On: Nvidia’s $589B DeepSeek rout

[chpatrick]

If the demand is for reaching AGI then we'll still need all the GPUs NVIDIA can sell, we'll just get there faster thanks to DeepSeek.

[CamperBob2 → chpatrick]

Bearish argument: "How can we adapt you to run on 8,000,000 NVidia GPUs instead of 80,000,000?"

That probably won't be the first question we ask AGI if/when we ever get there, but it will be near the top of the list.

[Jlagreen → CamperBob2]

Nvidia is already doing that.

What needed 1000k of Voltas, needed 100k of Amperes, needed 10k of Hopper, will need 1k of Blakwell.

Nvidia has increased compute by a factor of 1 million in the past decade and it's no where near enough.

Blackwell will increase training efficiency in large clusters a lot compared to Hopper and yet it's already sold out because even that won't be enough.