Array
(
[content] =>
[params] => Array
(
[0] => /forum/threads/aws-ceo-matt-garman-said-today-there-is-so-much-demand-they-have-never-retired-an-nvidia-a100-gpu.24492/
)
[addOns] => Array
(
[DL6/MLTP] => 13
[Hampel/TimeZoneDebug] => 1000070
[SV/ChangePostDate] => 2010200
[SemiWiki/Newsletter] => 1000010
[SemiWiki/WPMenu] => 1000010
[SemiWiki/XPressExtend] => 1000010
[ThemeHouse/XLink] => 1000970
[ThemeHouse/XPress] => 1010570
[XF] => 2030871
[XFI] => 1060170
)
[wordpress] => /var/www/html
)
Guests have limited access.
Join our community today!
Join our community today!
You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
AWS CEO Matt Garman said today there is so much demand they have NEVER retired an Nvidia A100 GPU
- Thread starter soAsian
- Start date
It's little like fully depreciated fabs - as long as GPU time / tokens are priced right, Amazon keeps on making money. In some ways better than depreciated fabs because main cost then is variable (energy) and is somewhat proportional to utilization.
• Most AWS servers and networking equipment: depreciated straight‑line over 6 years.
• AI‑intensive servers and networking gear (GPU/accelerator systems, associated high‑speed switches, etc.): depreciated over 5 years, with some older AI gear being written down even faster via accelerated depreciation
• Most AWS servers and networking equipment: depreciated straight‑line over 6 years.
• AI‑intensive servers and networking gear (GPU/accelerator systems, associated high‑speed switches, etc.): depreciated over 5 years, with some older AI gear being written down even faster via accelerated depreciation