Release Time :January 2, 2024

What Makes High-Density Data Centers Increasingly Popular as AIGC reshapes Infrastructure?


Kevin Kelly, in “The Inevitable”, argues that technology has a natural tendency to move in specific directions.


In the digital economy, with the rapid advancement and widespread adoption of artificial intelligence, it’s clear that this direction is towards computing power technology. Computing power has emerged as the core productivity driver of the latest technological and industrial revolutions, making the development of advanced data center infrastructure essential in today’s landscape.


At the beginning of this year, AIGC swept across the globe, accelerating changes in data center infrastructure like a catalyst. Companies that provide data center solutions will inevitably be part of this transformation, and Chayora, a provider of hyper-scale data center infrastructure, is among the key players.


Unlike just a few years ago, AIGC is now reshaping data centers and introducing new requirements for computing power, storage, cooling, and even the overall environment. To meet this surge in demand, developing build-to-suit high-density data centers solution will be crucial for unlocking productivity.


AIGC Reshapes Data Centers


In recent years, the rapid growth of the digital economy has led to data centers expanding in scale, increasing the number of devices, and experiencing rising power consumption costs.


The emergence of AIGC and large language models has brought significant “qualitative changes” to data centers, positioning them as the driving force behind the future of infrastructure. The reason for this transformation lies in the need for large language models to perform real-time training and inference on massive datasets, all of which require substantial computing power. For instance, OpenAI's latest GPT-5 model demands around 30,000 to 50,000 H100 GPUs, highlighting its intense demand for computing resources.

Currently, the swift advancement of AIGC has created a significant demand gap in computing power while also introducing new requirements for data center computing capabilities, storage, and energy efficiency. This shift is poised to impact the future development of data centers comprehensively.


Firstly, the rapid evolution of AIGC has sparked an extreme demand for computing power, simultaneously driving the need for higher computing power density in data centers. OpenAI’s analysis indicates that in recent years, the computing power required for AI training has surged by an astonishing 300,000 times—far outpacing Moore’s Law. In the Chinese market, the ongoing intense competition among the large model has also led to significant changes in data center computing power.



The growth of artificial intelligence and the demands of large language models will fundamentally reshape the data center industry. Most of existing data centers will struggle to meet the requirements of these advanced models. The launch of large multi-modal models like OpenAI's GPT-5 and Google Gemini this year marks a pivotal trend; these models will generate unprecedented volumes of data. Gartner predicts that by the end of 2023, 20% of content will be created by AIGC, and by 2025, artificial intelligence will account for 10% of all data generated, leading to ongoing, massive storage needs.

Secondly, the push for high-density computing means packing a large number of processors and devices into limited space, resulting in a surge in power consumption. This presents significant challenges for energy conservation and emissions reduction. Consequently, green and efficient computing power has become a critical requirement for many data center users. For instance, under the carbon peak and neutrality initiative, the “Three-Year Action Plan for the Development of New Data Centers (2021-2023)” aims to reduce the Power Usage Effectiveness (PUE) of newly built large data centers to below 1.3 by the end of 2023.


We can anticipate that the power density of single cabinet will exceed current expectations. The large model is currently mainly used for training. However, it will pose even higher requirement on the infrastructure with the increase of inference load in the future. Presently, the mainstream power density for a single cabinet is around 20-40 kW, but this may rise significantly to 40-60 kW in the future.


Clearly, the rise of AIGC has disrupted the traditional development pace of data center infrastructure technology, driving a comprehensive transformation of data centers. So, what will be the most important development direction for data centers moving forward?


High-Density Data Centers: A Growing Demand


Economist Joseph Alois Schumpeter once suggested that establishing a new production system requires “creative destruction” to accelerate the obsolescence of outdated technologies and methods. Today, in the rapidly evolving digital economy, there’s a pressing need to establish a new production system centered around AI technology, which can only be achieved through innovative data center infrastructure. This shift is essential for initiating a new cycle of digital economic growth.



This year, the development and construction of high-density data centers have garnered significant attention as they represent the cutting edge of infrastructure. The advantages of high-density data centers are twofold: they allow for large-scale deployment of high-density cabinets, maximizing computing power per square foot, and they incorporate advanced energy-saving technologies, such as liquid cooling, which enhance energy efficiency and promote greener, low-carbon operations. Global hyper-scale data center providers agree that the new generation of high-density data centers must be characterized by advanced technology, high computing power, high energy efficiency, and robust security.


