Did Jensen single-handedly pop the AI bubble with his juang?

  • frongt@lemmy.zip
    link
    fedilink
    English
    arrow-up
    0
    arrow-down
    1
    ·
    11 days ago

    As you observed, it’s about the efficiency. A top of the line GPU can fit the biggest, latest models, or it can train a bigger model with more parameters. The field is moving so fast that this makes a big difference.

    Remember that any server has a limited number of slots, too. You can’t keep adding 8gb GPUs to match whatever number of 64gb GPUs.