NextFin News - The Persian Gulf's richest states are pressing ahead with a build-out of artificial intelligence infrastructure worth hundreds of billions of dollars even as data centers have become repeat targets in the US war with Iran. State-backed firms in the United Arab Emirates have continued work on a sprawling AI campus in Abu Dhabi designed to add five gigawatts of computing capacity, after adding worker-protection measures at the peak of the fighting, people familiar with the matter said. The region is betting that sovereign money, American chips and hardened concrete can outlast whatever its adversaries can throw at its servers.
The Damage Is Real, but the Pipeline Is an Order of Magnitude Bigger
The contradiction could hardly be starker. In early March, Iranian drones and missiles struck three Amazon Web Services data centers — two in the UAE and one in Bahrain — forcing the facilities offline and disrupting banking, payments, food delivery and enterprise software across the Gulf. Careem, the delivery and taxi platform, payments companies Alaan and Hubpay, and enterprise software provider Snowflake were all hit with service disruptions, a visible reminder that cloud infrastructure is not an abstract concept when it goes dark.
AWS described the damage in a statement released as the strikes were still unfolding:
"In the UAE, two of our facilities were directly struck, while in Bahrain, a drone strike in close proximity to one of our facilities caused physical impacts to our infrastructure. These strikes have caused structural damage, disrupted power delivery to our infrastructure, and in some cases required fire suppression activities that resulted in additional water damage."
Within days, Iran's Islamic Revolutionary Guard Corps-affiliated outlets published a list of 29 "legitimate" technology targets across Bahrain, Israel, Qatar and the UAE, naming facilities operated by Amazon, Microsoft, Google, Oracle, Nvidia, IBM and Palantir. The list read like a who's who of the Gulf's AI ambitions — and it was not a one-off threat. A second disruption hit AWS's Bahrain region later in March, forcing the company to repeat its advice that customers migrate workloads out of the region.
Yet the concrete keeps getting poured. G42 and its consortium partners — OpenAI, Oracle, Nvidia, SoftBank and Cisco — are building Stargate UAE, a 1-gigawatt compute cluster inside a 5-gigawatt, 10-square-mile Abu Dhabi campus that its builders describe as the largest such deployment outside the United States. The first 200-megawatt AI cluster is expected to go live in 2026. Microsoft is adding 200 megawatts of capacity through G42's Khazna Data Centers before the end of 2026, part of a $15.2 billion UAE commitment that includes roughly $7.3 billion deployed between 2023 and 2025 and another $7.9 billion planned through 2029. Saudi Arabia, through the Public Investment Fund-owned Humain, has started construction on two campuses comprising 11 data centers of 200 megawatts each, and has signed $23 billion of technology agreements with AWS, xAI and Nvidia.
The question this build-out forces is not whether the Gulf wants to be an AI superpower. It is whether a region that has just become a shooting gallery for cloud infrastructure can finance, insure and secure its way to that status — and whether the war has changed the economics enough to slow the rush. So far, the answer from the ground is no.
The Real Cost Is Not Concrete — It Is Insurance, Security and the Price of Capital
The second-order effect is where the war actually bites. Physical damage is a one-time repair bill that a sovereign balance sheet can absorb. The enduring cost is a permanent repricing of risk that compounds on every new project. Analysts estimate a "security premium" for physical hardening — anti-drone systems, reinforced structures, redundant power and cooling — will add 15% to 20% to the capital cost of new projects, on top of construction inflation already projected at 6% to 8% for 2026. Across a build-out measured in the hundreds of billions of dollars, a 15% premium is not a rounding error; it is tens of billions of capital that must be found, priced and deployed before a single server turns on.
Insurance markets are being forced to confront a question they have never had to answer. The war-exclusion clause, according to specialty brokers, has never been invoked on a claim reported under a cyber or cloud-infrastructure policy. The AWS strikes may become the precedent-setting case: if insurers pay, premiums for Gulf cloud risk reset higher across the board; if they deny, developers lose the coverage they need to finance construction at all. Either outcome raises the cost of capital for the region's AI build-out, and it raises it permanently, because the risk has been demonstrated rather than merely modeled.
That premium does not fall evenly, and that is the mechanism that reshapes the industry. Sovereign-backed projects with state balance sheets behind them — G42 in Abu Dhabi, Humain in Riyadh, Qatar Investment Authority vehicles — can absorb a 20% cost overrun that would kill a speculative private developer. Brookfield Asset Management's $20 billion data center deal with the Qatar Investment Authority is proceeding, and AWS is moving forward with its planned $5.3 billion region in Saudi Arabia, precisely because the counterparties can self-insure. The war, in other words, is not stopping the Gulf data center boom; it is consolidating it into the hands of states that do not need to buy insurance at market rates.
Power Is the Region's Weapon — and Its Vulnerability
Beneath the security debate sits a simpler advantage that the war has not touched. AI data centers are, above all, power plants with servers attached, and the Gulf has some of the cheapest and most abundant energy on earth. The Abu Dhabi campus is designed to be powered by nuclear, solar and natural gas — an energy mix that lets its builders promise low-carbon compute at a cost per kilowatt-hour that Phoenix or Finland cannot match. That is why the hyperscalers signed up in the first place: not because the Gulf is safe, but because it is cheap, and because the chips they are allowed to deploy there are the same frontier systems their competitors cannot buy in China.
But power is also the vulnerability that keeps risk managers awake. A data center without electricity is scrap metal, and the Gulf's grids, while robust, are concentrated and exposed. The March strikes disrupted power delivery to AWS's infrastructure — not through a direct hit on a turbine, but through damage to the facilities themselves. The fix is redundant generation and microgrids, which is exactly what the sovereign builders are adding to their designs. It is also exactly what adds another layer to that 15% to 20% premium.
