| ▲ | rvnx 3 days ago |
| Long term consequences: China outperforms Nvidia, by producing cheaper, faster chips at a large scale, by getting inspired by the IP but using their own production lines. Through sanctions, the irony is that the west removed the incentive for China to respect IP laws. Well done. If they can solve the lithography/ASML issue by getting access to it, then they will be forced to win. |
|
| ▲ | palmotea 3 days ago | parent | next [-] |
| > Long term consequences: China outperforms Nvidia, by producing cheaper, faster chips at a large scale, by getting inspired by the IP but using their own production lines. Unlike your typical free market fanboy, the Chinese leadership isn't stupid. They were always planning to do that, sanctions or no. Realistically, all sanctions can do is mess with their timelines for some temporary strategic advantage, slowing some things down and forcing reallocation of investment away from other areas into the sanctioned areas. The US refraining from sanctions is likely the stupid move, because that lever of control will expire at some point. To not use it is to squander it. But if there's one thing the US government and its business elite is good at, it's squandering things. |
| |
| ▲ | CamperBob2 3 days ago | parent [-] | | "Planning to do it" is one thing, but thanks to Trump's erratic and corrupt trade policy, they now have a Manhattan Project-level incentive to make it happen. It's ridiculous to think they won't succeed, just by dint of sheer numbers alone. | | |
| ▲ | palmotea 3 days ago | parent [-] | | > "Planning to do it" is one thing, but thanks to Trump's erratic and corrupt trade policy, they now have a Manhattan Project-level incentive to make it happen. The plans weren't wishes, they were things they were actively working on to make happen. The point is they didn't need "Trump's erratic and corrupt trade policy" to motivate it, they were already motivated to do it anyway. The US's problem is that its actions are uncoordinated. Sanctions and tariffs need to be coupled with massive investments to build new capabilities, and the latter is usually lacking. For instance, tariff revenue (and then some) should be poured directly into subsidies for building new facilities that support critical industries (like rare earths and electronics manufacturing). And things would probably be counterintuitively more effective if there was more tolerance of waste For instance, China's subsidized hundreds of solar panel manufacturers, none of them make money and a lot have probably failed, but the vicious domestic competition has helped them dominate that technology globally. The US freaked out in a massive scandal when one subsidized solar panel maker went out of business. | | |
| ▲ | CamperBob2 3 days ago | parent [-] | | The plans weren't wishes, they were things they were actively working on to make happen. The point is they didn't need "Trump's erratic and corrupt trade policy" to motivate it, they were already motivated to do it anyway. Yes, they were "actively working on it"; no, they had made little significant progress despite throwing tons of money at the initiative. There were lots of stories along the lines of https://www.nytimes.com/2021/07/19/technology/china-microchi... from the early 2020s, not so many lately. Their internal posture will now be the same as Russia's post-1945 push for the Bomb. Continued failure will (possibly literally) place heads at stake. The US's problem is that its actions are uncoordinated. They are coordinated well enough, but with the goal of magnifying Cheeto Benito's personal influence and cultivating his in-group's fortunes. | | |
| ▲ | palmotea 3 days ago | parent [-] | | > Yes, they were "actively working on it"; no, they had made little significant progress despite throwing tons of money at the initiative. That's how things sometimes go when you're building up a capability. I'm sure they were going to work through the setbacks, regardless. >> The US's problem is that its actions are uncoordinated. > They are coordinated well enough, but with the goal of magnifying Cheeto Benito's personal influence and cultivating his in-group's fortunes. No. That problem is bigger than the Trump administrations, focusing on him is lazy. | | |
| ▲ | CamperBob2 3 days ago | parent [-] | | No. That problem is bigger than the Trump administrations, focusing on him is lazy. It's absurd to say that without elaborating on how anyone else was "just as bad," which I expect will be a key part of your next reply. Trump is fucking bad, and if you disagree after all we've seen, you're either arguing in bad faith, or you're not such a great person yourself. He is costing us every jot and tittle of soft power we ever wielded as a nation. Trump is the living embodiment of the old cliché about how in the Chinese language, the words for "threat" and "opportunity" are similar. His actions have comforted Russia, alienated Europe, and galvanized China. | | |
| ▲ | palmotea 3 days ago | parent [-] | | > It's absurd to say that without elaborating on how anyone else was "just as bad," which I expect will be a key part of your next reply. > Trump is fucking bad, and if you disagree after all we've seen, you're either arguing in bad faith, or you're not such a great person yourself. He is costing us every jot and tittle of soft power we ever wielded as a nation. Sorry dude, all of that is coming from inside your own head. You're so blinded by Trump that you're incapable of having this conversation. I don't want to put in the effort to try to fix that. Have a nice day. |
|
|
|
|
|
|
|
| ▲ | codedokode 3 days ago | parent | prev | next [-] |
| I don't know much about GPUs, but is there really any value in IP? I can learn to write HDL code all day long, but turning it into real transistors is the hard part. Code is worth nothing nowadays with AI. |
| |
| ▲ | PunchyHamster 3 days ago | parent [-] | | not IP in particular but APIs. Putting your existing CUDA codebase into new NVIDIA generation that might be noticeably more expensive but is still faster/more energy-efficient than competition. Competition's software stack have to be good enough that it is worth migrating over and I think till we get some kind of cross vendor API for that it won't happen for a while |
|
|
| ▲ | 1970-01-01 3 days ago | parent | prev [-] |
| I thought it was impossible for them to leapfrog without actively occupying the TSMC fab as it takes years and years just to dial-in the insane precision. |
