Rise of Liquid Cooling Technology: Chip Manufacturers Enter the Fray to Compete for the Multi-Billion-Dollar Market

Data from Omdia Research shows that by 2028, the global data center liquid cooling market will reach $16.8 billion, with a compound annual growth rate (CAGR) of 25%. In China, according to CINNO Consulting's forecast, by 2027, driven by the large-scale implementation of AI-related applications and the gradual maturity of the liquid cooling ecosystem, the market size is expected to exceed the 100-billion-yuan mark.

Furthermore, a research report by CITIC Securities mentioned that under the impetus of improving data center computing power and reducing energy consumption, liquid cooling technology will accelerate the replacement of air cooling technology and become a new technological development direction.

With the development of liquid cooling in data centers, the liquid cooling industry has gradually become more segmented, and the industrial chain has been continuously improved. One of the most obvious changes is that in previous data center conferences, liquid cooling technology rarely made an appearance. Even if it did, it was mostly confined to presentation PPTs and the display sessions of some full cabinet solutions.

However, starting from last year, at various data center-related conferences, in addition to the display of liquid cooling cabinets, more and more component manufacturers in the upstream and downstream of the industry have become active. Manufacturers of downstream liquid cooling components such as coolants, UQD connectors, and water 凝 tubes (water 凝 tubes refer to hoses used for liquid circulation in liquid cooling systems) are also constantly upgrading and iterating their products.

Data Centers Step into the Liquid Cooling Era

The large-scale application of liquid cooling in data centers can be traced back to around 2020. Before the advent of ChatGPT, data centers were mainly equipped with general-purpose CPU servers, with power consumption basically concentrated around 3kW.

At that time, driven by the newly launched 'dual carbon' strategy (a national strategy aiming to peak carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060), data centers had just begun to adopt green technologies. As the cooling system accounts for the highest proportion of energy consumption among non-IT equipment in data centers, it naturally became a breakthrough for new technologies. Technologies such as indirect evaporative cooling, natural air cooling, and AHU (Air Handling Unit) units have been widely applied in data centers.

Different from these cooling methods that still rely on 'air', the exploration of liquid cooling technology in data centers has enabled data centers to achieve heat dissipation through 'liquid' (mainly coolants). However, at that time, users' attitudes towards liquid cooling were twofold: on the one hand, they thought it was 'unnecessary'; on the other hand, the data center industry, which holds a relatively conservative attitude towards technology, had extremely high requirements for business stability, so its acceptance of new technologies was not high. At that time, there was a common view in the industry: 'It is better not to use new technologies than to introduce additional risks due to their application.'

Nevertheless, for the industry, liquid cooling had already begun to emerge in data centers during that period. A few service providers started to propose the concept of liquid cooling, but the majority of voices in the industry at that time believed that these concepts were just hype and were not optimistic about the application scenarios of liquid cooling in data centers.

Everything changed after the launch of ChatGPT in 2022. The new wave of AI has not only reshaped various industries but also brought about a subversive transformation in the architecture of data centers. The growth of general computing power has gradually slowed down, while the growth of intelligent computing power has been accelerating rapidly.

According to IDC data, the global AI server market size will reach $125.1 billion in 2024, increase to $158.7 billion in 2025, and is expected to reach $222.7 billion by 2028. Among them, the proportion of generative AI servers will rise from 29.6% in 2025 to 37.7% in 2028.

In response to this, Zhou Zhengang, Vice President of IDC China, once told the author: 'From the perspective of growth trends, during the period 2023-2028, the five-year CAGR of China's intelligent computing power is expected to reach 46.2%, while that of general computing power is expected to be 18.8%. It is not difficult to see from the data that currently, most of the incremental demand for computing power from users will come from intelligent computing power.' Zhou Zhengang emphasized.

In addition, according to the latest liquid cooling industry research by TrendForce (集邦咨询), it is estimated that the penetration rate of liquid cooling in AI data centers will increase significantly from 14% in 2024 to 33% in 2025, and will continue to grow in the coming years.

Behind the booming development of intelligent computing, people have begun to notice that the power consumption of cabinets is increasing. The original single CPU with 300W is no longer sufficient to support business operations. Single CPU chips with 500W have become the mainstream, and in extreme cases, the power of a single CPU has approached 1kW.

At the same time, the demand for GPUs (Graphics Processing Units) driven by AI has further increased the power consumption of individual cabinets. The 2.5kW standard cabinet has become a thing of the past. The continuous emergence of cabinets with 8kW, 10kW, and even 20kW or 50kW power has made the power consumption of individual cabinets higher and higher.

