AI Data Centres Outstrip Insurance Capacity Rapidly

The rapid expansion of artificial intelligence (AI)-driven data centres is creating a profound challenge for the global insurance and reinsurance industry, as the scale, complexity, and cost of these facilities increasingly outpace available coverage capacity. A recent report by Swiss Re highlights how multi-billion-dollar projects are stretching traditional underwriting models and exposing structural gaps in risk transfer markets.

Escalating Costs and Coverage Gaps

Modern AI data centres are no longer conventional infrastructure projects. Construction costs for a single facility can reach as much as $20 billion, and once high-value equipment such as graphics processing units (GPUs) and advanced computing systems are installed, the total insured value can double. This dramatic escalation significantly increases the amount of insurance required for a single site.

However, insurers face a fundamental constraint: reinsurance markets are often unable to provide sufficient capacity at competitive rates to match these soaring valuations. Financing arrangements for such projects typically require full coverage of construction and operational risks, yet insurers may only be able to underwrite a portion of the exposure. This mismatch is becoming a critical bottleneck for developers and financiers alike.

Surging Demand for Data Centre Insurance

Demand for specialised insurance cover is rising sharply alongside the growth of AI infrastructure. Industry projections suggest that global data centre insurance premiums could increase from $10.6 billion today to approximately $24.2 billion by 2030, reflecting both the expansion of capacity and the increasing complexity of associated risks.

Yet this growth is not without complications. Insurers must navigate a rapidly evolving risk landscape where traditional actuarial models may no longer apply, particularly as new technologies introduce unfamiliar vulnerabilities.

Geographic and Catastrophe Risk Concentration

One of the most pressing concerns identified by Swiss Re is the geographic concentration of data centres in high-risk regions. In the United States, more than a quarter of data centre capacity is located in areas that experience at least three large hail events annually. Furthermore, around 40% of facilities are situated in zones prone to three or more EF1-strength tornado days each year.

This clustering effect significantly increases the likelihood of large, correlated losses. A single severe weather event could damage multiple facilities simultaneously, amplifying insurers’ exposure and challenging diversification strategies.

Key Risk Drivers in Data Centres

The report identifies several major sources of risk that are reshaping underwriting considerations:

Fire remains the most severe risk in terms of financial impact. Although it accounts for a relatively small proportion of incidents, it generates a disproportionately large share of losses. Emerging technologies are exacerbating this threat: lithium-ion battery backup systems are increasingly embedded within server racks, introducing new ignition sources directly into sensitive data environments.

Water-related risks are also gaining prominence. Advanced cooling systems, particularly direct-to-chip liquid cooling used in AI servers, heighten the probability of leaks, installation faults, and maintenance failures. Such incidents can lead to extensive equipment damage and prolonged downtime.

Power Demand and Business Interruption

Business interruption represents another critical exposure. According to data from the Uptime Institute, power supply issues account for approximately 45% of data centre outages.

AI workloads significantly intensify power requirements. While traditional servers typically consume between 5 and 15 kilowatts per rack, AI servers can demand over 100 kilowatts. This surge is prompting developers to integrate on-site power generation and battery storage systems, which in turn introduce additional risks, including fire, explosion, and the release of toxic gases.

Accumulation and Systemic Risk Challenges

A further complication for insurers is accumulation risk—the potential for a single event to trigger multiple claims across interconnected systems. Large data centres are often insured through separate programmes covering buildings, equipment, and energy infrastructure. This fragmentation can obscure the total exposure.

Shared systems such as cooling networks, power supply, and fire protection mechanisms mean that a single failure could impact multiple tenants simultaneously. Consequently, insurers may face cascading losses across several policies, complicating claims management and capital allocation.

Outlook: A Structural Challenge for Insurers

The findings from Swiss Re suggest that the insurance industry must adapt rapidly to keep pace with the evolution of AI infrastructure. Traditional underwriting approaches, capacity limits, and risk models may need to be rethought to accommodate these mega-scale, technology-intensive assets.

Without innovation in risk sharing, pricing, and capital deployment, the gap between insurable demand and available supply is likely to widen. As AI data centres continue to proliferate, bridging this divide will be essential—not only for insurers, but for the broader digital economy that increasingly depends on these critical facilities.