[artemisart]

Correction: it's 8x the TFLOPS of a DGX (8 H100), not 1 H100. But it's true that if it stays at $3M it's probably too much and I don't think the memory bottleneck on gpus is large enough to justify this price/performance.

[Tepix]

So, the corrected statement is:

"56x the size of H100 but only 64x the performance improvement"

Doesn't sound too shabby.

[throwup238]

The company started in 2015 so I think they are (were?) banking on SRAM scaling better than it has in recent years.

[bee_rider]

If you have a problem that you can’t easily split up into 64 chunks, I guess it makes more sense, right?