Compute and models are becoming political assets
China published its first frontier AI export controls (MOFCOM Announcement No. 48). Commenters say they largely match earlier guidance: restrictions on top models with an exemption for open-source collaboration. One notes the irony that Chinese models gained popularity precisely because they are open. A satirical post, 'Sam Altman: All the RAM Should Belong to Me, So I Can Charge People to Use It', points at fear of memory and compute being hoarded.
OccupySaaS, a protest-style idea against SaaS pricing, got a lukewarm reception. One reaction called it Big Tech's 'we demand you use AI, not like that' moment, and another doubted anyone could keep a coalition aligned.
The common thread is that access to models and hardware is turning into leverage: states can restrict it, the biggest labs can hoard it, and customers complain but have few exits. Local models are the obvious hedge, which is why the local-coding thread matters here too.
So what?
Do not assume the open Chinese models you rely on stay available on today's terms. Keep your product portable across at least two model providers, and track memory and GPU supply as a cost risk.