The rapid expansion of artificial intelligence (AI) and cloud computing is driving a significant increase in data centre construction worldwide, according to Allianz Commercial. However, with this expansion comes significant risks.
The company’s latest report outlines how the demand for AI technologies is resulting in large-scale development projects, particularly in the United States and China. Market research suggests that up to US$7 trillion will be spent on data centres by 2030, with technology firms in these countries leading the investment, while Europe’s growth remains comparatively slower.
Allianz Commercial notes that Amazon, Microsoft, and Google Cloud collectively accounted for nearly two-thirds of global cloud revenue in the second quarter of 2025. Their combined capital expenditure, along with that of Chinese firms such as Alibaba and Tencent, is set to reach hundreds of billions of US dollars this year. Much of this spending is directed toward the infrastructure and energy resources needed to support high-performance AI and cloud computing.
The report highlights that the costs of constructing data centres have increased sharply, with average-sized facilities now requiring between US$500 million and US$2 billion. Some projects have exceeded US$20 billion.
Allianz Commercial’s construction specialists point out that the complexity of these projects means they often require project-specific insurance policies, tailored to address risks such as power supply issues, faulty workmanship, fire, and natural catastrophes.
As the industry accelerates its adoption of AI and automation to support these developments, insurers are also contending with new forms of risk. The shift towards digital transformation is no longer just about efficiency or speed; it now requires a strategic approach that prioritises business objectives and accountability.
This means that, alongside the physical risks of large-scale construction, insurers must also address challenges such as data privacy, cyber security, and the potential for algorithmic bias in automated decision-making.
The integration of AI into underwriting, claims triage, and customer service is also reshaping the risk landscape for insurers. One significant concern is that AI systems trained on historical data may inadvertently reinforce existing biases, leading to inequitable outcomes. Industry experts emphasise that while AI can enhance decision-making, human oversight remains essential to ensure fairness and accuracy in both underwriting and claims processes.
The United States is expected to remain the largest market for data centres, projected to account for about two-thirds of global data centre power demand by 2028, with 81 gigawatts of capacity.
China’s market is also expanding rapidly, with Greater Beijing alone representing roughly 10% of global hyperscale capacity. Europe, while behind the US and China, is experiencing a 43% annual increase in pipeline activity, with London and Dublin leading the region.
Chris Fancher (pictured above, left), US head of construction property at Allianz Commercial, notes, “The bigger data centres have a huge footprint. The scale of a US$20bn+ facility can involve tens of thousands of workers on site at peak times, with significant equipment and building supplies moving in and out.”
He adds that tight project timelines and the risk of faulty workmanship can lead to losses or costly delays.
Christian Sandric (pictured above, right), regional managing director of Allianz Commercial Asia, says, “As the demand for data centres in the region surges, it is crucial that parties fully understand the risks involved during construction and operation.”
He adds that, beyond the main risks, stakeholders must also consider cyberattacks and environmental impacts, such as the effect of cooling systems on local water bodies.
“These complex and extensive risks call for specialist insurance and expert risk-management guidance, and clients need to work with an experienced team of underwriters who knows the business and can support the project from beginning to end, including multi-year coverage and policy extensions as needed,” Sandric said.
