I am an open weights model proponent so obviously I am intrigued and glad that Qwen exists (though at the moment I personally use US-trained open weights models — Muse Glimmer, Gemma 4, and I am testing Nvidia’s own Bolt —because they better fit my requirements).
I am aware that there’s a tension there, but the fact that I am glad the Chinese open weights models exist doesn’t change my mind about what Nvidia will almost inevitably have to do about Jensen Huang (or my own view of him in the scenario where he is supposedly naive about espionage!)
I suppose it colours my judgement about his actions re: Megaspeed if they were indeed a deliberate attempt to push availability of GPUs to Chinese customers to the limit; Singapore-washing is not something I am personally that troubled by.
alt Hacker News
I am an open weights model proponent so obviously I am intrigued and glad that Qwen exists (though at the moment I personally use US-trained open weights models — Muse Glimmer, Gemma 4, and I am testing Nvidia’s own Bolt —because they better fit my requirements).
I am aware that there’s a tension there, but the fact that I am glad the Chinese open weights models exist doesn’t change my mind about what Nvidia will almost inevitably have to do about Jensen Huang (or my own view of him in the scenario where he is supposedly naive about espionage!)
I suppose it colours my judgement about his actions re: Megaspeed if they were indeed a deliberate attempt to push availability of GPUs to Chinese customers to the limit; Singapore-washing is not something I am personally that troubled by.