SPRY356 October   2026 ADS7142 , DRV8426

 

  1.   1
  2.   en-us Overview
  3.   At a glance
  4.   Introduction
  5.   Comparing today's thermal management architectures
    1.     Air-to-air
    2.     Air-to-liquid
    3.     Liquid-to-air
  6.   Evolution of liquid cooling technology
  7.   The need for leak sensing
  8.   Immersion cooling
  9.   Future AI data center cooling trends
  10.   Conclusion
  11.   About the author
  12.   Additional resources
  13.   Trademarks

At a glance

  • Rising heat loads: The rack power density in artificial intelligence (AI) data centers has climbed from about 5kW to more than 100kW, with some AI clusters expected to reach 1.5MW.
  • Cooling architecture evolution: As racks exceed 100kW, data centers are moving from air-based cooling to liquid cooling and immersion cooling.
  • Predictive thermal management: Server-tray-level liquid flow control and edge AI-based sensor fusion will help data centers detect leaks faster and predict failures.