In response to growing concerns about sustainability, Amazon has disclosed for the first time the water usage of its data centers worldwide, highlighting ongoing efficiency improvements even as operations expand.
- Water use fell 2% per kilowatt-hour despite growth in data center footprint.
- Air cooling dominates, with evaporative methods reserved for peak heat periods.
- Amazon claims water efficiency notably better than industry averages and key rivals.
Infrastructure signal
Amazon's disclosure marks a rare industry insight into data center water consumption, revealing a total of 2.5 billion gallons used globally in 2025. This figure reflects consumption at a rate of 0.12 liters per kilowatt-hour, representing a 2% efficiency gain versus 2024 despite expanded infrastructure.
The company primarily leverages air cooling 90% of the time, resorting to evaporative cooling only during peak heat events. Additionally, Amazon has optimized server hardware to tolerate higher operating temperatures, reducing reliance on intensive water-based cooling. These infrastructure choices directly influence cloud resource sustainability and operational costs.
Developer impact
Developers relying on Amazon’s cloud infrastructure can anticipate a platform increasingly optimized for sustainability without compromising availability or performance. The shift toward elevated server temperature tolerances allows for stable deployments under more variable environmental conditions, potentially reducing downtime risks related to cooling system constraints.
Operational efficiencies in water use may indirectly benefit teams through cost stabilization or potential pricing benefits as Amazon’s investments in efficient cooling systems and infrastructure engineering reduce waste and overhead. Observability tools and APIs should continue to reflect stable service metrics aligned with these reliability improvements.
What teams should watch
Infrastructure, sustainability, and operations teams should monitor further updates on Amazon’s water usage reports, especially as regulatory environments evolve following local moratoriums like Seattle’s one-year pause on new data center builds. Potential impacts on cloud availability and regional infrastructure investment decisions may emerge from such policies.
Engineering teams should also watch for advancements in cooling technology and server thermal tolerance strategies that could affect deployment standards and environment specifications. Monitoring competitor water-use disclosures might also provide broader context for strategic cloud infrastructure planning and vendor comparisons.