How AI labs are solving the power problem
newsletter.semianalysis.com
How AI labs are solving the power problem
1–10 of 270 posts
Re: How AI labs are solving the power problem
#2I found Boom's pivot much less confusing after this article.
Re: How AI labs are solving the power problem
#3Power problem: solved
Natural Gas supply problem: worsened
Carbon in the atmosphere problem: worsened
Re: How AI labs are solving the power problem
#4What about renewables + battery storage? Does it take much longer to build? I can imagine getting a permit can take quite a long time, but what takes so long to set up solar panels and link them to batteries, without even having to connect them to the grid?
Re: How AI labs are solving the power problem
#5Our kids are not going to be happy we spun up more CO2 generation for this.
Re: How AI labs are solving the power problem
#6> An AI cloud can generate revenue of $10-12 billion dollars per gigawatt, annually.
Citation needed.
Re: How AI labs are solving the power problem
#7Part of what bothers me with AI energy consumption isn't just how wasteful it might be from an ecological perspective, it's how brutally inefficient it is compared to the biological "state of the art" — 2000kcal = 8,368 kJ. 8,368 kJ / 86,400 s = 96.9 W.
So the benchmark is achieving human-like intelligence on a 100W budget. I'd be very curious to see what can be achieved by AI targeting that power budget.
Re: How AI labs are solving the power problem
#8Power problem: solved Natural Gas supply problem: worsened Carbon in the atmosphere problem: worsened
Yeah, that headline made me think "Oh good, there's some solution on the horizon that won't require absurd amounts of electricity."
Not so.
Re: How AI labs are solving the power problem
#9Power problem: solved Natural Gas supply problem: worsened Carbon in the atmosphere problem: worsened
Coincidentally the USA is more than self sufficient in natural gas and is a net exporter. Drill baby drill!
Re: How AI labs are solving the power problem
#10Part of what bothers me with AI energy consumption isn't just how wasteful it might be from an ecological perspective, it's how brutally inefficient it is compared to the biological "state of the art" — 2000kcal = 8,368 kJ. 8,368 kJ / 86,400 s = 96.9 W. So the benchmark is achieving human-like intelligence on a 100W budget. I'd be very curious to see what can be achieved by AI targeting that power budget.
That’s about the energy a laptop or two uses at full tilt.