How much of this capex is infrastructure that can support future generation of compute vs current-gen GPUs that will become obsolete?
How much of this capex is infrastructure that can support future generation of compute vs current-gen GPUs that will become obsolete?
I can't find the reference now unfortunately, but I read it was about 50:50, so half is the data center land / building / cooling / power generation, and half is in the GPUs themselves.
This is notably very different from the dot-com build out of dark fibre, where digging the holes cost the vast majority of the money, and the fibres and network equipment cost very little in comparison.