Foreword

The global race to construct AI and digital infrastructure is accelerating at an extraordinary speed, with hyperscale operators and institutional investors committing billions of dollars to new developments. While public debate has focused on energy demand and water consumption, physical climate risk is an emerging challenge. Extreme weather events including flooding, coastal inundation, forest fire, cyclones, and extreme heat are becoming more frequent. Data centres, which operate around near-continuous uptime expectations, face outsized financial and operational consequences from even short disruptions. Swiss Re projects that global insurance premiums associated with data centres could rise from US$10.6 billion today to US$24.2 billion by 2030. Planned data centres provide a window of opportunity to influence future performance through site selection, engineering standards, and resilience investment.

About this report

This report presents findings from an analysis of physical climate risk and resilience across 2,595 planned data centre facilities globally. It considers physical damage, operational disruption from heat, infrastructure dependency, and indirect risks including supply chain impairment. The analysis uses engineering-based infrastructure archetypes to assess how design choices influence performance. To identify vulnerabilities that may not be visible under lower-impact scenarios, the report uses the high-emissions RCP 8.5/SSP5-8.5 scenario as a stress-testing benchmark. Risk is measured using ‘Maximum-to-Date Value at Risk’ (MVAR), with properties classified as High Risk (MVAR ≥ 1.0%), Moderate Risk (MVAR ≥ 0.2% – 1.0%), or Low Risk (MVAR < 0.2%).

Data centres and climate risk by region

Regional analysis indicates that South East Asia has the highest proportion of planned data centres at risk, with 20% classified as high risk. This is followed by East Asia at 13% and South Asia at 12%. North America contains the largest number of analysed assets (1,599), with 5% identified as high risk. Several regions face significant increases in average damage risk by the end of the century, with Oceania projected to see a 339% increase and Europe a 289% increase between 2026 and 2100.

Key findings

Climate resilience is already an operational risk for digital infrastructure. In 2026, 154 planned data centres were identified as high risk under low resilience settings, with almost half of these located in North America. France is identified as the world’s fifth riskiest country for planned data centres despite receiving US$69 billion in foreign investment. Within France, the Nouvelle-Aquitaine region emerged as a major hotspot. In the United States, Oklahoma recorded the highest concentration of risk, with 67% of planned facilities modelled to be high risk. The report finds that while higher resilience standards can materially reduce risk, they do not eliminate it everywhere; locations like Vietnam and Thailand remain high risk even under advanced resilience settings.

Global rankings: where resilience requirements may be greatest

Vietnam and Thailand rank first and second globally for risk of damage from climate change extreme weather in 2026, driven primarily by coastal inundation. Subnational rankings identify Nouvelle-Aquitaine (France) as the top-ranked state for risk, with 100% of analysed assets classified as high risk. Other high-risk states include Oklahoma (US) at 63% and Rio de Janeiro (Brazil) at 50%. Extreme heat is emerging as a major operational challenge, with planned assets in Brazil, India, Mexico, Indonesia, and Spain recording some of the most acute projected disruption risks.

Indirect risk: operational disruption and infrastructure dependency

Data centres do not operate in isolation and remain dependent on surrounding systems including electricity networks, water supply, and transport. Analysis of a synthetic portfolio of 138 centres across Europe indicates that the risk of operational disruption increased ten-fold when indirect risk was included. Productivity loss, which measures the expected proportion of time an asset is unavailable, is modelled to triple by the end of the century. This suggests that direct asset analysis may capture only a fraction of the climate-related risks facing corporate operations.

Conclusion

The expansion of AI and digital infrastructure represents one of the largest investment waves in history. This analysis demonstrates that climate resilience is becoming an increasingly important consideration for infrastructure designed for high availability. While engineering specifications can reduce modelled risk, location-specific due diligence remains essential. Investors, operators, and governments should use the current planning window to identify vulnerabilities and embed resilience before construction commences to ensure long-term operational continuity and value creation.