@MrEnyecz

This video didn't age well. We have the studio with 512GB RAM, Framework Strix Halo with 128GB RAM, but x86, yet there is no DIGITS yet and it will have jut 128GB RAM. For a ton of money.

@bdhaliwal24

The memory bus size is 1/2 of Apple, so large models will be painfully slow anyway.   They would never canabalize their data center cards with tgis

@Azaurus1

The most common use case i see is the brains of a house robot.

@gdotone1

This cost how much? 3000. Not selling a bunch

@arunachalpradesh399

lol 128gb ram with super computer , no intrest , it will able to run only small models may be deepseek 70B. presently apple M3 ultra studio with 512gb is still the best which can sit on desk. with some tweak and setting , it can run deepseek r1 671B, not super fast but atleast 12 token per secound, which is faster 2 times faster than human writing.