BeePetLove1784123951

Rising Temperatures Could Make AI Data Centers More Expensive to Cool
As demand for AI continues to grow, a new study suggests that climate change is reducing the effectiveness of one of the most affordable and energy-efficient methods used to cool data centers.
Researchers from the University of Hawaiʻi at Mānoa, publishing in Scientific Reports, analyzed 45 years of hourly weather observations, climate model projections, and a global database of data center locations. Their findings indicate that rising temperatures and increasing humidity are more frequently exceeding the recommended limits for direct air free cooling, a technique that relies on cool outdoor air instead of mechanical cooling systems.
According to the study, these unfavorable conditions have become significantly more common over the past 45 years, particularly in tropical regions and the southeastern United States. As a result, a growing number of data centers now face conditions that limit the use of direct air free cooling for at least one-quarter of the year.
The researchers project that these limitations will continue to expand through the middle of the century as temperatures and humidity continue to rise. They also found that the most extreme heat and humidity events are becoming more severe at a faster rate than average conditions, creating additional challenges for data center reliability, emergency planning, and infrastructure design.