Microsoft Tests Microfluidic Cooling Technology to Enhance AI Chip Efficiency

Microsoft Tests Microfluidic Cooling Technology to Enhance AI Chip Efficiency

Microsoft is validating microfluidic cooling technology in data centers by etching tiny channels on the surface of chips, allowing the cooling liquid to directly contact the heat dissipation area. Compared to traditional air cooling and liquid cooling, this method can efficiently dissipate heat even at high temperatures of 70°C while significantly reducing energy consumption.

This technology has already been successfully tested on server chips and AI GPU prototypes, and it may support stacked chip designs in the future, further unlocking performance potential. Additionally, Microsoft is exploring hollow fiber technology to accelerate data transmission and is collaborating with companies like Corning to expand the production capacity of related materials, optimizing data center operational efficiency from both cooling and transmission perspectives.

Source: Bloomberg

Related posts

  1. Seamless Transition Between Swimming and Flying: Students Create a 3D-Printed “Submersible Drone”
  2. Alibaba’s Breakthrough in Self-Developed AI Chips: A Booming Opportunity for the Domestic Computing Power Industry Chain
  3. From ‘General’ to ‘Customized’: AI ASIC is Redefining Chip Competition Rules
  4. Understanding Linux Network Device Drivers (Part 6)
  5. The New Favorite of AI Chips: An Overview of the AI ASIC Chip Industry Chain
  6. How Are Billions of Transistors Packed Inside Chips?
  7. Medical Robotics Technology and Development Trends
  8. An Attempt to Modify GCC Source Code on Windows Platform to Another Language!
  9. Nano Letters – Monolithically Integrated Metasurface on a PCSEL for Depth Perception – National Yang Ming Chiao Tung University
  10. Summary of the Interchangeability of Arrays and Pointers in C Language
  11. Engineering Crop Flower Morphology Facilitates Robotization of Cross-Pollination and Speed Breeding
  12. What Are the Advantages and Disadvantages of the ARM mbed Platform from a Developer’s Perspective?
  13. Technical Explanation of 16-bit and 32-bit Segments in Assembly Language
  14. Dual-Mode BLE 5.0 Bluetooth Module for Wearable Products
  15. Sky-High Costs! The Development Cost of a Chip from Design to Tape-Out Reaches 4.5 Billion!
  16. Confirmed Risks of ‘Backdoors’: The U.S. Embeds Trackers in AI Chip Shipments
  17. Introduction to the Domestic EtherCAT Slave Chip FCE1100 and FCP32C335 DSP Chip Function Board
  18. felix86 Emulator Update: RISC-V Processors Can Run AAA Games Like The Witcher 3
  19. Current Status and Prospects of Temperature Sensor Applications Based on WiFi Technology

Leave a Comment