The Iran conflict is a double-edged shock for AI as it disrupts the physical and financial foundations of the current AI commercial boom, while simultaneously accelerating military and intelligence adoption of AI systems, writes eToro analyst for Romania, Bogdan Maioreanu.
While the current situation might slow down the general adoption of AI, it is transitory and dependent on how long the conflict will last and how much damage will be inflicted on the global economy.
The current computing demand is high, and available processing power is becoming scarce while the renting price of servers is steadily increasing. According to the latest industry report of CBRE, the overall vacancy rate in data centres in major metropolitan areas with high concentrations of infrastructure fell to a record low 1.4% at year-end 2025. At the same time, the average asking rate for computing power, in a 250-to-500-kW data center cluster, jumped by 6.6% year-over-year to a record $196.25 per kW/month at the end of last year. While rate growth continued, the pace slowed after three consecutive years of double-digit growth, but it still almost doubled from 2021.
AI data centers are huge electrical power consumers. Attacks on Gulf energy infrastructure and higher oil and gas prices raise operating costs and can slow data-center build‑out and model deployment. The United States, China and Europe are projected to remain the largest regions for data centre electricity demand over the coming years. According to data at the end of 2024, the main fuel to produce electricity in Europe and the United States is natural gas, while in China, it is coal. In a more granular way, in the European Union, the main source is in fact nuclear, followed by wind and then gas. In Romania, the main source is hydro, followed by nuclear and gas.
The current conflict in the Middle East increased the price of natural gas in Europe by 59%, while in the US, it is still stationary, having almost the same price as before the conflict. This may make more queries go toward the US, which is cheaper, but degrades performance as the data centers will be overwhelmed and queues increase. A prolonged conflict in the Gulf region might continue to fuel the rise of natural gas prices, adding to the operating costs of the data centers in other regions.
With limited capacity available, hyperscalers were devising expansion plans in the Gulf (UAE, Qatar, etc.). But the latest developments highlight how the region is exposed to missile and drone risks, making some sites effectively uninsurable and forcing a rethink or relocation to places like India, Southeast Asia, or Northern Europe. And this will take time, increasing pressure on existing capacities. In addition, the current capacities in the Gulf area are at risk of being attacked. Iranian military communications have explicitly labelled major U.S. tech and AI companies (e.g., Amazon, Microsoft, Palantir, Oracle, NVIDIA, OpenAI) as legitimate targets, and drones have already damaged Amazon’s AWS facilities in the UAE and Bahrain, causing outages. This is showing how the concentration of AI workloads in a few hyperscale clouds makes them vulnerable to localized strikes, which can cause global service degradation or outages and highlight AI’s dependence on a fragile energy‑and‑cloud stack.
While all these are risks for the AI industry as a whole and for its main players, the conflict in the Middle East is accelerating the military development, training and usage of AI. The conflict might be fought with a mix of human and artificial intelligence, including AI‑enabled targeting, surveillance, and decision support for military operations. The current situation incentivizes further investment in autonomous systems, reconnaissance, cyber‑AI, and decision‑support tools, reinforcing AI’s role as a core military technology even as the civilian AI boom faces macro and infrastructure headwinds.
The conflict in the Middle East, the closing of the Strait of Hormuz and the risks energy prices are bringing to global inflation are making Central Banks keep the interest rates higher for longer. Also, the private credit sector that funds much of the data‑center builds, right now, is experiencing a lot of stress. This might reduce available capital for AI infrastructure and make large projects easier to delay or cancel. In this context, the rising energy and capital costs, combined with insurance and security premia, might push up the cost per token or per model training run, which can compress margins and slow the rollout of the most advanced AI models.
Despite all this, the situation is transitory and will depend on how long the conflict lasts and how quickly the global economy recovers after its end. So far, the ETF that tracks the Magnificent 7 stocks has lost only 5% from the beginning of the hostilities, showing that the markets might be mildly concerned about the prospects of the industry.
