Hacker News
new
past
comments
ask
show
jobs
points
by
mattlondon
1 hours ago
|
comments
by
ianm218
48 minutes ago
|
[-]
GPUs are much more efficient at parallelizing requests for LLMs so it's going to much more efficient to centrally host. Maybe big companies it would make sense to get their own though.
reply