AI for Good blog

How immersive liquid cooling and AI are turning data centers green

Energy | Environment & Climate change

By Erin Kalejs 

Data centers are the beating heart of the digital economy. If you don’t know what they are, here’s a quick description: data centers are the factories of the “Digital Era”, “the hidden engine” behind any mobile, PC, IoT, where computers are used to collect, store and manage data. Data in the cloud doesn’t exist somewhere in the sky, and unfortunately there are yet no clean “data clouds”.  

Using air cooling, data centers require an enormous amount of energy and use of natural resources like water to run smoothly. Today, estimates of that amount vary greatly depending on the source from 1% to 15% of the world energy consumption. According to the International Energy Agency, data centers use approximately 200 terawatt-hours (TWh) of electricity, or nearly 1% of global electricity demand, contributing to 0.3% of all global CO2 emissions. 

No matter the sources one looks, overall, it’s quite an inefficient system in place that struggles to keep up with the rise of computing needs and big data (IDC predicts that the collective sum of the world’s data will grow from 33 zettabytes this year to a 175ZB by 2025, for a compounded annual growth rate of 61 percent). It is clear that drastic action must be taken in order to reduce data centers’ energy consumption on a global scale.  

The good news is that green technologies such as Immersion4 in combination with AI, are changing the way data centers operate to make them more sustainable based on four pillars; no water consumption, no CO2 emissions, no GHG (greenhouse gas) emissions and ethics. It sounds almost unattainable, doesn’t it? In a recent AI for good webinar, “Energy saving vs sustainable AI and data” Immersion4’s Founder and CEO, Serge Conesa explained how it’s happening.  

The driving forces behind air vs immersive liquid cooling 

The major issue impacting traditional data centers is their inefficient use of energy. This is based on the simple fact that we cool electronics using air. Air is an insulator. We need to change the paradigm. As long as we use air, we need to have special dedicated cooling infrastructure to protect the electronics and the PCBs against heat, external events, such as dust, corrosions, oxidation, humidity, temperature swings, and so on and so forth.” 

“When it comes to energy consumption within data centers, 60 to 70% goes to cooling and 30 to 40 at most goes to the IT load which is at the end of the day a big loss of energy,explains Serge Conesa. “Any data center today is being paid for data resiliency not for energy efficiency.”

Immersion4’s simple yet effective solution to reducing data centers overall harmful environmental impact can be summed up in two words: liquid cooling. Thus, eliminating the need for air cooling and its dedicated infrastructure and offers the opportunity to use the thermal waste from the datacenter to reduce the carbon footprint of the built environment: “simplicity is the highest form of sophistication”.  

Immersion4 uses a “Liquid to LiquidTM” thermal transfer combining “DTMTM systems using ICETM  coolant” connected to a sealed closed water loop. This is paradigm shift, no more air temperature control needed in the datacenter yet only liquid temperature control.  Electronics of any kind immersed within the ICETM coolant liquid circulating and collecting the heat generated allows “any IT Load to operate no matter if the servers are overclocked, and independently of any kind of application you’re using.” 

Using liquid instead of air to cool IT Load solves a ton of all problems (such as humidity, corrosion, oxidation, dust, electrostatic, temperature swing, whiskers, to name a few) providing a stable temperature environment control which is crucial for the longevity of any electronics. 

Another big challenge towards achieving sustainability is the problem of E-waste.  

“One of the crucial problems we are facing today is E-waste,” he says. “If we use Immersion4 technology, generally by immersing any kind of electronic components with the connectivity, electrical data, and fiber optics we basically do not need all those protections called conventional coating and brominated chemical layers which are needed when we use air cooling.” 

He added, “I am looking forward to the day when we create PCB’s and components that are dedicated to liquid immersive cooling because that will be the day that we make E-waste history allowing urban mining recycling anywhere in the world.” 

Watch the replay below to find out more about enhanced efficiency and reduced energy consumption in the field of AI. 

Technology for sustainable energy consumption and conservation 

While using and creating renewable sources of energy is crucial, Conesa pointed out that conservation is key. “The first principle that drives Immersion4 is Eco conservation, using energy for what matters and what matters in the datacenter is the IT load not the cooling. It’s also not about using dedicated building because any building should be able to become a data center. For me, there is no smart city without smart energy”. 

Additionally, air cooled data centers require “creating new power plant, upgrading the grid and renewable energy is not enough in term of quantity, resiliency and consistency which is why there is not a single data center today running only on renewable energy.” 

AI is also playing a role in making data centers more sustainable. AI-powered software can help data centers more effectively manage their infrastructure, maximize the use of their CPUs and deliver significant energy savings.  

New processors for data centers also play an important part as they increase computing power without increasing energy consumption. As Conesa mentions “the issue is not only about a processor being too hot. Today CPU, GPU, FPGA, ASIC, memory, etc have become heat sensitive impacting performances and overclocking capability. Using immersive cooling allows tremendous savings in energy consumption and dedicated cooling infrastructure.  

When we solve the core problem which is cooling electronic, AI can be used efficiently. Combining data collection, optimized application software using the right algorithms, virtualization to manage data with the needed processing contribute greatly to enhance sustainability.  These are the keys to have a complete sustainable chain from the electronics component to the applications.” 

Embracing innovation to create change 

In the AI for Good interactive Q&A session with the audience, a question came up: how compatible is Immersion4 with current servers? which Conesa replied: “Any server, any router or switch is compatible with Immersion4being able to take the legacy of any servers from any brand is very important. The only modification you will have to make will be to replace the hard drive with an SSD hard drive, because a mechanical hard drive doesn’t work with the liquid so the performance will be greatly degraded. He adds, “the good news is, because of the market trend all servers today are equipped with SSD hard drive.” 

“The only thing we need to change right now is our behavior because the technology is here, Serge Conesa.

Advanced cooling technologies go further with AI. By embracing and applying it to redesigning true sustainable data centers the possibilities are endless since, “Immersion4 DTMTM technology is climate and building agnostic for any application from EDGE to HYPERSCALE”.  

Conesa ended the session by pointing to the elephant in the room “it is not you and me or our kids who are guilty for using their phones and consuming data, it’s the big engine on the back – the data factory aka the data center.” 

Watch AI For Good’s latest webinar “Saving the energy equivalent of 1500 nuclear plants with green tech” to learn how green technologies can help reduce the world’s energy consumption.