Why Under the Sea?
Some suggest that building data centers at the bottom of rivers would be more convenient and cost-effective since river water also flows. However, the answer is: not really!
The primary reason for constructing data centers under the sea is that seawater has a higher specific heat capacity than freshwater. This allows the heat generated by data centers to dissipate more quickly. In other words, for the same cooling effect, the cost of cooling with seawater is lower than with freshwater. Additionally, data centers often require astronomical data throughput and high operational stability. Submarine cables are currently the most mature high-speed broadband connection, making underwater data center deployment advantageous. They can connect to submarine cable buses nearby, avoiding land-based construction challenges and ensuring long-term reliable operation.
Why Not Riverbed?
Taking the Yangtze River as an example: although the Yangtze River is China's largest river, with a length of 6,387 kilometers and a total basin area of 1.8 million square kilometers, it is not suitable for building underwater data centers. The reasons include:
The average depth of the Yangtze River basin is not deep, ranging from 1.5 to 5 meters, which does not meet the depth requirements for underwater data centers.
The summer months of July and August are flood seasons for the Yangtze River, prone to flooding disasters that underwater data centers cannot withstand.
The period from December to April is the dry season of the Yangtze River. During the dry season, even navigation is challenging, not to mention the cooling of underwater data centers.
Building data centers on the riverbed is not conducive to riverbed sand discharge, hinders navigation, and seriously affects the ecological environment.
Why Not Reservoirs?
While large reservoirs do not require navigation, their value lies in rich ecological resources. Moreover, reservoirs primarily function to control floods and provide drinking water. During droughts, reservoirs may dry up, making them less stable than seawater. In contrast, an underwater data center off the California coast releases 800 kilowatts of thermal energy per hour. If the water is stagnant, the surrounding water temperature may increase by about 0.7°C per hour. Large reservoir water volumes and the speed of heat exchange through water are insufficient to support underwater data centers and can significantly impact the reservoir's ecological environment.
Why Not Lakes?
Lakes are smaller than seas, with no ocean currents, and water flow is relatively stagnant. Therefore, lake water may not cool as effectively, potentially causing lake water temperature increases and affecting the ecosystem.
Alibaba Cloud once established a data center with a total floor area of 30,000 square meters at Qiandao Lake, with a total capacity of 50,000 servers. However, this data center was not built at the bottom of Qiandao Lake. Instead, it extracted deep lake water through a completely sealed pipeline, passing through the data center to help cool servers. After natural cooling, the water finally returned to Qiandao Lake. The data center's Power Usage Effectiveness (PUE) ranged from 1.27 to 1.28, much higher than Highlander's 1.076. In conclusion, the quantity and volume of data centers are large, making them unsuitable for placement in reservoirs or lakes. The ocean is the most suitable option.