Post

Matt Harding
@hardingmatt.bsky.social

Love it - What spec are we talking (and suprised they are not Mac users who get the super unified memory)? I explored eGPU for this - just too expensive right now but someday Jennifer. Someday...

Matt Harding
@hardingmatt.bsky.social

Nice - And vendors are currently charging at a price which people are willing to pay at least (rather than actually being cheap) - My concern is there is an AI bubble and if it pops, I'll need local so I don't have to relearn how to code 😉! (plus be a bit greener)

Marian Zeis
@marian.zeis.de

My guess it will more slowly deflate or go the Enshittification route. Also the gap between local model and codex/Claude models is closing. Let's see, nobody knows really...

John Patterson
@jasper07.secondphase.com.au

I wonder if enterprise developers will opt for local small language models that are more accurate and cost-effective for specific workloads, buying something like a DGX Spark seems like a nice NUC upgrade.