Cyclical Shock or Structural Shift? The Structural Force Is Winning
This is the judgment that determines the conclusion. The attacks are a cyclical shock layered on top of a structural shift — and the structural force is winning.
The cyclical leg is real and it is painful: a conflict-driven spike in security costs, a temporary insurance repricing, and a near-term scramble for redundancy that is pushing some workloads to Europe and the United States. History offers a template. After the 2019 drone strikes on Saudi Aramco's Abqaiq processing facility — which knocked out about 5.7 million barrels of oil a day, roughly 5% of global production — the kingdom hardened its energy infrastructure, bought air defenses and diversified its export routes. It did not abandon oil production. The shock raised the cost of doing business; it did not change the business. The Gulf is running the same playbook with compute.
The structural leg is stronger, and it is what separates this cycle from 2019. The Gulf's data center push is not a speculative bet on a commodity price. It is a state-led repositioning of national economies away from hydrocarbons, backed by sovereign wealth funds that collectively manage trillions of dollars, anchored by long-term offtake from US technology giants, and wired into export controls that deliberately lock advanced semiconductors inside a US-aligned security perimeter. The Pax Silica framework, expanded to Qatar and the UAE in January 2026, formally brought both countries into a US-led coalition on semiconductor and AI supply chains alongside the UK, Japan, South Korea, Singapore, Australia and Israel. Gulf compute is now a geopolitical asset, not just a commercial one — a sanctioned, chip-controlled, US-aligned infrastructure class whose value depends on the credibility of the perimeter around it.
A regime shift of that kind does not reverse because a few facilities get hit. It reverses only if the security guarantee itself breaks, or if the export-control architecture that makes Gulf compute scarce and valuable collapses. Neither is on the table.
The Counter-Thesis: Proximity to Iran Is a Structural, Not Cyclical, Disadvantage
The strongest argument against the above is that geography is destiny, and data centers are footloose in a way oil fields are not. Unlike Abqaiq, which is hard to replace and globally indispensable, a hyperscaler deciding where to locate its next gigawatt can choose Phoenix, Finland or Singapore as easily as Abu Dhabi. If Iran can reliably reach Gulf infrastructure with drones and missiles, the risk is not a one-time premium — it is a permanent locational discount that compounds with every escalation. The 29-target list is not a threat that expires with a ceasefire; it is a published targeting database that any future adversary can reuse, and it names not just facilities but the companies that operate them.
There is evidence this is already being priced at the margin. AWS's $5.3 billion Saudi region is proceeding, but after the March strikes the company told customers with workloads running in the Middle East to migrate critical data and workloads to alternate regions, recommending facilities in the United States, Europe or Asia Pacific as latency and data-residency rules allowed. A builder that advises its own customers to leave the region is not behaving like an investor with unlimited conviction. If missile and drone accuracy keeps improving, and if Iran or its proxies demonstrate the ability to penetrate hardened facilities repeatedly, the sovereign builders could find themselves owning increasingly stranded, over-insured capacity while hyperscalers diversify elsewhere. The region's AI dream, as some investors have called it, would then be measured not in gigawatts built but in gigawatts stranded.
The rebuttal rests on what capital is actually doing, not what it is saying. So far, diversification is happening inside the region, not out of it. Microsoft's 200-megawatt expansion, G42's campus, the Qatar-Brookfield deal and the Saudi sovereign pipelines are all proceeding on schedule. The market is demanding a higher return and more security, not fleeing. The falsifying signal is specific and observable: if two or more hyperscale operators publicly defer or cancel announced Gulf projects within a 12-month window after a confirmed strike on a hardened facility, the "structural shift" thesis is wrong and the locational-discount thesis takes over. Until that prints, the burden of proof sits with the skeptics.
Who Wins, Who Is Exposed, and What Breaks the Thesis
The winners of this war economy are the sovereign builders and the security, power and engineering suppliers that serve them — the firms that can turn a 15% to 20% risk premium into a recurring revenue stream. State-backed operators gain share by default as private capital prices out, and the region's hyperscale partners gain a politically protected foothold in the world's fastest-growing AI infrastructure market. The exposed are the pure-play commercial colocation developers without sovereign balance sheets, the insurers who wrote Gulf cloud policies on pre-war models, and any project whose economics depended on 2025's risk-free capital.
What to watch, by horizon:
- Short term (next six months): whether the AWS war-exclusion claims get paid and how quickly insurers reprice Gulf cloud risk. A denied or delayed payout would freeze private capital faster than any missile.
- Medium term (through 2027): whether announced projects — the 5-gigawatt Abu Dhabi campus, Humain's 11 facilities, AWS's Saudi region, Brookfield's Qatar deal — actually reach commercial operation on schedule. Announcements are cheap; energized racks are not.
- Long term: whether the US security guarantee holds. Gulf compute is now a chip-controlled asset class; its value depends on the credibility of the perimeter around it.
Scenarios. Base case: the build-out continues with a permanently higher cost of capital and security overhead, concentrated in sovereign hands, and the Gulf captures a larger share of global AI infrastructure than anyone modeled in 2025. Upside case: a ceasefire or sustained de-escalation compresses the risk premium, insurance markets normalize, and the region's cheap power and sovereign capital make it the default location for the next wave of AI capacity. Downside case: repeated successful strikes on hardened facilities trigger hyperscaler cancellations, war-exclusion claims are denied, and the region's AI capacity becomes a stranded, state-owned white elephant.
The closing judgment: The Gulf is not betting that war won't touch its data centers. It is betting that state money, American chips and hardened concrete can outlast whatever Iran can throw at them — and so far, the concrete is winning.
Explore more exclusive insights at nextfin.ai.
