Pushing AI System Cooling To The Limits Without Immersion
The Next Platform, Thursday, May 15th, 2025
Here is a question for you. What is harder to get right now: 1,665 of Nvidia's 'Blackwell' B200 GPU compute engines or 10 megawatts of power for a four year contract in the Northeast region of the United States?
Without question, it is the latter, not the former, and both will cost on the order of $66 million.
The fun bit is that those GPUs will probably actually take 13.4 megawatts of juice to operate in a GB200 NVL72 rackscale system configuration, and that means they will burn around 88.5 megawatts over four years. And if you don't need a rackscale coherent memory domain for the GPUs because you are using the GPU machinery for AI training instead of inference (which operates at a scale of tens of thousands of GPUs), you will burn about the same power but you can do it with twice as much space and half the power density.