Altman says local inference is 'terrible allocation' of scarce DRAM, community pushes back
TheZachMueller · x · 2026-10-30
Sam Altman argued that running open-source models locally is "a terrible allocation of scarce resources": the world's limited DRAM sits in personal computers used 5% of the time, and in a supply-constrained environment utilization should be maximized in the cloud.
The take immediately drew mockery — critics framed it as "the man who wants to sell you inference telling you to stop doing local inference," noting the conflict of interest and ignoring local models' real benefits like privacy, offline use, and no subscription.
More from Fun
- Tech Halloween: AI-Generated Berkshire Hathaway Trio Photo Goes Viral — johncoogan · 2026-10-31
- A robot arm trained to rebel flips off visitors at SF's Scaling Intelligence AI art show — charles_irl · 2026-10-31
- Anthropic 'kitten article' backlash didn't go the 'cross-ideological dating' route critics expected — mimi10v3 · 2026-10-31
- Strategy games are unsolvable puzzles: masters approximate with an in-head engine — BraydonDymm · 2026-10-31
- Claude $200 plan axed yesterday, back today: users ride the mood swing — kimmonismus · 2026-10-31
- Why 'AI researchers' is an apt name: they're only artificially intelligent — eigenron · 2026-10-31