Re:Meta building cloud business to sell excess AI capacity
by digdugdirk on Thursday July 02, 2026 @07:53AM
(#48760000)
It's an inherently different thing. Fibre is infrastructure. It'll still function decades later. GPUs at this scale are consumables. I've heard 3-5 years lifespan before they fail out. This might be a low estimate, but even if you double it - they're 50% of the cost of datacenters. We're flushing entire countries worth of economic value down an Nvidia shaped toilet.
Re:Meta building cloud business to sell excess AI capacity
by bwb on Thursday July 02, 2026 @08:02AM
(#48760105)
Anyone got any good testing data? Surely someone knows rough failure rate from constant usage?
Re:Meta building cloud business to sell excess AI capacity
by rwmj on Thursday July 02, 2026 @09:18AM
(#48761098)
It's not that they fail (which they also do), it's that they become obsolete. They consume more energy to do fewer operations relative to newer nodes. This is the basis behind Moore's law.
Re:Meta building cloud business to sell excess AI capacity
by bwb on Thursday July 02, 2026 @09:35AM
(#48761390)
Ah gotcha, ya that will be interesting to see, but does seem like a more likely life of 7 to 10 years is possible.
I am curious on actual hardware failure rates due to heat etc.
Re:Meta building cloud business to sell excess AI capacity
by rightbyte on Thursday July 02, 2026 @11:33AM
(#48763165)
I tried to do a calculation.
https://news.ycombinator.com/item?id=48762428
So 10% failure rate per gpu year. Some might be repairable and there might be a failure rate acceleration as time passes etc. So very approximate numbers.
Are we talkinga bout loosing 50% of the hardware as it fails? Or far less?