Follow The Lee Company
Global data center capacity is expected to double by 2030 as nearly 100 GW of new data centers are built. Under the pressure of this rapid growth, the way that data centers are designed and operated is shifting. As power scarcity and infrastructure become growing concerns, data center operators are looking to get more processing capacity from increasingly limited resources.
In this article, we’ll discuss the forces driving the growth of data centers and how selecting the right component supplier can help thermal design engineers maximize processing potential, energy densities, and system uptime. As liquid cooling is becoming a critical part of data center infrastructure, finding a fluid control supplier who can deliver in the following areas is becoming increasingly important:
Data centers are the backbone of many tools you use every day – from your cell phone to navigation and content streaming. The first data centers emerged in the 1940s, when computers were so bulky that they required large, dedicated facilities for operation. With the advent of new technologies like cloud computing and virtualization in the early 2000s, these facilities have become increasingly sophisticated and are now seen as an essential piece of society’s infrastructure.
As day-to-day life continues to become more digital, there is growing demand for computing power. This explosive growth is driven primarily by the widespread adoption of AI. By 2027, AI is projected to comprise 28% of the data center market. Thanks to the massive potential for this technology, the AI boom that is fueling demand for data centers is expected to continue for quite some time.
Data centers are huge consumers of electricity, and AI data centers can require considerably more energy per square foot than traditional data centers. For example, NVIDA’s H100 GPU can have power needs up to 400 Watts, the H200 can require up to 800 W, and their next generation Blackwell chip can consume up to 1,200 W per GPU. As higher performance GPUs and CPUs are used, data centers will require more power. This trend of denser chip designs, combined with increasing demand for compute power, means that power demand from data centers globally is expected to double by 2030.
The increasing energy demands of data centers is drawing attention among politicians, industry professionals, electric grid operators, developers, and consumers – and with good reason. By 2030, data center energy consumption could account for up to 9% of U.S. electricity use.
To help address this concern, engineers are focusing on maximizing the power usage effectiveness (PUE) of data centers. Organizations like the Open Compute Project are also working with industry leaders to identify opportunities to make data center infrastructure more sustainable.
Cooling systems are a critical piece of data center infrastructure. To maximize their longevity and performance, CPU and GPU chips need to be kept at stable temperatures. As denser chip technologies that need more power are adopted, more advanced cooling technologies become necessary. If data center cooling technology can be made more efficient, it would meaningfully reduce the power requirements of data centers – making growth more sustainable.
Many data centers are now adopting liquid cooling technologies over traditional air cooling. Liquid cooling provides more efficient performance and is more effective at removing the heat generated by more powerful chips. Current liquid cooling technologies include immersion cooling and direct-to-chip (D2C) liquid cooling, which uses liquid-cooled plates mounted directly onto chips to cool electronics.
Microfluidic control at the chip level takes cooling technology to the next level, read more here.
Scaling data center capacity to meet global demand while minimizing unnecessary energy consumption is an ambitious, but critical task. With liquid cooling systems becoming the future of data center infrastructure, fluid control component suppliers are emerging as an important partner for industry leaders.
Here are four ways that the right supplier can help you minimize energy use and maximize performance in your data center.
Data center cooling systems must provide reliable, precise flow across a range of thermal loads for optimal chip performance. When you choose products that offer efficient and precise performance in the smallest possible envelope, those benefits directly translate to the performance of your system – helping you maximize computing power while reducing ongoing energy use. For example, using miniature components at the tray level enables you to fit more chips in a tray, adding more processing power to your data center. Products that can provide stable flow rates at low pressure drops require less pump power, reducing the ongoing energy consumption of your system. Ideally, you also want components that can respond to changing system needs and direct flow based on where the most heat is being generated. The right supplier will be familiar with how critical these performance needs are to reducing power consumption and maximizing computing performance in your system – and they will have proven solutions ready to ship. Look for suppliers who are already providing components to data center cooling systems or other critical industries where low pressure drops, responsive performance, and constant flow rates are required, as their components can offer field-proven performance.
With data center cooling technology evolving so rapidly, you need a supplier that can deliver components on short lead times. A supplier with robust engineering resources and facilities is more likely to be able to deliver components on time, even under demanding schedules. And when your design needs continue to change, you’ll want to have the support of a responsive team who can quickly modify existing components to ensure that you always have a working solution in hand.
Scalability should also be considered. Don’t let a supplier’s production limitations hamstring your timeline: choose a supplier who regularly supports industries with high volume production requirements. Suppliers with this type of experience are more likely to have the equipment and resources needed to scale with you. If you’re unsure, ask the supplier how they can guarantee that they will meet the production volumes you’re expecting based on your schedule.
Data center infrastructure is expensive, and component failure in liquid cooling systems can cause problems ranging from minor inefficiencies to the catastrophic damage caused by leakage. Building your system with high quality components will help minimize risks and maximize your computing power. When evaluating a supplier’s level of quality, ask about the manufacturing quality controls they use and consider how rigorous their component testing is. How does the supplier ensure that their components will perform as promised, with no part-to-part variations in performance? Suppliers who have experience working in heavily regulated industries such as the automotive or aerospace markets are more likely to have even stricter production measures in place.
Having the support of a knowledgeable and responsive engineering team will also help ensure that you can meet your system’s cooling requirements efficiently. Ideally, the supplier should act as an engineering partner who can address your questions and even recognize potential performance issues before they happen. For example, the right supplier can provide mounting options that easily integrate into your system and components that won’t have material compatibility issues with your fluids or system materials.
Liquid cooling in data centers is a relatively new technology. Opportunities to improve the efficiency and performance of these systems, such as responsive chip-level cooling technology, are continuing to emerge. A supplier who is knowledgeable about the latest microfluidic control strategies for these systems and has experience supporting other critical fluid control systems is an asset to any engineering team in this space. Consider not just whether the supplier has customers in the industry, but how actively involved they are – do they participate in industry events, organizations, or forums?
The Lee Company is proud to support the industry leaders and organizations developing the next generation of cooling technology for data centers. For over 75 years, we’ve partnered with customers in the aerospace, automotive, and industrial markets to solve their toughest fluid control problems. With our vast component catalog, it’s likely that we already have the solution you need available or can quickly modify a standard component to meet your needs. We regularly support customers with tight timelines and can typically deliver, subject to application complexity and technical requirements.
We have multiple solutions available that specifically address the needs of the data center cooling industry by offering precise, stable flow performance at low pressure drops in the smallest possible size. Our components are manufactured under the strict quality standards developed for markets such as aerospace and automotive and are 100% tested for flow performance. With our state-of-the-art manufacturing facilities and experience supporting high volume applications, we can easily scale production to meet your needs on schedule.
Our engineers have extensive microfluidic control experience and are familiar with the unique needs of liquid cooling systems for data centers. We specialize in working on an engineer-to-engineer level to provide solutions that help our customers optimize performance in their systems. Our team is ready to work with you to ensure that your liquid cooling system is as efficient and effective as possible. Contact us today to get started.
Always verify flow calculations by experiment.
*There are many parameters to consider when determining V-Factor. Click here for more information.
