Comment by bobxmax
Who cares? Are you not seeing the exponential decrease in GPU inference costs? H100s costs 3x what they did a year ago.
Who cares? Are you not seeing the exponential decrease in GPU inference costs? H100s costs 3x what they did a year ago.
I'm not sure I understand how B (H100s costs 3x what they did a year ago) follows from A (exponential decrease in GPU inference costs).