Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about Google's groundbreaking contribution to liquid cooling infrastructure through this 20-minute conference talk from the Open Compute Project. Discover how Google has successfully deployed liquid cooling at gigawatt scale across over 2000 TPU Pods over seven years, achieving remarkable 99.999% uptime since first implementing liquid cooling in TPU v3 in 2018. Explore the technical details and strategic context behind Project Deschutes, Google's fifth-generation cooling distribution unit that the company is contributing as an open specification to accelerate industry-wide adoption of liquid cooling solutions. Gain insights into critical design considerations including high cooling performance optimization, manufacturing quality standards, reliability and uptime requirements, deployment velocity strategies, serviceability protocols, and operational excellence practices. Understand how physical infrastructure—encompassing power, cooling, and mechanical systems—serves as the critical foundation for AI's continued scaling in modern data centers, and learn about the supply ecosystem advancements that make large-scale liquid cooling deployments possible.
