i’m starting to really understand why those that can afford it are building local llm/ai servers now. after a while a bunch of this stuff really starts to add up on the cost front so bringing a bunch of that local and inhouse makes a big difference, i’m starting to route requests to different models
Originally published at https://sett.ee/2026/04/15/im-starting-to-really-understand.html
pinterest epic wins pinboard → brand advocate for nokia, 1000heads, verisign → won vloggie for node666 (san fran 2006) → television for time team history hunters 1999 (burton on trent) → sold me.dm to evan williams in april 2011 → went to phil campbell, alabama to help raise money after tornado ripped up the town (was on sky news, bbc news)→ CNN for sxsw 2013 about austin south by southwest event → video chat with robert scoble from rackspace → music video can you spot me? → won the digital derry contest for 5k euros → crowdfunded digital signage concept called pi street → now living life through digital blockchains.