Nearly 80% of datacenters face climate hazards: what this means for AI emissions accountability

Nearly 80% of datacenters globally are exposed to extreme climate hazards including flooding, extreme winds, and wildfires, according to new research from climate risk analytics firm First Street. The same infrastructure driving the artificial intelligence boom is therefore vulnerable to the very physical risks that AI workloads are accelerating through their greenhouse gas emissions. First Street's findings point to disrupted operations, increased time offline, and inflated insurance and repair costs as the practical consequences for affected facilities.

The timing of this research lands at a moment of peak scrutiny on AI energy consumption. Concerns have been mounting across the ESG investment community about the scale of electricity demand and associated emissions tied to rapid datacenter expansion. The physical risk dimension identified by First Street adds a second layer of materiality that sustainability teams at technology companies and their investors will need to address.

For corporate ESG managers and CFOs responsible for CSRD reporting, physical climate risk is not optional disclosure. Under ESRS E1, companies must assess and disclose their exposure to climate related physical risks across both their own operations and their value chains. Datacenters used by a company, whether owned or contracted through cloud service agreements, could plausibly fall within the scope of that assessment depending on how material the dependency is.

Procurement and infrastructure teams at large organisations should consider whether their cloud and computing suppliers are themselves disclosing physical risk exposure. A datacenter that goes offline due to a flood or wildfire creates operational disruption that cascades through every digital workflow, including the compliance and reporting tools that ESG teams depend on. Supplier resilience is therefore both a business continuity and an ESG due diligence question.

The First Street study arrives as AI infrastructure buildout accelerates across Europe and North America, with regulators and investors paying closer attention to the twin exposures of operational emissions and physical climate vulnerability. The paradox of AI contributing to the climate hazards that now threaten AI infrastructure itself is the kind of systemic risk narrative that is increasingly difficult for asset managers and sustainability officers to set aside.