With the increasingly widespread application of artificial intelligence (AI), the operating costs of data centers have also increased. According to the Wall Street Journal, some data centers in the United States have had to raise prices to customers in order to cope with the additional expenses brought by AI. At the same time, data shows that the capacity demand for large-scale data centers and managed data centers is rapidly growing, while the capacity demand for enterprise built data centers is relatively shrinking.
The Wall Street Journal quoted CBRE Group, a commercial real estate service company, as saying that in Northern Virginia, the world’s largest data center market, it has more than 275 facilities. This year, the amount of electricity available for lease has shrunk to 38.4 MW, compared with 46.6 MW last year. In this region, the monthly electricity cost per kilowatt for data center customers is as high as $140, an increase of nearly 8% from $130 in 2022. In contrast, in Silicon Valley, customers’ monthly electricity costs per kilowatt are as high as $250, an increase of 43% from $175 in 2022.
The cost pressure faced by data centers not only comes from power consumption, but also includes processing capacity, advanced chips, supply chain challenges, AI technology, and other aspects. According to the Wall Street Journal, “AI applications consume more energy than traditional software because they need to process more data. Within a few days, a single AI model can consume tens of thousands of kilowatt hours of power. The technology used by generative AI models, such as OpenAI’s ChatGPT Chatbot, is 100 times larger than standard AI tools.” In addition, construction The costs of labor and public utilities are also constantly rising.
Corresponding to the increase in data center costs is the rapid expansion of the cloud service market. According to Synergy Research Group, over the past decade, companies’ spending on cloud services has grown at an annual rate of approximately 42%, from $10 billion in 2012 to $227 billion in 2022. On the contrary, the expenditure of enterprises on data center hardware and software is only growing at a rate of about 2% per year.

Synergy Research Group also provides distributed data on global data center capacity usage. In 2022, ultra large data centers accounted for 37% of all global data center capacity, compared to only 17% in 2017. Large scale data centers refer to large facilities operated or leased by cloud service giants such as Amazon, Microsoft, and Google. Half of them are owned data centers, and the other half are leased data centers. Non ultra large scale hosting facilities account for 23% of the market share. The self built data centers of enterprises accounted for 40% of the market share in 2022, a significant decrease from 60% in 2017.
Synergy Research Group predicts that over the next five years, “although the proportion of enterprise built data centers in total capacity will decrease by more than two percentage points annually, their actual capacity will only slightly decrease. The proportion of managed facilities in total capacity will remain relatively stable.” The company also predicts that by 2027, ultra large data centers will account for over 50% of the capacity share, And the self built data center of enterprises will be reduced to less than 30%.
One of the main factors driving the growth of large-scale data centers is the prosperity of digital consumer services. Services such as social networks, e-commerce, and online games require a large amount of data storage and processing capabilities, and large-scale data centers can provide these capabilities. According to Synergy Research Group, there are currently approximately 900 data centers owned or leased by ultra large data center operators worldwide.