The rapid proliferation of data centers globally and the rise of energy-intensive artificial intelligence (AI) are highlighting a critical issue in the sector: overheating-related outages. The outage at CME Group, the world’s largest stock exchange operator, on Thursday night, which halted trading in numerous markets ranging from currencies and commodities to bonds and stocks, has brought the fragility of cooling systems back into the spotlight.
The Dallas-based CyrusOne has more than 55 facilities in the US, Europe, and Japan.
The company announced that engineering teams are working on-site to fix a cooling failure that occurred at its data center near Chicago.Why Do Data Centers Heat Up?
High-capacity AI and cloud servers consume large amounts of energy to process data, and this process generates significant heat. The constant operation of stacked server racks leads to traditional air cooling being insufficient. According to Daniel Mewton of Slaughter and May, server chips must remain within specific temperature ranges; Otherwise, it malfunctions or shuts itself down.
New Solutions for Cooling
The industry is turning to liquid cooling, which uses water or special liquids instead of air. Although this method is approximately 3,000 times more efficient than air in removing heat, it carries risks such as leakage, corrosion, and high water consumption. Large technology companies are trying to reduce the need for external cooling. Last year, Microsoft announced a data center design with zero water consumption, explaining that they recirculate heat using closed-loop water recycling technologies.
Some facilities are also working on systems that can recover and reuse the waste heat generated.
How Common Are Outages?
Mewton emphasizes that data center outages due to cooling problems are generally “extremely rare.” The main reason for this is the strict contractual obligations of data center operators that guarantee almost full-time employment.
These developments further increase the strategic importance of cooling technologies in an era of rapid growth in artificial intelligence and data processing capacity.
