Data centres used to run AI services have become a new target in the war with Iran. Whereas earlier attacks focused on oil installations and military infrastructure, the digital layer on which artificial intelligence operates has now also appeared on attackers' radar.
According to analysts, this shift reflects how significant the strategic value of AI infrastructure has become, both for the economy and for national security. Data centres are therefore no longer merely a business interest, but a geopolitical battleground.
From oil to data: a shift in targets
In earlier phases of the conflict with Iran, the focus was on physical, highly visible targets: oil refineries, military bases and other components of the traditional power base. That logic is still present, but is now supplemented by a new category. AI data centres today represent a concentration of computing power, data and economic value comparable to strategic energy sources.
Analysts point out that damaging or disabling data centres has direct consequences for business continuity, military applications of AI and public services. The threshold for treating digital infrastructure as a target has demonstrably fallen in recent years, partly because dependence on such systems has increased sharply.
Specific details about which attacks have taken place, who carried them out and what the exact damage is cannot be verified on the basis of available source material. The report at Emerce indicates that analysts have identified the trend, but does not cite documented individual incidents with traceable perpetrators.
Why AI infrastructure is such an attractive target
AI systems require large amounts of computing power, concentrated at relatively few physical locations. Large language models, image recognition and military decision-support systems run on clusters of specialised chips, primarily graphics processors and dedicated AI accelerators, that are difficult and costly to replace.
That concentration makes data centres vulnerable. An attack on a single facility can disable services that are indispensable to hundreds of thousands of users or to critical government systems. At the same time, data centres are relatively straightforward to locate: their power consumption, cooling systems and network connections are visible in satellite imagery and public permit documents.
Added to this is the fact that AI is increasingly intertwined with defence and intelligence services. Disrupting that infrastructure therefore has potentially greater consequences than simply taking a commercial service offline.
Broader context: data centres as critical infrastructure
The recognition of data centres as a strategic target is not a new idea in security circles, but its practical manifestation in an active conflict gives the discussion a different dimension. Western governments have in recent years taken steps to formally classify digital infrastructure as critical infrastructure, with corresponding security requirements and response mechanisms.
In the European Union, the NIS2 directive has been in force since 2023, requiring organisations in sectors such as digital infrastructure to manage cyber risks in a structured manner and to report incidents. The Netherlands has transposed this directive into national legislation. The growing attention to physical threats against data centres, alongside the more familiar cyber attacks, confronts policymakers with additional questions about surveillance, site selection and redundancy.
For the European AI sector, the development is relevant because the continent is investing heavily in its own AI infrastructure, partly in response to the dominance of American and Chinese technology companies. The Netherlands plays a role in this as a data centre hub, due in part to its favourable position for international fibre-optic connections and the presence of large campuses around Amsterdam and the Randstad. The question of how that infrastructure is secured, both physically and digitally, takes on more direct urgency for both operators and policymakers as the threat landscape shifts.