By 2027, the market size of green data centers will increase by approximately $147 billion
With the rapid development of generative artificial intelligence, such as the popular ChatGPT, the demand for computing power in data centers has surged. Artificial intelligence applications read large amounts of data and consume more electricity than traditional software. The GPU used for training generative AI models has high power consumption and also requires additional cooling energy.
According to estimates, in terms of macro data, artificial intelligence may account for 3% to 4% of global electricity demand by 2030. Due to the surge in artificial intelligence servers, power consumption in data centers has been significantly increasing. McKinsey predicts that by 2030, the electricity consumption in data centers will more than double.
The increase in electricity consumption and costs is a key factor driving market growth. The energy consumption of the data center is very high, and with the increasing demand for more powerful applications such as autonomous vehicle, streaming media and 5G, the energy consumption of the data center may grow exponentially, and they need a lot of power to supply different equipment.
To address the enormous energy demand challenges in data centers, various measures need to be taken, including energy-saving hardware, innovative cooling solutions, green energy, and broader sustainable development strategies.
The use of energy-saving chips is the cornerstone of improving energy efficiency in data centers. Energy saving chips have advanced architecture and power management functions, playing a crucial role in minimizing data center hardware power consumption. These chips can more effectively allocate and utilize hardware resources, thereby improving performance per watt. For example, compared to previous generations of Intel Xeon processors, the fourth generation Xeon improved average performance efficiency per watt for target workloads by 2.9 times when using built-in accelerators. In 2022, the energy efficiency of Nvidia's H100 GPU AI chip is almost twice that of the previous generation product A100.
In addition, another effective measure to reduce energy consumption in data centers is to adopt more efficient cooling solutions on a large scale, reducing the proportion of cooling energy consumption, and a key indicator is "Power Use Efficiency" (PUE). In the past decade, despite a 6-fold increase in computing output and a 25-fold increase in storage capacity, the energy usage of global data centers only increased by 6% from 2006 to 2018. This significant efficiency improvement is attributed to the decrease in PUE.
According to estimates, the green data center market is expected to increase by $146.95 billion between 2022 and 2027, with a compound annual growth rate of 24.63%. With the increasing deployment of liquid cooling (especially direct liquid cooling DLC) on a large scale, the PUE of data centers entering the liquid cooling era will be below 1.3. Liquid cooling technology can not only improve the overall cooling efficiency of data centers, but also meet the cooling needs of high-power density chips, reduce dependence on high power consuming air conditioning systems, and promote sustainable environmental development.
