Post

I love some good ol-fashioned data center porn But I have some questions engineering.fb.com/2026/10/19/d...

They started with GB200 NVL72 racks with the following stats

They upgraded to GB300 NVL72 racks with a bullet point that it has over 1 exoFLOP of FP4 compute. The GB200 rack says it has 1.4 exoFLOP of FP4 compute they just wrote it as "1,440 PFLOPS"

Are you telling me that last year with GB200 12% of your GPUs were "interrupting" every day??? The stat says 1000 failures per 8000 GPUs per day. That's 1 in 8. IDK about you, but to me that sounds like a completely broken product

This kinda kills me. They buy racks of 72 GPUs but only use 64 (8 spares) because the jobs work better with powers of 2. An 11% stand by capacity is a lot of money They also have 2% extra racks just waiting around

Their goal is 1M GPUs and let's just use 72 GPU/rack as a round number. That means they have 13,888 active racks and 277 SPARES!!! A GB300 NVL72 rack costs about $5,000,000 which means they have (list price) $1.3B of spare capacity

But gpu has a relatively high burn/failure rate, so without redundancy, they will need to repair/decommission full rack

For business use, the most important part is whether it is profitable or not