I got an F in Thermodynamics, shortly before I flunked out of UVA's Aerospace Engineering program.
So I leave matters such as cooling satellites to experts, and recently I came across a concern about data centers in space: can they scale to the size we need for AI? Forget issues of space debris and power. The Reverse Salient, in Hughes terms, appears to be cooling.
The article here discusses, in somewhat technical terms that should still be understood by the non-engineer, the reasons that what engineers call "radiational cooling" in space may not work for the amount of heat generated by orbital data centers. Perverse thought, no? Space is either hideously hot in the sun or deadly cold in the shade. But the problem involves moving heat away from the object in question. There's no breeze in vacuum; any water would boil off or freeze solid. Read the article for more technical details.
This year we will find out if orbital data centers can scale. I hope so. We simply do not have the energy resources or land for the types of AI economy the billionaires envision. The Carbon pollution alone might seal our fate.
NVIDIA, many other US firms, and the Chinese space program, are testing the concept now or will be this year. I wish them luck. We need to get this right.
I'm increasingly of the belief that we'd be better off had AI not been invented, a thought I also have about smart phones and social media, but like those technologies AI is not going to be wished away.
Image source: Daily Galaxy, Creative Commons image
No comments:
Post a Comment