Against the backdrop of high power consumption, traditional air cooling can no longer meet the needs of high-density and efficient heat dissipation in data centers simultaneously. In recent years, liquid cooling has gradually become the mainstream cooling method for data centers.

The Liquid Cooling Industry Gradually Becomes 'Segmented'

Chip manufacturers have gradually recognized the application of liquid cooling technology in data centers and have joined the liquid cooling industry.

For example, Intel has made in-depth layouts in both cold plate liquid cooling and immersion liquid cooling directions.

In terms of immersion liquid cooling, Intel collaborated with Green Cloud Tech (绿色云图) and launched the G-Flow immersion liquid cooling cabinet-related technology in 2024. This technology adopts single-phase immersion liquid cooling. Notably, in this solution, the coolant used is the latest PFAS-free immersion coolant (a type of hydrocarbon synthetic oil) developed by ExxonMobil, which is more environmentally friendly than traditional fluorinated coolants.

In the field of cold plate liquid cooling, Intel collaborated with Invengo (英维克), a domestic provider of precision temperature control and energy-saving solutions, to launch the world's first Intel-certified cold plate liquid cooling integrated solution based on the Xeon 6 platform.

As a chip manufacturer, Intel's layout in liquid cooling technology also provides some references for end-users in choosing liquid cooling solutions. Earlier, Hu Shixuan, CEO of Green Cloud Tech, told the author: 'For users, the endorsement of hardware manufacturers is one of the key considerations when choosing liquid cooling technology. For instance, after users adopt liquid cooling solutions, hardware manufacturers such as chip and server manufacturers were previously reluctant to provide maintenance services.' Hu Shixuan said, 'However, this situation has significantly improved. Hardware manufacturers like Intel have begun to deploy applications in the field of liquid cooling.'

In addition to its layout in liquid cooling technology, Intel has also made in-depth efforts in building the data center liquid cooling ecosystem in China and launched the 'China Data Center Liquid Cooling Innovation Acceleration Program'.

In this regard, Du Liguang, Director of Platform Technology R&D, China, Data Center and AI Group at Intel, told the author: 'The overall idea of Intel's China Data Center Liquid Cooling Innovation Acceleration Program is to carefully divide the industrial chain from upstream to downstream, and cooperate with leading manufacturers in each key liquid cooling component segment.'

'By cooperating with leading manufacturers, we empower them to develop new products and new materials,' Du Liguang further pointed out. 'Then, we promote these high-quality manufacturers, which have been empowered through cooperation with Intel, to our OEM (Original Equipment Manufacturer) and ODM (Original Design Manufacturer) partners. Finally, through these partners, we reach Intel's end customers, thereby achieving win-win and multi-win outcomes for the industry.'

The 'First Step' of the Acceleration Program

Not long ago, under the guidance of the Liquid Cooling Innovation Acceleration Program, Intel took the lead in establishing the 'Intel Universal UQD Interoperability Alliance' and conducted the first batch of certifications for five independent hardware suppliers in the liquid cooling ecosystem, including Invengo, Danfoss, Liminda Technology (立敏达科技), Lanke Electric (蓝科电气), and Zhengbei Connection (正北连接).

In the liquid cooling system of a data center, quick connectors are one of the important components. They need to maintain 'zero leakage' during a service life of several years and thousands of plug-in operations to ensure the normal operation of the server. However, due to differences in product size, end design, sealing form, and quality among different manufacturers, it not only brings cost challenges to procurement and maintenance but also requires more efforts in compatibility verification, connector replacement, and liquid circuit debugging during subsequent equipment upgrades.

In this regard, Du Liguang took quick connectors as an example and pointed out: 'Generally speaking, the service life of a data center cabinet is 8-10 years, but the update and iteration cycle of servers is usually 3-5 years. Especially in the AI era, the iteration of servers may be even faster. On this basis, if the specifications of internal components of the cabinet can be unified, so that there is no need to replace the cabinet when replacing servers later, it can save enterprises a considerable amount of costs.'

In addition, in the early days, most manufacturers in the field of liquid cooling components for the data center industry were small in scale, with relatively scattered specifications and uneven quality. However, as the application of liquid cooling has experienced explosive growth, these problems have gradually been amplified. Du Liguang said: 'With the increase in the number of UQD applications, on the one hand, more and more data centers are using liquid cooling; on the other hand, in terms of product form, more and more components are used in each server or rack.'

'All these have led to higher requirements for the reliability of UQDs,' Du Liguang further pointed out. 'No one wants the liquid cooling products they use to leak frequently, regardless of the reason.'

