News ASML explores Hyper-NA chipmaking tools as the next step in shrinking transistors — tools would debut in 2030, but significant technology and cost...

[Admin]

ASML mulls Hyper-NA lithography with a higher than 0.7 numerical aperture in 2030s.

ASML explores Hyper-NA chipmaking tools as the next step in shrinking transistors — tools would debut in 2030, but significant technology and cost... : Read more

 

[yeyibi]

In the past, there were so many foundries, that when one got stuck, being unable to advance, some other foundry developed a new idea, and beat them, accelerating the advance of the technology.

Today the number of foundries is so small and expensive, that new ideas that think out the box got so rare that progress has been stopped. Patents accumulate unused, unable to attract new investors.

 

[bit_user]

In the past, there were so many foundries, that when one got stuck, being unable to advance, some other foundry developed a new idea, and beat them, accelerating the advance of the technology.

You don't just need multiple competitive foundries, but also competitive equipment suppliers. This article is really about ASML being in the driver's seat about whether to go ahead with hyper-NA. We shouldn't consider it a good thing that a single company is relied upon to make such decisions for an entire industry.

Today the number of foundries is so small and expensive, that new ideas that think out the box got so rare that progress has been stopped. Patents accumulate unused, unable to attract new investors.

I'm not too concerned about willingness to try new ideas. The fabs roll out different developments in slightly different orders and timelines, also with their own twists on the approach. For instance, TSMC was first to embrace EUV, Samsung was first to utilize GAA, and Intel was first to deploy backside PDN. I think, if there were good ideas that are feasible, they wouldn't be ignored. This business is so incredibly high-stakes that you really can't afford to leave any stone unturned.

Probably the main reason fabs are following similar roadmaps is that IMEC does so much work in analyzing different ideas and assessing their merits and the roughly optimal order to deploy them. Of course, having a common equipment supplier is surely another big factor.

 

[yeyibi]

I'm not too concerned about willingness to try new ideas. The fabs roll out different developments in slightly different orders and timelines, also with their own twists on the approach. For instance, TSMC was first to embrace EUV, Samsung was first to utilize GAA, and Intel was first to deploy backside PDN. I think, if there were good ideas that are feasible, they wouldn't be ignored. This business is so incredibly high-stakes that you really can't afford to leave any stone unturned.

The story of computer hardware is full of cases where the engineers didn't got support for their ideas, so they left the company, started their own, and made a revolution.
For example Cray left his company to build his own supercomputers, and his ideas are today in all consumer processors.

Many of them failed, but the ones who succeeded changed the industry.

That's no possible today, because to be relevant is necessary to have billions in new investment.

 

[bit_user]

The story of computer hardware is full of cases where the engineers didn't got support for their ideas, so they left the company, started their own, and made a revolution.

I think this doesn't really apply so much to a field as well-studied as lithography. Yes, there are some approaches that are fundamentally different, like Canon's, but it takes many $Billions of investment, to get an idea like that off the ground. By the time they have it working, lithography will already have surpassed it, making it unclear where the additional investment would come from to try and keep it going.

If we restrict ourselves to the set of incremental changes that fabs can implement, I believe there aren't really untapped opportunities. Too many really smart people are working on this field, in at least 3 different companies.

For example Cray left his company to build his own supercomputers, and his ideas are today in all consumer processors.

That was more than 50 years ago, when the industry was in its infancy. He left not because they rejected his ideas (some of which indeed made it into several CDC machines), but because he wanted to chase a different market than his employer. So, it was a business-level disagreement, rather than technical.