The Iran war has created an instructive paradox: it is accelerating military AI adoption even as it damages the material systems that make AI possible. On March 1, Iranian drone strikes against data centres in the UAE and Bahrain, along with the closure of the Strait of Hormuz, revealed how exposed digital infrastructure has become to regional conflict. This is the first major stress test of AI infrastructure’s geopolitical vulnerability. It reveals the extent to which computational power, model training and defence innovation depend on fragile supply chains that span the Gulf, East Asia and China-controlled material chokepoints.
That vulnerability is severe because AI is not an immaterial technology. Its performance depends on helium for chip fabrication, bromine for photoresist chemistry, tungsten for tooling and munitions, advanced semiconductor components and uninterrupted flows of energy resources such as liquefied natural gas and crude oil. The conflict has therefore damaged both the demand and supply sides of the AI economy. Defence agencies are turning more rapidly to algorithmic systems, while manufacturers face rising energy costs, logistics paralysis and shortages of essential inputs. Asia, which entered 2026 with a formidable lead in chip fabrication and AI deployment, now faces a structural shock that may delay its climb.
The war’s most immediate effect has been to reassert the importance of material chokepoints. March’s missile strikes on Qatar’s Ras Laffan complex removed about a third of the semiconductor-grade helium supply from the global market, producing a sharp price shock. Repairs are expected to take up to five years, in part because the global turbine shortages needed for replacement infrastructure already constrain maintenance schedules. Helium matters because it is indispensable in extreme ultraviolet lithography and other advanced chip processes used by leading-edge foundries such as TSMC and Samsung. Without it, production of 5nm-class and more advanced nodes becomes expensive and uncertain.
The Strait of Hormuz compounds the problem. Its effective closure has transformed a regional conflict into a global energy shock, rapidly inflating the energy bills of semiconductor plants, data centres and industrial customers across East Asia. The impact is especially severe in South Korea, where roughly 70 percent of oil imports are tied to West Asia.
At the same time, Israel and Jordan, which together account for a significant share of global bromine production, have seen their output chains disrupted, threatening photoresist chemistry that is fundamental to chip manufacturing. China’s tungsten export controls, already associated with a dramatic price surge, further intensify the crisis because tungsten carbide is essential for semiconductor tooling. The result is a layered disruption affecting fabrication, logistics, energy and downstream AI deployment simultaneously.
These material disruptions strike at a moment when Asia had become the principal engine of AI trade expansion. By one estimate, Asia contributed nearly two-thirds of global AI trade growth in the first half of 2025. In that context, the crisis, aside from raising prices, also challenges the strategic assumption that Asia is a stable place to build the physical backbone of AI.
Delays at ports, rerouting requirements and the rising cost of insurance and transport all introduce friction into supply chains that were already optimised for speed rather than resilience. That friction matters most for high-bandwidth memory (D-RAM) production in South Korea, where the combination of energy volatility and helium shortages threatens output and reliability.
AI training costs, meanwhile, have risen by an estimated 15-20 percent due to the energy shock alone. For firms contemplating large-scale model training, cloud expansion or data centre localisation, that increase is a direct reconsideration of project economics. TSMC faces a different but equally serious vulnerability: its dependence on helium makes 5nm-and-below production more exposed to external disruption, including the GPU supply chains that serve Nvidia and other AI accelerators. The broader consequence is a pause in corporate decision-making. Companies are already delaying AI and digital transformation projects, because cost uncertainty has widened abruptly.
The war has also clarified a deeper paradox in defence modernisation. The Pentagon has confirmed the use of “advanced AI tools” in combat operations against Iran, highlighting how rapidly military organisations are integrating machine intelligence into targeting, intelligence fusion and operational planning. Yet, every such system rests on hardware that depends on semiconductors, and every precision-guided munition, missile defence radar and stealth aircraft is similarly dependent on the same fragile industrial ecosystem. The war is thus accelerating demand for military AI at the very moment it is constraining the industrial inputs required to produce it.
That contradiction is sharpened by dependence on China-controlled materials. Most advanced defence chips use gallium nitride, whose exports to the US China embargoed in 2024. As highlighted, China also controls more than 80 percent of tungsten production, which is essential for munitions. Conventional warfare consumes the same upstream materials that support advanced industry. China also dominates gallium and germanium production, both of which are essential for guidance electronics in missiles. This creates a coupling trap: the cheapest and most expensive munitions alike depend on materials controlled by a strategic competitor.
The characterisation of Iranian strikes on data centres as “the first war against AI” identifies a strategic reality. AI has become a wartime target precisely because it is now central to not only warfighting but also economic sustainability. What emerges, then, is a structural lesson. The Iran conflict has exposed the fact that AI supremacy depends equally on geopolitical stability and material access, not innovation alone.
Asia’s rise in chip and AI leadership faces a constraint that no amount of software sophistication can remove: the underlying physical systems sustaining computation remain vulnerable to conflicts thousands of miles away. The scramble for military AI reveals a further irony, since the nations seeking algorithmic advantage depend on adversaries for the raw materials. Until alternative supply chains are built, Asia’s technological ascent will remain vulnerable to West Asian stability and Chinese material dominance.
Noor Al Mazrouei | Head, Artificial Intelligence Department, Trends Research & Advisory, Abu Dhabi
(Views are personal)