Europe is already showing signs of strain. The annual inflation rate rose to 2.5% in March, according to preliminary Eurostat figures, driven by an increase in energy prices of 4.9%. Governments are reaching for measures from the 2022 playbook: caps on gas prices, windfall taxes on energy companies, and fuel tax reductions. With European gas prices climbing sharply, approximately 70% since the start of the conflict according to the EU's energy commissioner, the region's dependence on imported energy is translating directly into economic vulnerability.

This pressure feeds directly into cloud infrastructure. Data centres, responsible for around 70 terawatt-hours (TWh) of electricity demand in the EU by 2024 according to IEA estimates, are fundamentally energy-dependent systems. They are accountable not only for data processing and deployment, but for training and running AI models, cooling, and maintaining physical infrastructure. As the IEA reported in April 2025, global data centre energy use is projected to double by 2030, with AI as the primary driver.

As demand for AI accelerates, so does their energy consumption. Between rising energy prices and increasing cloud operating costs, the assumption of a cheap cloud begins to erode.

Europe's structural digital dependency

Europe's push towards digital sovereignty has made progress through regulations and investments in AI infrastructure and R&D, but it remains structurally incomplete. The region continues to rely heavily on external providers for its cloud and data infrastructure.

European companies hold around 15% of the local market, according to Synergy Research Group, a share that has held roughly steady since 2022. US hyperscalers, Amazon Web Services (AWS), Microsoft Azure, and Google Cloud, account for approximately 70% of the European cloud market. As TNW has previously reported, European cloud hosts offer alternatives, but none come close to challenging the scale of the big three.

The imbalance positions Europe primarily as a consumer, exposing a vulnerability: by importing cloud services, Europe is also importing the geopolitical risk embedded in the systems it depends on.

The Iran war did not create this exposure, but it has brought it into focus. Trump's tariffs had already reignited the push for European sovereign cloud; the current conflict sharpens the urgency.

Europe now faces pressure on both sides of its digital economy: higher operating expenses driven by volatile energy markets, while remaining primarily dependent on foreign-controlled cloud infrastructure.

Infrastructure location is now a risk variable

Cloud infrastructure is no longer just a technical decision. It is becoming a geopolitical one. Recent events have made this shift unmistakable.

In early March, three AWS data centres in the Gulf were struck by Iranian drones, the first known military attacks on a hyperscale cloud provider. Meanwhile, Iranian state media has explicitly identified companies including Nvidia, Apple, Microsoft, and Google as potential targets within the conflict.

Data centres and cloud platforms are no longer neutral infrastructure. They are increasingly being treated as strategic assets embedded in conflicts. Their significance extends beyond storage and compute power to the functioning of entire digital ecosystems, where disruptions can affect connectivity, latency, and cloud service availability across industries.

The implications are economic as much as technical. The Gulf region has been central to the future of cloud and AI expansion, attracting tens of billions of dollars in data centre investment, driven by access to capital, energy, and infrastructure.

Instability in this region introduces a new layer of risk to cloud architecture, where the possible consequences range from operational disruption to broader macroeconomic pressure. As a result, decisions around cloud architecture and deployment need to be re-evaluated in the light of geopolitical risk management.

Spending is being reshaped, not reduced

The impact of the Iran war on tech highlights how companies and countries are reshaping capital allocation within the industry. While multiple big tech companies had planned to continue investing and expanding within the Gulf, execution is now being questioned on grounds of geopolitical stability, national security, and economic continuity.

IDC has outlined scenarios in which a conflict lasting up to three months could reduce global IT spending growth by approximately one percentage point, from around 10% to 9% in 2026. Longer-duration conflict would produce a more pronounced impact.

The research firm notes that cybersecurity and AI budgets appear likely to be ring-fenced, even as discretionary and device spending faces downward pressure.

Investments by national sovereign wealth funds may be redirected to local security needs. Capital allocation towards cybersecurity appears to be becoming non-negotiable, as geopolitical tensions raise the likelihood of large-scale cyberattacks and disruptions in the digital supply chain.

Simultaneously, AI remains a strategic investment priority, linked to competitive advantage and productivity gains. The war is acting as a capital allocator in the tech industry, centring the pressing issues that require investment while constraining the less vital ones.

AI resilience instead of sovereignty

AI sovereignty is the capacity for countries and companies to control their AI stack: infrastructure, data, models, and operations. A recent BCG Henderson Institute study, examining AI policy across more than 30 countries, argues that for most nations, full AI sovereignty is an illusion.

The more practical path, BCG suggests, is AI resilience: using, adapting, and governing AI domestically at scale, while minimising strategic dependencies. The question of what that sovereignty actually looks like in practice is one Europe has been wrestling with for years.

This implies that countries and companies need to approach AI strategy as a balance between internal capabilities and external, strategic interdependencies. However, resilience assumes that the countries acting as external providers will remain stable, and the current geopolitical environment suggests the underlying infrastructure is becoming less predictable.

In Europe, the tension is particularly visible. While efforts towards strategic autonomy, aimed at reducing reliance on US and Chinese technology, had already begun, the Iran war has highlighted how the continent's AI strategy still lags behind global competitors.

As TNW has reported, the race to resolve Europe's AI sovereignty concerns has been underway for some time, but building on foundations that remain structurally exposed introduces a new risk: resources may be allocated towards short-term crisis management, delaying the development and deployment of a resilient European AI ecosystem.

The end of assumed stability

The war in Iran is exposing the vulnerabilities of the current global system at micro and macro levels. For Europe, it highlights external dependency in crucial markets, such as energy and data infrastructure, that over time could become a strategic constraint: by slowing its AI development and competitive positioning among world powers, it also poses a threat to its growth and productivity.

As AI strategies shift from sovereignty to resilience, a new challenge emerges. Resilience depends on the stability of the systems it is built on, and that stability can no longer be taken for granted. Achieving controlled interdependence requires diversification: broader and multiple alliances to reduce geopolitical exposure, alongside targeted investment to scale domestic capabilities, mainly in strategically sensitive areas.

Ultimately, the Iran war did not create these vulnerabilities, but it acts as the stress test for a system that was optimised for a stable and predictable world that no longer exists. If the previous era was defined by the efficiency of borderless expansion, this new era of resilience marks the end of the cheap cloud, as the costs of security and energy independence are finally priced into the digital stack.