How can insurers keep up with AI data centre risks?

Increased power and battery storage can cause fire and equipment failure.

Artificial intelligence (AI)-driven demand could nearly triple global data centre power consumption by 2030, forcing insurers to reassess facilities exposed to strained electricity grids, climate hazards, and cyberthreats.

“Power availability and energy resilience have become leading concerns as data centre facilities hyperscale,” Terence Williams, head of commercial risk for the Asia-Pacific region at Aon Plc, said in an August statement. “AI-driven demand could increase global data centre power consumption by 165% by 2030,” Williams said.

Developers are adding power generation, battery storage, microgrids, and hybrid energy systems to address grid constraints. These systems can also create fire, equipment failure, and maintenance risks, Williams said.

Higher computing demand is increasing power density, cooling needs, and equipment values at data centre sites, he added.

Aon increased capacity under its Data Centre Lifecycle Insurance Program from $3.5b to $5b, Williams said. The program assesses risks from design and construction through power generation and operations.

Natural hazards and climate conditions are also affecting data centre locations and construction standards across the Asia-Pacific region, HDI Global SE said in an August report.

Tokyo faces earthquake, storm, and flood risks. Typhoons can produce wind speeds of as fast as 200 kilometres per hour, while climate projections point to more severe heat and rainfall extremes, HDI Global said.

The insurer said developers should assess floodplain, earthquake, wildfire, and severe-storm exposure when selecting sites.

Mumbai and Singapore could experience more than 200 days a year above 35°C from 2081 to 2100, based on climate projections cited by HDI Global.

It warned that standard commercial building requirements might not provide enough protection for data centres facing future climate conditions.

It recommended emergency roof overflows, raised floors, elevated critical equipment, floor drainage, and barriers to prevent water or fire damage from spreading.

As operators move into secondary cities for land and electricity, Aon and HDI Global expect insurers to assess climate exposure, water availability, power security, construction standards, and emergency arrangements before providing coverage.

Questions to ponder:

• How can data centre operators secure reliable power without creating additional insurance risks?
• Should insurers require stronger climate and fire protections before covering AI-focused data centres?
• How will rising power density and equipment values affect data centre insurance costs?