Pfftwell, fair enoughthe ride never ends. Land was right about technocapital, it's just surprisingly state-directedZephyr: Good pointCXMT will generate over $130B in gross profit in the…
This is a funny bit but 1) Kremlinology also sucked2) Even though it did, I think it was vastly higher quality than the current China Hand…
DeepSeek is a bit special in that things they release are often not knock-your-socks-off in papers, but they *work*. V4-Flash demonstrates this forcefully. It's a no-bullshit…
DeepSeek is a funny company because while they're doing tremendously much for open source, it's not pandering to the "local AI" and "normal people" with 4090s.…
Updates ??Stranger Ranger: There has never, in recorded history, been a war wherein one side lost its entire chain of command—including its supreme political and religious…
RT Sebastian RaschkaI put together a new article on setting up local coding agents with open-weight models. Everything runs 100% locally.I thought it might be useful…
RT Maxime LabonneFun surprise: DeepSeek used my open-perfectblend dataset to train their new DSpark drafterTime to promote it again! It's an open-source reproduction of "The Perfect…
Honestly this makes the whole benchmark look even more absurd. Grok 4.20 over Opus 4.8 (max), Kimi K2.5 > GLM 5.2 and Opus 4.7, Opus 4.6…
1) burger cope, ASML owns plenty of current IP2) China has a light source prototype already, incidentally developed by ex ASML head of light source3) none…
Yes.And this means choosing Chinese AI over MAGA MaoismAlexander McCobin: @pmarca More and more companies will need to choose between the an authoritarian communist approach to…