The management of data centers encounters numerous persistent challenges, one of the most pivotal being efficient cooling. Cooling significantly impacts overall performance and costs, making it a complex, energy-intensive, and costly aspect of data centers. With ever-rising computing demands and increasing rack density, heat generation escalates while the available free space for effective cooling diminishes.
Developing and implementing effective cooling in data centers is no easy feat. The entire industry is actively striving to devise better solutions, alongside significant contributions from major government institutions worldwide. Enter COOLERCHIPS – an initiative by the US Department of Energy that is anticipated to have significant implications for the global data center industry. Let’s delve into the core details and innovative elements of this project.
Understanding COOLERCHIPS
COOLERCHIPS is an initiative under the Advanced Research Projects Agency–Energy (ARPA-E) and stands for Cooling Operations Optimized for Leaps in Energy, Reliability, and Carbon Hyperefficiency for Information Processing Systems. This substantial funding project, announced by the Department of Energy in late June 2023, aims to advance data center cooling technologies over the next one to three years, with each project targeting a minimum tenfold improvement in cooling efficiency.
Cooling constitutes about 40% of a data center’s electricity consumption and already contributes to 2% of total US electricity use. The impending widespread adoption of artificial intelligence (AI) is projected to dramatically increase data center usage, subsequently accelerating heat generation, computing demand, and energy consumption, thereby necessitating better cooling solutions.
Challenges with overheating servers existed even before the era of AI. For instance, processor thermal designs are expected to reach 500W by 2025, and GPUs are already approaching 700W. The intensified need for enhanced cooling systems poses a significant hurdle to achieving maximum power and efficiency. Given these challenges, COOLERCHIPS aims to address these concerns within three years while ensuring sustainability.
Initiatives by Grant Recipients
Among the 15 grant recipients are prominent companies and institutions, including Intel and Nvidia. Nvidia’s substantial grant of $5 million targets an innovative approach merging two widely used cooling methods: direct liquid cooling and immersion cooling. The initiative seeks to optimize these methods in a confined setup, resulting in a compact, double-cooled server suitable for traditional racks without any further modifications.
Intel’s $1.71 million grant involves a two-phase immersion cooling system that utilizes 3D printing for creating highly efficient, nature-inspired structures. This coral-like heat sink, integrated within a 3D vapor chamber, is designed to minimize thermal resistance and improve overall cooling performance. The resultant water vapor helps dissipate heat from the chips into the outside air, enabling enhanced cooling and energy reduction.
Additional Cool Initiatives
Other companies, such as HP, Flexnode, and the University of Texas at Arlington, are working on unique liquid and hybrid cooling solutions to tackle various data center cooling challenges. These diverse projects cater to a wide array of data center types and cooling scenarios, covering new, old, edge, and modular setups.
By successfully implementing these initiatives, the hope is that the majority of these projects will effectively address the critical cooling needs prevalent in today’s data centers.