5nm Barrier Broken

[thebirdybird]

Thought i'd share something cool. Someone, figured out how to go past 5nm.

Scientists just developed the world's smallest transistor

 

[Shintai]

The problem is to make it economically viable. Not to develop it.

 

[Ruoh]

The problem is to make it economically viable. Not to develop it.

Chicken and the egg.

 

[Shintai]

Chicken and the egg.

Not really. The problem already exist with 16/14nm that kicked a huge chunk of companies off the road. 10nm will reduce the amount that can afford to get it manufactured even more, 7nm even further and so on.

To make an example, a 28nm IC design cost around 50M$. At a real 5nm for example, that price may currently end up around 800M$. And then its not even an optimized one. For that we may talk about 4 billion $. On top of that you have to add R&D etc for development.

 

[Ruoh]

Not really. The problem already exist with 16/14nm that kicked a huge chunk of companies off the road. 10nm will reduce the amount that can afford to get it manufactured even more, 7nm even further and so on.

To make an example, a 28nm IC design cost around 50M$. At a real 5nm for example, that price may currently end up around 800M$. And then its not even an optimized one. For that we may talk about 4 billion $. On top of that you have to add R&D etc for development.

Think about the companies that would be using this. You think they'll let $4Billion+ in R&D stop them?

 

[Shintai]

Think about the companies that would be using this. You think they'll let $4Billion+ in R&D stop them?

There is only one single company that can possible do that for an IC design. That's Intel.

Remember R&D for the uarch etc comes on top of that price.

For a 1 billion $ IC design cost, you need about 10 billion $ revenue on that single chip. This is why companies are falling off left and right or reduce their amount of chips significantly.