The above-mentioned issues have also brought greater challenges to the liquid cooling industry, which is why Intel has launched the Liquid Cooling Innovation Acceleration Program and established the UQD Interoperability Alliance at this juncture.

During the process of promoting UQD standardization, the alliance also encountered some problems. In this regard, Du Liguang gave an example to introduce the difficulties encountered in the UQD testing process. Taking the simplest interoperability test as an example, in the early stage of testing, Intel adopted a machine-operated method for plug-in testing. However, it was soon discovered that with machine operation, the angle of each plug-in was the same. In actual application scenarios, maintenance personnel cannot achieve exactly the same plug-in action every time. Therefore, the test results obtained through machine operation may not be applicable in actual application scenarios.

'Based on this, we changed our strategy and conducted the test through manual plug-in operations by real personnel. The entire test required more than 3,000 manual plug-in operations,' Du Liguang emphasized.

This meticulous approach is just a microcosm of the rigor of the entire testing process. The relevant person in charge of Lanke Electric, which participated in the testing, also highly praised the rigor of this UQD alliance. He said: 'UQDs have a variety of application scenarios, and their operating conditions vary under different scenarios and working conditions. Unlike previous UQD tests we participated in, this test conducted comprehensive and rigorous tests on various application scenarios and extreme conditions to ensure the high reliability of UQDs in all situations.'

'In the testing process, both the test standards and potential failure modes were covered in the test plan. Moreover, the test equipment and test personnel required for the test plan were highly professional. The test report was carefully considered and revised repeatedly to ensure that it can provide great reference value for Internet giants, integrators, and server manufacturers,' the person in charge further pointed out. 'In the end, what we provided is not only a report but also a highly valuable industry design plan and acceptance standard.'

As the demand for AI computing power surges at an annual rate of over 40%, and the power consumption of individual cabinets jumps from several kW to dozens of kW, or even hundreds of kW, liquid cooling technology is no longer an 'alternative solution' for data center heat dissipation, but a 'must-answer question' to support the evolution of the digital economy towards high density, high energy efficiency, and high reliability.

The unification of UQD connector specifications is just the beginning. Liquid cooling technology has shown an increasingly segmented trend in data centers. Under the influence of this trend, more and more liquid cooling components will achieve generalization and standardization. At that time, the cost of applying liquid cooling in data centers will further decrease, and may even be lower than the deployment cost of air cooling.

Kres(Suzhou) Power Systems Technology Co., Ltd is an emerging company focused on fluid power solutions, and serve for the markets of Energy, Construction, Agriculture, Manufacturing, Transportation, Medical, Data Centers, etc., and was founded in 2022 with leading-edge engineering capabilities and innovative products to keep up with the changing customer needs. Including OCP standard UQD, UQDB, BMQC, manifold, EPDM hose and hose assemblies, FF9 full flow dual interlock quick disconnect coupling (a better replacement for FD83 series products), API heat transfer products, Gates EPDM hose and Wika products.

KRES FF9 connectors are specifically engineered for fluid transmission, electronic cooling, and data center liquid cooling applications, where full flow, fluid compatibility, and safety are of paramount importance. The FF9 series ball valve quick-disconnect couplings adopt a full-flow design and are now available in 2-inch, 1.5-inch, and the popular 1-inch sizes. The FF9 double interlock liquid cooling quick-change connector features a swivel structure at its tail, facilitating easy connection and disconnection. Additionally, its two connector halves are completely identical, simplifying maintenance, installation, and service processes.

· 100% full-flow interlock ball valve: The FF9 double interlock ball valve has a diameter of up to 21mm, ensuring large flow while reducing pressure drop, achieving large flow and low pressure drop, and has obtained SGS ROHS reports and Huace performance test reports.

· KRES interlock ball valve adopts a double interlock design with two locking mechanisms to prevent accidental disconnection and ensure data security.

· The FF9 liquid cooling quick-change ball valve has identical connector halves, making maintenance, installation, and service simple.

· As a liquid cooling quick-change ball valve for data centers, it has reliable and stable performance with minimal leakage, meeting the maintenance needs of safe machine relocation or shutdown.

· Standard body material of the liquid cooling quick-change connector: 304 stainless steel, with wide fluid compatibility, and other materials can be provided upon request.

· Standard sealing material of the liquid cooling quick-change connector: EPDM ethylene propylene diene monomer synthetic rubber, and other materials can be provided upon request.

· Operating temperature range of the liquid cooling-specific ball valve: -40℃ ~ 120℃