Chinese semiconductor industry

measuredingabens said:

Wasn't the estimate for the construction time of an SSMB facility around a year? Has that changed? Depending on where they're at in the construction then there could be an operational facility potentially in 2024.

Click to expand...

How long it takes to build a synchrotron isn't the point, it's what the facility at Xiong'an is intended to do. If it's still an early proof of principle device - which it very well could be - then SSMB EUV is still some ways away while the experiments get done. If China is willing to take more risk by pursuing a more aggressive roadmap, then the facility could serve as a direct prototype for a production synchrotron, with commercial entities like SMIC experimenting with it and beginning low rate production fairly soon (around 2025).

In any case, an LPP production prototype has been/will soon be delivered and the most credible rumours indicate that it should be in commercial production within 3 years.

ZeEa5KPul said:

To say the media is stupid is to say water is wet. It's best just to ignore them.

When SSMB will be ready for at least the commercial equivalent of LRIP depends on the nature of the facility being built at XNA and its place on the technological roadmap. If it's a small-scale proof of principle demonstrator then I would say that SSMB EUV would be post 2027 at the earliest. If China's more ambitious and builds an advanced prototype of the intended industrial synchrotron, then we'll see SSMB EUV much earlier, possibly even a Ma Weiming situation between LPP (steam) and SSMB (electromagnetic).

Click to expand...

---

- $176 million USD per facility ()
versus
scanner - $150 to $200 million USD per unit ()

---

- $720 million USD per facility ()
versus
scanner - $300-400 million USD per unit ()

---

An entire synchrotron facility costs as much as a single EUV scanner (or two). At 7nm, normally 15 DUV systems and 5 EUV systems are demanded, depending on chip type and company, according to @tinrobert article () For reference, there has been 180 EUV scanners sold by end of 2022, with another 60 EUV to ship this year. Assuming there all vanilla EUV, that's over $32 billion USD by end of 2022 alone.

This alone makes SSMB-EUV highly attractive and cost-effective option, as a single facility can produce EUV beams for multiple lines of wafer fabrication, improving throughput and efficiency in semiconductor manufacturing process. The Tsinghua SSMB-EUV in Xiongnan should strive to be more ambitious given an entire synchrotron facility cost only a single EUV scanner or two - and SMIC needs few dozens EUV, possibly more to satiate huge domestic demand. This can replaced by an SSMB-EUV lithography "factory" or few of these.

weig2000 said:

ChinaTalk on Huawei's new chip and Mate 60 pro, etc.

From the introduction of the video:

Huawei’s breakthrough Kirin 9000s: what is it, why is it a big deal, and what if anything should the US do about it? Joining me, I have on two fantastic semiconductor analysis, Doug O'Laughlin of Fabricated Knowledge and Dylan Patel of Semi-Analysis. We get into: - How this chip illustrates Chinese engineering excellence and the porous nature of the current export control regime - Why we can expect AI chips on par with the A100 coming out of China in the next two years - What steps the US government could take to tighten export controls and set back the Chinese semiconductor ecosystem - How China has come to dominate both the lagging edge and the EV space.

---------------------------------------------------------------------------------------------------------------

Overall impression:

• Dylan Patel & Co are highly complimentary about Huawei's Kirin 9000S and SMIC's process technology.
• "China has been constantly underestimated;" "China's process technology is on par with the US now (Intel)."
• Barely mention Chinese SEMs, appear to be quite ignorant of the players and rapid progress that have been made
• "EUV is the last battle."
• Talk a lot about the EV/BYD analog
• Talk about Patel's List of 20 to stop China's semiconductor development and advance once for all.

Click to expand...

Jiejia Weichuang's five-in-one clustered perovskite laminated vacuum coating equipment was successfully shipped​

Recently, JJVicron's five-in-one clustered perovskite laminated vacuum coating equipment successfully rolled off the production line and was successfully shipped after passing the customer's FAT acceptance. The five-in-one clustered perovskite stack vacuum coating equipment will have a significant impact on the field of thin film preparation. This revolutionary technology integrates multiple technologies such as RPD, multi-source evaporation, RF sputtering, pulsed DC sputtering and DC sputtering, and can continuously deposit various thin films in a vacuum environment, including TCO (ITO, IWO, ICO, IZO), Cu, SnO2 and NiO, etc., while maintaining the best characteristics of the interface, providing a powerful equipment support for research in the fields of materials science, electronic engineering and optoelectronics.

