In 2024, the shift toward strategic collaborations with liquid cooling and infrastructure providers will provide streamlined, effective administration of dense computer clusters, making modular data centers an adaptable and efficient alternative for data center operators. This method not only fulfills the expanding demands of modern computing but also improves operating efficiency, decreases energy consumption, and allows for quick growth.
Fremont, CA: The combination of artificial intelligence (AI) and sophisticated cooling technologies is causing substantial adjustments in data centers. The most notable example is the rise of direct-to-chip liquid cooling, an intelligent approach redefining thermal management in high-density GPU clusters.
With the growing need for AI and its power-intensive computations, a substantial thermal problem emerges: efficiently controlling AI workload cooling. As a result, liquid cooling systems, particularly single-phase direct-to-chip cooling, are expected to be more widely used to manage the intense heat generated by tightly assembled server racks and high-density GPU clusters.
Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.
In 2024, the shift toward strategic collaborations with liquid cooling and infrastructure providers will provide streamlined, effective administration of dense computer clusters, making modular data centers an adaptable and efficient alternative for data center operators. This method not only fulfills the expanding demands of modern computing but also improves operating efficiency, decreases energy consumption, and allows for quick growth.
Direct-to-chip (D2C) liquid cooling is evolving from a luxury performance increase to a must-have for powering high-density clusters in colocation, edge computing, and federal applications.
This year, single-phase direct-to-chip liquid cooling is recognized as the cooling sector's frontrunner. To efficiently dissipate heat, cold plate systems include microchannels or microjets. Direct-to-chip and immersion cooling approaches have been competing for supremacy in the field of cooling technologies.
Single-phase immersion cooling is being improved to handle increased power densities. This method currently incorporates fans and various power devices into the oil tank.
Recent advancements have not attained over 1,000W of cooling capacity, and operating circumstances have yet to exceed chip manufacturers' compliance norms. For example, these experimental setups frequently result in chip temperatures reaching 80 degrees Celsius and necessitate high flow rates of roughly seven liters per minute. Despite improving performance at the expense of efficiency, immersion cooling has a market use case.
Two-phase liquid cooling, on the other hand, is gaining popularity as demand for AI-driven applications grows. However, this technology faces regulatory challenges as the industry transitions to more environmentally friendly and operationally effective cooling methods.
More in News