And why the US and China are fighting over silicon in the first place.Subscribe and turn on notifications 🔔 so you don't miss any videos: http://goo.gl/0bs...
What people are missing is that China doesn’t need to make the best semiconductors, at least not at first, since there already exists a market for them and will buy them up at any performance level, and not just in China. Does Iran or Russia or Venezuela care if Chinese semiconductors are only 80-90% as good when the alternative is no chips?
Exactly, even 20nm semiconductors work just fine for the vast majority of applications. the main advantage of 5nm and lower is in reduced power consumption, and there are a few domains, like mobile devices, where this is relevant. Given that China can already produce 5nm chips, seems to me that they’ve effectively closed the gap with the west in practical terms.
The other aspect of all this is that we’re now running into physical limits of what’s possible using silicon substrate. There is herculean effort involved in making 1-2nm chips for diminishing gains, and then there’s no room to grow after that. So, western companies don’t really have room to grow going forward.
Incidentally, China is researching alternative computing substrates such as graphene, and Chinese researchers even managed to produce a 12 inch wafer using MoS2 substrate already. I think this is the real path forward, and it could make silicon look like vacuum tubes overnight. Even a crude chip on a different substrate could have far better performance than anything possible on silicon with decades of optimizations to follow.
I wouldn’t say there is no room to grow after 1-2nm chips, I’d say that is where things get interesting, since we could be working with science-fiction level technology after that point. But I do agree with there being diminishing returns at first.
There’s no room with silicon because you start getting into problems like quantum tunnelling which makes it basically impossible to isolate signals on the chip. It’s basically like trying to make an ever smaller vacuum tube, at some point you hit a limit. Where things get interesting is where we move away from using silicon. There are a number of substrates that we know are possible, and have been shown to work in the lab. It’s not a question of whether it’s possible, but just a question of putting this stuff into mass production. This is why that MoS2 wafer is so exciting. It’s the first alternative substrate that looks like it might be possible to mass produce in the near future.
What people are missing is that China doesn’t need to make the best semiconductors, at least not at first, since there already exists a market for them and will buy them up at any performance level, and not just in China. Does Iran or Russia or Venezuela care if Chinese semiconductors are only 80-90% as good when the alternative is no chips?
Exactly, even 20nm semiconductors work just fine for the vast majority of applications. the main advantage of 5nm and lower is in reduced power consumption, and there are a few domains, like mobile devices, where this is relevant. Given that China can already produce 5nm chips, seems to me that they’ve effectively closed the gap with the west in practical terms.
The other aspect of all this is that we’re now running into physical limits of what’s possible using silicon substrate. There is herculean effort involved in making 1-2nm chips for diminishing gains, and then there’s no room to grow after that. So, western companies don’t really have room to grow going forward.
Incidentally, China is researching alternative computing substrates such as graphene, and Chinese researchers even managed to produce a 12 inch wafer using MoS2 substrate already. I think this is the real path forward, and it could make silicon look like vacuum tubes overnight. Even a crude chip on a different substrate could have far better performance than anything possible on silicon with decades of optimizations to follow.
I wouldn’t say there is no room to grow after 1-2nm chips, I’d say that is where things get interesting, since we could be working with science-fiction level technology after that point. But I do agree with there being diminishing returns at first.
There’s no room with silicon because you start getting into problems like quantum tunnelling which makes it basically impossible to isolate signals on the chip. It’s basically like trying to make an ever smaller vacuum tube, at some point you hit a limit. Where things get interesting is where we move away from using silicon. There are a number of substrates that we know are possible, and have been shown to work in the lab. It’s not a question of whether it’s possible, but just a question of putting this stuff into mass production. This is why that MoS2 wafer is so exciting. It’s the first alternative substrate that looks like it might be possible to mass produce in the near future.
Even western science fiction starts to look dated when you read a but about these new materials in initial development.