Many technologies in one

This equipment combines multiple technologies such as RPD, multi-source evaporation, RF sputtering, pulsed DC sputtering and DC sputtering to achieve continuous deposition of thin films in a vacuum environment. This innovative technology can be applied in a variety of fields, providing new possibilities for research in fields such as materials science, electronic engineering, and optoelectronics.

Phead128 said:

---

Please, Log in or Register to view URLs content!

- $176 million USD per facility (

Please, Log in or Register to view URLs content!

)
versus

Please, Log in or Register to view URLs content!

scanner - $150 to $200 million USD per unit (

Please, Log in or Register to view URLs content!

)

---

Please, Log in or Register to view URLs content!

- $720 million USD per facility (

Please, Log in or Register to view URLs content!

)
versus

Please, Log in or Register to view URLs content!

scanner - $300-400 million USD per unit (

Please, Log in or Register to view URLs content!

)

---

For reference, there has been 180 EUV scanners sold by end of 2022, with another 60 EUV to ship this year. Assuming there all vanilla EUV, that's over $32 billion USD by end of 2022 alone.

This alone makes SSMB-EUV highly attractive and cost-effective option, as a single facility can produce EUV beams for multiple lines of wafer fabrication, improving throughput and efficiency in semiconductor manufacturing process. The Tsinghua SSMB-EUV in Xiongnan should be more ambitious given a single EUV scanner cost as much as an entire synchrotron facility - and SMIC needs few dozens, possibly more to satiate domestic demand.

Click to expand...

How much the cost of ASML EUVL is related to CYMER power source, 40% or $80 million? So if we do the math, a SSMB between the size of Shanghai and Beijing Synchrotron costing at $500 million powering at least at minimum 10 EUVL units at $120 million each for a total of $1.7 billion, compare that to ASML HI NA EUVL (since both are the same category tech wise) at $3billion ($300 million x 10). I think everybody with a functioning brain can do a business calculation which is better.

tphuang said:

one thing to note is that despite all of huawei's new devices, China as a whole is still very dependent for American chips in consumer electronics here

Please, Log in or Register to view URLs content!

this example of leishen X4 (thunderbolt) router uses broadcom chip & 8 FEM which are likely the broadcom FEMs

Production domestically is still limited and Huawei likely uses up all of it

Click to expand...

What can you really do about that?

These chips being made by TI, Renesas, Broadcom like DAC, Wi-Fi, DSP, worth like $2 - $10 USD.

Would the US really restrict these kinds of commodity chips? They aren't hard to design, just not worth the effort. If you make it worth the effort, these western companies are simply going to shut down.

Burn Lin, a former vice-president of R&D at TSMC, said the tech giant has had a factory in Washington state for more than a decade and enlists many local workers.

“We never can make it achieve the same performance as [our] own Taiwan factory,” he noted. “We can never bring it up to the same level, no matter how hard we try.”

Taiwanese workers, like their American counterparts, have their own frustrations.

Kao said many Taiwanese voters were “nervous” about US reshoring rhetoric and were questioning whether bringing Taiwanese engineers would serve Taiwan’s interests, not just America’s.

“Basically, are we being bullied to do something we don’t want to do? What’s in the best interest of the United States, we understand. But is it in our best interest?”

supersnoop said:

What can you really do about that?

These chips being made by TI, Renesas, Broadcom like DAC, Wi-Fi, DSP, worth like $2 - $10 USD.

Would the US really restrict these kinds of commodity chips? They aren't hard to design, just not worth the effort. If you make it worth the effort, these western companies are simply going to shut down.

Click to expand...

What do you mean what can you do about it?

Ramp up your production and make it to all the OEM supply china? Right now, we have a capacity issue that can be addressed over next few years.

When has chinese industries ever stopped once they start to win?

tokenanalyst said:

They underestimated and they keep underestimating a lot of development that are happening in the Chinese industry, there is a real probability that technological wise Chinese semiconductor materials and equipment players are getting closer and closer to their foreign counterparts because this guys talk a lot about the things they read in SCMP or Nikkei, maybe except Dylan, but I have never see a technical decent detail of the things the China have and doesn't have. The exception being a really good presentation about the Chinese semiconductor supply chain that I read in 2019, sadly i lost the link, that presentation end with a warning against export controls.​

Click to expand...

I think you're talking about this . Or another article from the same site.

As for Dylan, he is a consultant with fairly decent connections. I think it's highly unlikely that he puts out his best stuff in public. I actually do pay for his reports (I subscribed back when he was only 30$ per month), and the paid sections are considerably better than the public sections. I assume that there is a similar step-up in detail and information for the actual reports he produces for important clients.

I wouldn't jump to conclusions about whether he knows or doesn't know something.