I read the article expecting the same thing, but it said that Microsoft, Nvidia, and Palantir recently released a big statement warning the US not to restrict open models.
Which is weird. I don't believe for a minute that you and I are on the same side as those jackals, but I guess they are not as dumb as I expected on this particular day (be sure to check again tomorrow, though).
This whole thing is stupid and I want this bubble to burst. I don't look forward to the economic consequences for people like my parents who live on retirement savings, but lord, this whole economic system is just totally cooked and it can't go on like this.
If they understand that there's no moat for their AI R&D business, perhaps they're looking to a possible future where they could lower their cost of R&D by picking up open models and running them for inference. Especially if the current spending collapses.
6 replies
Didn't Huang declare the war lost at the start of the year?
Incoming: US cries about how unsafe open models are and that they should be banned. That the only "Safe™®©" model is one you buy from Altman or whatever anthropics chief scumbag officers name is. AI bubble pops and "we" (read: you and me, Johnny and Jane public) have to bail them out to Save US Innovation™©® from China...
I read the article expecting the same thing, but it said that Microsoft, Nvidia, and Palantir recently released a big statement warning the US not to restrict open models.
Which is weird. I don't believe for a minute that you and I are on the same side as those jackals, but I guess they are not as dumb as I expected on this particular day (be sure to check again tomorrow, though).
This whole thing is stupid and I want this bubble to burst. I don't look forward to the economic consequences for people like my parents who live on retirement savings, but lord, this whole economic system is just totally cooked and it can't go on like this.
If they understand that there's no moat for their AI R&D business, perhaps they're looking to a possible future where they could lower their cost of R&D by picking up open models and running them for inference. Especially if the current spending collapses.
Qwen 3.8 27B dropping soon for all the lucky bastards with >=24GB VRAM.
It's a shame there's still no 3.8 MoE model that can run on consumer hardware. Still stuck at 3.6 if you need to leverage offloading experts to CPU