Too hot to compute: data centres face rising heat and water risks

Saw some pretty eye-opening research from Verisk Maplecroft about rising climate risks for global data centres. Thought it was worth bringing here since a lot of us depend on multi-region infrastructure without really tracking these risks.

Key points from the report:

- Over 50% of the top 100 data centre hubs already face high heat and water stress risks. - By 2040, nearly 75% will be in extreme heat zones, with major jumps in cooling demand and operating costs. - Cooling already makes up ~40% of total data centre power usage, and that’s expected to increase sharply. - A typical mid-sized DC burns through 1.4 million litres of water per day for cooling. With rising temps, that’s going to get worse. - By 2030, more than half of these hubs will also be in high water stress areas, meaning not just cost issues but also risks of outages, or even political blowback from local communities.

Thought-starter:

Is anyone here factoring in environmental stress when designing multi-region failover or disaster recovery setups? Right now, most of us optimise for latency and cost, but it feels like these environmental risks are going to creep into operational reliability decisions, especially with regulatory pressures increasing too.

Tools like CarbonRunner.io that route workloads based on both carbon intensity and water stress data, turning off regions with unsustainable risk factors. Curious if anyone else is doing something similar or if this is still too early-stage for most teams.

Source (Maplecroft report): https://www.maplecroft.com/products-and-solutions/sustainabl...