is efficient in peacetime and catastrophically fragile in war. Within weeks, Kyiv tore it apart and improvised a new one under fire. Two moves defined the transformation, and each offers a template for the Gulf.
The first was emergency cloud migration. In the war’s opening days, Ukraine’s parliament passed emergency legislation amending its data protection laws to permit government data to leave national territory. Over roughly three months, working with AWS, the government migrated some 15 petabytes of data from more than 50 institutions onto foreign cloud servers beyond the reach of Russian missiles. The state’s registries, tax records, and critical services were effectively backed up across borders. A Russian warhead could destroy a building in Kyiv; it could no longer destroy the Ukrainian state’s memory.
On energy, Russia made Ukraine’s power grid a primary target, which meant every data facility had to assume grid failure. New Ukrainian infrastructure is built around redundant, decentralized power: dedicated solar arrays paired with battery storage and plans for small modular reactors to guarantee uptime through extended blackouts.
The Gulf should transfer Ukraine’s innovations to its own infrastructure build-outs, adapting them to the region’s own needs and creating a new standard for compute security with the Gulf’s mark for the years to come. It’s no surprise, then, that Ukraine showed up in the UAE, Saudi Arabia, and Qatar after the Iran war began. Ukrainian President Volodymyr Zelensky toured the Gulf in March, selling Ukraine’s drone and missile defense expertise as Iranian attacks hit the region.
Still, the Gulf is not Ukraine. It is far wealthier, faces a different adversary, and is building infrastructure two orders of magnitude larger: gigawatt-scale campuses rather than wartime capacity. But the design principles translate directly, and in several respects, the Gulf can implement them better than Ukraine ever could.
The current announced road map consists of 10 to 12 gigawatts of capacity spread across the UAE, Saudi Arabia, and Qatar. These gigawatts aren’t concentrated in one place; they’re divided across three countries. And by virtue of the region’s geography, they aren’t repeating Ukraine’s prewar situation: clustering strategic capacity in target-rich zones.
The first task in wartime is hardening existing infrastructure. In practical terms, compute that serves sovereign, military, and security workloads should be hardened in its construction or built underground, much like other sensitive state assets shielded that way to avoid aerial targeting: placed below ground, away from the coast, and beyond easy reach of low-cost drones and cruise missiles. Concretely, that means battle-tested hardening across the entire campus footprint: earth berms and reinforced concrete to defeat blast, fragmentation, and drone-borne munitions; fully buried power substations, transformers, fuel storage, and cooling loops; cleared surrounding terrain to deny cover to low-flying drones and improve defensive fields of fire; and modular, compartmentalized designs with on-site generation, redundant microgrids, and automated failover so a facility can absorb a partial strike and keep running at reduced capacity rather than collapsing entirely.
Energy resilience, moreover, must become nonnegotiable. The Gulf’s cheap, abundant power is its core competitive advantage in the AI infrastructure era, but the war has demonstrated that energy infrastructure itself is a target. Ukraine’s response to heavy Russian bombardment through co-locating solar with storage, with modular nuclear on the horizon, maps naturally onto the Gulf, which already leads the region in utility-scale solar and civil nuclear power. Each major campus should be capable of operating independently of the national grid for extended periods. Gulf states should rethink their understanding of energy resilience as a form of air defense by other means.
The Iran war should force a critical rethinking. Gulf states have spent years building data sovereignty regimes that run counter to cloud deployment. These Gulf capitals need not abandon sovereignty, but they should build the wartime exception in advance. Pre-negotiated frameworks with hyperscalers should specify which workloads can surge to facilities abroad, under what controls, and on whose authority, executable in hours rather than months. The continuity of the digital state should never depend on emergency legislation drafted under fire.
Finally, doctrine must evolve. The March strikes confirmed that data centers have joined refineries, ports, and desalination plants as infrastructure that adversaries will deliberately target. Gulf countries’ air and missile defenses performed credibly in the war’s opening months, intercepting the overwhelming majority of incoming ballistic missiles and drones. But that architecture was designed to protect cities and military bases, not gigawatt-scale compute clusters. Whether through dedicated point defenses tied to major campuses, prioritization within layered national systems, or both, compute must now be defended with the same seriousness as energy infrastructure. The costs of that defense must be priced into project economics from day one.
Iran’s strikes on Gulf data centers carried an identical message to Russia’s on Ukraine: This region is too dangerous for the compute capacity you are building. The Gulf’s answer should mirror Kyiv’s response—not retreat but a redesign of the buildup.
The Ukraine war gave us clear clues on what infrastructure redesign requires: hardened and dispersed facilities, resilient energy systems, flexible cross-border data legislation, and air defense systems, all cost-priced into project finances from the start. None of this is cheap, and none of it featured in the Gulf states’ gigawatt-scale original vision. The Gulf’s positioning as one of the main global compute nodes will be decided not by how many gigawatts it announces but by how many are actually built during this long Iran war and, most importantly, still running the day after Iran’s next attack.
