The artificial intelligence is reshaping the global market for data center, pushing investments to unprecedented levels and transforming structures created to host servers and data into real critical infrastructures. According to a new report from Allianz Commercial, annual spending in the sector could more than double within a few years, from around $500 billion in 2024 to over $1.000 trillion as early as 2027.
This race isn't just about building data centers. The new investment cycle also involves energy production and distribution, power grids, cooling systems, telecommunications infrastructure, and semiconductors. Increased capacity, however, also comes with a leap in complexity and risk. Fires, extreme weather events, blackouts, water damage, and operational interruptions can translate into losses of tens or hundreds of millions of dollars.
The United States and China lead the race, while Europe seeks new hubs.
La growth will remain heavily concentrated in the two major technological powers. United States e China should represent approximately the 62% of new global capacity installed by 2030, even though the geography of data centers is becoming progressively more complex. In Europe, Germany, the United Kingdom, and Ireland remain among the key markets, while Allianz sees faster expansion prospects in Spain, Finland, and Denmark, favored in some cases by greater energy availability and more favorable permit conditions.
More stronger growth expected in Asia-Pacific Outside of China, installed capacity in the region is expected to rise from the current 9 GW to over 28 GW by 2030, while Malaysia could see a more than tenfold increase.
“Artificial intelligence is transforming the latest generation of data centers from specialized real estate assets to essential critical infrastructure,” he says. Thomas Lillelund, CEO of Allianz Commercial. The scale of the investments, he emphasizes, makes factors such as energy availability, supply chain reliability, construction controls, site selection, and insurance programs capable of considering the actual concentration of risks increasingly crucial.
Energy, jobs and climate become the new bottlenecks
The limit to growth is no longer just the availability of capital. What weighs most heavily are physical and operational constraints, starting with access to electricity and the time it takes to get connected to the grid. Added to these factors are the availability of specialized equipment, permits, and a shortage of skilled workers.
In the United States, the construction sector is already having to deal with a estimated deficit of approximately 439.000 skilled workers, while a further 349.000 workers could be needed during 2026.
More and more relevant is the climate factorApproximately 79% of global data center capacity is located in areas with a high risk of natural disasters, while 54% is exposed to chronic heat and drought. Some of the fastest-growing markets for AI-related infrastructure, such as Northern Virginia in the United States, Johor in Malaysia, and Marseille in France, are also among the most exposed. In the Americas, according to Allianz, risks such as floods, fires, and wind affect 86% of capacity, while in Asia-Pacific, heat and drought threaten 89%. The ability to withstand extreme events therefore becomes a strategic component right from the design stage of new facilities.
Fires and operational shutdowns cause claims costs to explode
The increase in size of campuses and technological density also changes the nature of the potential lossesAllianz Commercial's analysis of data center-related claims shows that fire are responsible for more than half of the approximately 700 million euros of losses examined, in front of natural disasters. They are followed by malicious acts, a category which also includes criminal and cyber incidents, and power failures.
The damage caused by the awhere instead represent the most frequent type of accident. From an economic point of view, however, the one that weighs most heavily is interruption of business, because even a relatively small problem can knock out high-value computing capacity and simultaneously impact multiple customers and infrastructure.
In large hyperscale or colocation campuses, servers, cooling systems, batteries, energy infrastructure, construction sites, and multiple operators can coexist. A single incident can thus simultaneously generate claims for property damage, construction, liability, cyber risk, and business interruption. Case studies analyzed by Allianz show how cooling system problems, fires during construction, or startup delays caused by power supply anomalies can cause losses of between $50 and $100 million per single event.
Insurance to reach $24 billion by 2030
With insured values and risks increasing, even the The data center roofing market is set to change scaleAllianz Commercial expects the sector to grow from the current 11 billion dollars to over 24 billion by 2030, more than doubling in the space of a few years.
The insurance model is also expanding beyond traditional property protection. New solutions will need to integrate construction, engineering, property damage, business interruption, liability, and cyber risk, along with energy and operational continuity.
The value at stake can be enormous. The construction of a single AI campus can exceed $20 billion, even before fully considering the increased insured values resulting from the installation of high-performance computing equipment. For many large projects, comprehensive insurance coverage has now become a prerequisite for obtaining financing.
“For insurers, the key question is not just the value of the building, but the concentration of value and dependencies within and around it,” explainsto Christian Kolbe, Global Head of Construction Claims at Allianz Commercial. Energy, cooling, batteries, fiber connections, and business continuity are now part of the same risk landscape. In a sector that AI is transforming into one of the key infrastructures of the digital economy, resilience must therefore be incorporated into projects from the earliest stages.
