China says 'No' to GPU price-hikes by preparing a big chipmaking leap with particle-accelerator driven SSMB-EUV

[Delicieuxz]

An EUV lithography machine? How about an EUV lithography factory?

China is aiming to leap-frog current top-end EUV technologies with steady-state microbunching (SSMB) EUV, that would give China what's been called a "lithographic cannon", enabling faster, higher-yield, and cheaper production of cutting-edge 2nm-and-lower microchips.

China sets out to develop EUV lithographic ‘cannon’

China Aims to Use Particle Accelerators to Build Chips and Evade EUV Sanctions

According to the South China Morning Post (SCMP), China plans to use an innovative approach to manufacturing processors by harnessing particle accelerators, potentially positioning itself as a global leader in advanced chip manufacturing. SCMP says the method seeks to evade traditional lithography machine limitations and US sanctions on EUV technology, potentially reshaping the semiconductor industry landscape.

The Chinese research team, led by Tsinghua University, is developing a unique laser source using particle accelerators. Their goal is to sidestep the constraints of conventional lithography machines, which are pivotal in microchip production. The proposed particle accelerator will be roughly the size of two basketball courts, between 100-150 meters in circumference, and will serve as a high-quality light source for chip fabrication.

Professor Zhao Wu from Stanford University introduced the underlying technology, termed steady-state microbunching (SSMB). SSMB captures the energy emitted by charged particles during acceleration, transforming it into a continuous, pure EUV light source. Compared to the prevalent ASML EUV method, SSMB boasts superior power and efficiency, potentially reducing chip production costs.

China to Challenge ASML with a better technology than EUV

ASML prioritizes downsizing chip manufacturing machines for international trade. In contrast, the Chinese initiative centers on domestic production. It aims to construct an extensive facility consolidating numerous lithography machines around a central accelerator. This advancement can fuel abundant and cost-effective chip production. It positions China as a leader in industrializing advanced chips, including 2nm and beyond.
...
While numerous researchers are pursuing this technology, Chinese scientists have taken a distinct approach. This project has been in progress since 2017, but it gained significant visibility recently due to Huawei’s notable advancements in chip manufacturing.
...
Compared to the existing ASML EUV technology, [China's SSMB tech] proves to be a superior light source. It offers higher average power and increased chip production while maintaining lower unit costs.

ASML utilizes laser-induced plasma to generate an EUV source, involving the projection of strong laser pulses onto tin microdroplets. This impact crushes the droplets, producing EUV pulse light. Following intricate filtering and focusing processes, an EUV light source with an approximate power of 250W is generated.

However, the EUV beam, before reaching the chip, goes through reflection from 11 mirrors, resulting in around a 30 percent energy loss at each reflection. Consequently, when the beam reaches the wafer, its power is reduced to less than 5W, presenting challenges especially as manufacturing scales down to 3nm or 2nm.

In contrast, SSMB technology alleviates these concerns. SSMB beams attain a notably higher output power of 1000W, and due to their narrow bandwidth, they require fewer reflecting mirrors, leading to an inherent increase in terminal power.

 

[michalrz]

Holy shit...

I mean, "stretching ring" jokes aside, that sounds pretty damn complicated. I'll try to read up on it, interesting even though I barely understand what's up

 

[Zarathustra[H]]

This is difficult stuff, and China isn't exactly known for in-house innovation.

Let's see if they can do it.

And I bet this has absolutely nothing to do with "saying no to price hikes".

They are just tired of being on the wrong side of export restrictions for their defense industry.

 

[Delicieuxz]

This is difficult stuff, and China isn't exactly known for in-house innovation.

Let's see if they can do it.

And I bet this has absolutely nothing to do with "saying no to price hikes".

They are just tired of being on the wrong side of export restrictions for their defense industry.

Yeah, the lower GPU prices thing was just a joke, given how fed up PC gamers and enthusiasts are with the price increases.

But last December, ASML's CEO basically accused the US of using sanctions on China to steal European business, saying the excuse of restricting China's defence industry doesn't make sense because chips for strictly military application are typically 10 - 15 years old, and there are no restrictions on those exports.

If the CEO of ASML is correct, then chip export restrictions to China could just be a dirty, anti-competitive move. Huawei seems to have deliberately unveiled its new phone featuring a homebuilt, SMIC-produced 7nm 5G chip while Gina Raimondo was in China for a diplomatic visit.

 

[cjcox]

If China can truly do this, maybe the USA will start buying everything from them.

 

[Sycraft]

This is difficult stuff, and China isn't exactly known for in-house innovation.

Let's see if they can do it.

And I bet this has absolutely nothing to do with "saying no to price hikes".

They are just tired of being on the wrong side of export restrictions for their defense industry.

Ya they don't have a lot of choice, it isn't a money thing but a "We can't get the good tech" thing. So they are going to try and make their own.

...however as you say, in house innovation is something China does not tend to do well. It's not just a lack of experience thing either, part of the issue is there is a massive culture of BSing/fraud and that applies to academia as well. So people will just straight out make shit up, and they don't get called out on it. That doesn't work when it comes to science because, as Feynman put it "For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled." You can write a paper claiming you've discovered something amazing. You can have a lab that gins up fake results saying it works... but when you actually try to build something if what you said isn't true it won't work.

We'll see, I certainly won't write China off, they are a big country with lots of resource and lots of smart people, but cultural issues are hard to overcome. When you are doing cutting edge R&D you can't cut corners, you can't BS, and you can't indulge in people's egos. You have to focus on what is real, what works.

 

[Zarathustra[H]]

Yeah, the lower GPU prices thing was just a joke, given how fed up PC gamers and enthusiasts are with the price increases.

But last December, ASML's CEO basically accused the US of using sanctions on China to steal European business, saying the excuse of restricting China's defence industry doesn't make sense because chips for strictly military application are typically 10 - 15 years old, and there are no restrictions on those exports.


View attachment 601440


If the CEO of ASML is correct, then chip export restrictions to China could just be a dirty, anti-competitive move. Huawei seems to have deliberately unveiled its new phone featuring a homebuilt, SMIC-produced 7nm 5G chip while Gina Raimondo was in China for a diplomatic visit.

View attachment 601441

I think it depends on what you are making.

A lot of stuff (guidance systems on missiles, etc) certainly can use older chips, but some of the more cutting edge tech, like some of the key components you find inside of an F-35 that allows for its best in the world sensor fusion and AESA radars I'm not too sure about.

And remember, it's more about preventing them from outpacing us in the next gen, not the current. Next gen which is under development right now includes things like the Air Force's NGAD program and the Navy's F/A-XX both of which are said to rely heavily on AI integration, as well as the use of "cheap" and expendable autonomous loyal wingmen. This isn't stuff that can be done with 15 year old tech.

 

[Sycraft]

And remember, it's more about preventing them from outpacing us in the next gen, not the current. Next gen which is under development right now includes things like the Air Force's NGAD program and the Navy's F/A-XX both of which are said to rely heavily on AI integration, as well as the use of "cheap" and expendable autonomous loyal wingmen. This isn't stuff that can be done with 15 year old tech.

Also with military tech: Most of China's stuff is based on Russian designs. Some of it pretty straight copies or Russian supplied, some modified. The world is currently seeing how that functions against western weapons in Ukraine and the answer is "not so well." This is also not necessarily the latest and greatest shit from the west, just what they are willing to give Ukraine. China may well be worrying that they are already more behind than they thought they were based on the real world results being seen, and as you say it isn't stopping there, there are already new weapons systems in development.

 

[DouglasteR]

I can only hope for it to be true or feasible.