For instance, Chayora’s recently launched Ingenuity high-density data center solution is specifically designed to support high workloads, such as those generated by AIGC. It delivers up to 40 kW of power per rack, meeting the demands for computing power and storage in scenarios involving large model training and inference. To address the limitations of traditional air cooling in high-density data centers, the Ingenuity solution utilizes an air-and liquid cooling technology. This innovative approach allows for flexible and efficient cooling solutions tailored to cabinets with varying power densities within the same facility. By combining these methods, the solution optimizes the Power Usage Effectiveness (PUE) and proves more sustainable in terms of cost, efficiency, and environmental impact.



A prime example of this is the Chayora Tianjin Data Center Campus, which has achieved the Uptime Institute’s T3 certification in design, construction, and operation. This facility boasts an impressive annual average operation and maintenance PUE of just 1.25, showcasing its highly energy-efficient infrastructure that meets the growing computing power demands in the Beijing-Tianjin-Hebei region.


According to IDC’s “Global Artificial Intelligence Spending Guide”, China’s AI investment is projected to reach $38.1 billion by 2027. Industry experts highlight that the country is actively promoting the development of digital infrastructure through initiatives like “Eastern Data and Western Computing”. With strong policy support and a focus on industrial transformation and upgrading, data center infrastructure is poised to become a key area of market investment over the next five years. As a result, the demand for high-density data centers in China is expected to rise steadily. Solutions like Chayora Ingenuity are set to unleash significant industrial value in the market.


Chayora's Differentiated Approach to Data Center Transformation


Currently, the data center landscape is evolving rapidly. On one hand, new technologies—such as liquid cooling—are advancing quickly, alongside fast-changing industry standards. Experts emphasize the critical role of standardization in fostering industry growth. On the other hand, the rise of technologies like AIGC is diversifying user workloads, leading to shifts in market demand. Traditional data centers are struggling to keep pace, resulting in a widening gap between supply and demand.


So, how can this issue be addressed? Chayora is focusing on “built-to-suit data centers solutions”. Their approach tailors solutions to meet specific needs, covering the entire process from planning and design to system integration and operational management. This model is more agile than traditional data centers, allowing for quicker construction, better adaptability to business needs, and enhanced cost control. Chayora’s Ingenuity high-density data center solution exemplifies this customization, offering flexibility in design, space, power supply, and cooling to meet the high-density requirements of customers comprehensively.



The Tianjin data center serves as a prime example of Chayora's commitment to build-to-suit solutions . Here, the design takes into account factors like room layout, power, and networking to provide clients with ample flexibility and room for business expansion. In addition to applications in AIGC, these tailored services also effectively support sectors such as autonomous driving, healthcare, manufacturing, and smart cities. Notably, the civil engineering of the Tianjin campus was specifically designed to facilitate high-density deployment. Therefore, customers can benefit from optimal solutions without extensive modifications. This approach not only prevents resource waste but also significantly reduces data center delivery times, helping customers balance costs with the demands of high-density deployment.


So, how does Chayora bridge the gap between the supply and demand of data centers? The answer lies in three core capabilities:


Firstly, As an infrastructure provider specializing in hyper-scale data centers, Chayora prioritizes strong communication with clients in the Chinese market and industry experts. From the initial design phase, they engage in discussions to understand and address specific needs related to construction scale, timelines, and systems. With over a decade of experience in building and operating hyper-scale data centers in China, Chayora boasts a complete local professional team that plays a vital role in facilitating user communication and collaboration.


Secondly, Chayora has laid a solid foundation for build-to-suit solutions with a comprehensive portfolio of customized products and solutions. Chayora’s strategy centers on customization, allowing them to offer a range of flexible infrastructure solutions tailored to different business requirements. For instance, in response to the rapid deployment needs of AIGC clients, Chayora’s Ingenuity high-density data center solution can be adapted at three levels: modular setups, dedicated floors, or entire data centers, enabling customers to enter the market quickly.


Thirdly, through its partnership with EdgeConneX, Chayora can assist Chinese customers in expanding globally, providing access to over 50 popular markets in Asia and beyond, including Malaysia and Indonesia.


Looking ahead, AI is set to become as integral to everyday life as electricity. In light of this inevitable shift, data centers must prepare for future transformations. As part of this evolution, Chayora remains dedicated to developing high-standard data center infrastructure in the Chinese market so as to effectively serve local customers.


This article is reprinted from "Big Data Online.”Click to read the original article.

/ Industry Insights /