How safe are data centres in times of war?

In open, hot warfare, the aim of combatants is to cripple the enemy’s economy and force surrender — especially when military victory is beyond reach. This is an age-old strategy, deployed since the budding of civilisations, the domestication of plants and animals for food, and the emergence of trade. During the Peloponnesian War of 431 to 404 BCE, the Spartans infamously raided Attica and destroyed crops during the farming season, with the aim of disrupting the agricultural foundation of the Athenian economy and forcing them to fight outside their impenetrable walls.

Food supply is one of the many soft spots of a nation’s economy. Another common target is infrastructure — from roads and bridges to water and power structures, communication networks, schools and hospitals. The Iran conflict has once again underlined how energy supply — in this case, oil — can be used as a lever to choke the economies of the Gulf. This has led to calls, especially in the UK, for energy diversification and greater investment in renewables.

As the global economy flexes to the new realities of Industry 4.0 — blending physical, digital, and biological systems through artificial intelligence (AI), internet of things (IoT), robotics, and biotechnology — another critical infrastructure has emerged as a target for combatants: cloud data centres. These sprawling facilities prop up computing power, storage, and application hosting for much of the developed world.

What are cloud data centres?

The cloud refers to servers, data storage, databases, and software that can be accessed over the internet, as opposed to via a local computer. This enables consumers and businesses to rent computing power and storage from any location and device, from providers like Amazon Web Services (AWS), Microsoft Azure, and Google Cloud. The physical infrastructures that enable these services — cloud data centres — are physical facilities that house the requisite hardware, servers and storage capacity. This ensures the cloud is scalable, on-demand, and reliable.

Data centres are also increasingly critical to the AI revolution, providing the 24/7 computing power that trains and runs large-scale AI models. Unlike traditional data centres, these facilities utilise highly specialised hardware, like graphic processing units (GPUs), tensor processing units (TPUs), and liquid cooling mechanisms. Without these centres, access to AI would all but cease.

Where are the biggest data centres?

With their massive resources, access to rare earth minerals, and cohorts of highly skilled labour, China and the US are the key global superpowers driving data centre investment.

In Hohhot, China, sits one of the world's largest data centres: China Telecom’s Inner Mongolia Information Park. It spans almost 10 million square feet and has a capacity of 150 MW. The facility leverages the area's cold climate for natural cooling, and supports massive cloud computing and data processing operations. Other notable centres in China include the Zhangjiakou data centre cluster, a major hub near Beijing, and the China Unicom AI centre, which focuses on leveraging locally developed AI chips.

Meanwhile, in Tahoe Reno, Nevada, a major colocation facility is under construction: Switch’s Citadel campus. At over 7 million square feet, its 650 MW power capacity eclipses China Telecom’s Inner Mongolia Information Park, and aims for 100% renewable energy.

The UK, with its world-renowned tech-academic base, research facilities, and impressive clutch of AI unicorns, is also looking to build out its data centre infrastructure. In Northumberland, the $13 billion AI-ready QTS centre will support up to 720 MW when finished, significantly upscaling the UK’s cloud capacity.

The data that these centres contain includes vast amounts of digital information, from personal data to multimedia content, to critical enterprise applications and AI workloads. They act as central repositories of information, cloud backups, healthcare data, web infrastructure, and more. As such, data centres are exposed to immense geopolitical risk and have become potential targets in hot warfare.

What happens when the cloud goes down?

While direct damage to major, active, hyper-scaler cloud data centres is rare, recent cases have revealed the potential ramifications.

On 2 March 2026, AWS announced that some of ​its data centres in the United Arab Emirates (UAE) and Bahrain were damaged by suspected Iranian drone strikes, which caused considerable disruption to its cloud services. Multiple availability zones in the region were impacted, reaping structural damage, power outages, and fire suppression water damage to equipment. Products impacted included S3 (an object storage service), Lambda (a serverless compute service), and DynamoDB (a managed NoSQL database). This affected financial institutions (FIs), payment services, and apps across the region.

Such incidents raise questions around data sovereignty, and nations’ dependence on a small group of global hyper-scalers. Indeed, in a wider conflagration, data centres could be knocked out to shut down combatants’ digital backbone and induce pandemic-level economic impacts. Russia, for instance, has already displayed willingness to take down Ukraine’s energy infrastructure, while Iran and the US have threatened to hit the region’s desalination and power plants, respectively. Data centres could become yet another casualty in such escalations.

How is digital infrastructure protected?

Data centres, ideally, are designed for protection against both physical and cyberattacks.

With the former in mind, these infrastructures are often housed in reinforced, windowless bunkers, in underground locations — and rely on distributed availability zones to shift workloads in the event a site is destroyed. Many centres are even circumscribed by razor-wire fencing, bomb barriers, and patrolled by armed guards and anti-drone systems. What's more, remote power sources, such as solar arrays or protected generators, are used to maintain operations during power grid attacks.

In terms of cyber defence, data centres are often air-gapped, meaning sensitive, defence-related data is stored in servers disconnected from the public internet to prevent hacking.

Looking ahead, the physical risk and terrestrial regulatory hurdles around data centres could be circumvented by placing them in space. The benefits would include continuous solar power and a natural vacuum to facilitate cooling while in orbit. Firms like SpaceX and xAI are already exploring this idea, but outsized launch costs, harmful solar radiation, and maintenance logistics will prove considerable hurdles.

The digital paradox

Industry 4.0 has revealed a paradox inherent in industrial development. Indeed, one of the benefits of digitisation has been its ability to abstract away from the reliance on physical systems. In many ways, this has been delivered. Organisations around the world, large and small, now benefit from on-demand IT resources, delivered over the internet: from computing power, to storage, databases, and software. This enables users to pay only for what they use, thus improving efficiency, scalability, and security.

Yet, in doing so, access to these services has become highly centralised; not just around vendors like AWS, Azure, and Google Cloud, but around a handful of data centres that, in some cases, supply entire regions. Unsurprisingly, leaders of states are scrambling to understand the risks, diversify, and build protective bubbles around these critical infrastructures.