News Intel CEO Gelsinger says China is ten years behind in chipmaking capabilities, and it will stay that way

[Admin]

Intel CEO Pat Gelsinger says China's semiconductor industry is a decade behind, and it is going to stay this way.

Intel CEO Gelsinger says China is ten years behind in chipmaking capabilities, and it will stay that way : Read more

 

[peachpuff]

Without access to them, Chinese companies will have to develop their wafer fab equipment and means to purify gases, resists, and other chemicals for leading-edge chip production.

Develop with stolen ip? 🤔

 

[ivan_vy]

Pat talks to investors, TSMC N7 was released in 2018, Samsung N7 in 2019 and Intel N7 in 2021, the gap is 5 years not 10, it's a hell of a long time in tech development but N7 is still viable for consumption of mass products for office, schools and embedded systems.
The bleeding edge might not be in risk to be overtaken by China but the mass market is.
https://en.wikichip.org/wiki/7_nm_lithography_process

 

[nightbird321]

Anyone else here remember Intel's confidence about their own fab advantage 10 years ago?

 

[endocine]

"China will continue to develop its industry" s/develop/industrialespionage/

 

[kwohlt]

Pat talks to investors, TSMC N7 was released in 2018, Samsung N7 in 2019 and Intel N7 in 2021, the gap is 5 years not 10, it's a hell of a long time in tech development but N7 is still viable for consumption of mass products for office, schools and embedded systems.
The bleeding edge might not be in risk to be overtaken by China but the mass market is.
https://en.wikichip.org/wiki/7_nm_lithography_process

Because it's not a linear trajectory. Surpassing 7nm-class lithography without access to more ASML EUV is difficult. Eventually the industry will transition to high-NA EUV over the next 5 years and China will need to develop this tooling from scratch.

Massive amounts of inputs have been sanctioned and will now need to be created domestically to offset this - this is a huge, but certainly possible, burden to overcome that'll absolutely set the Chinese semi-conductor industry back by years relative to western rivals.

 

[Maolo76]

10yrs behind.. Maybe. But the smaller node has limited use at the moment. The majority is 7nm which cvina can produce.
I predict China is embarking on a dual path on catching up and researching and developing different path to manufacturing chips. They are working on photonic, carbon type chips, etc. So just like china dominate the EV sphere. They maybe on the same journey to leapfrog the legacy companies. Just like an article said.. They are developing disruptive technology that will shake up the traditional semiconductor industry.

 

[artk2219]

Its closer than the soviet union ever got, they were a good 15 - 20 years behind on average, then again global tensions with China are not as bad as they were with the Soviet Union. That said, it would definitely put them in the good enough for most uses spot, which means they will hopefully be more innovative in how the equipment is used, it could be interesting honestly.

 

[kwohlt]

10yrs behind.. Maybe. But the smaller node has limited use at the moment. The majority is 7nm which cvina can produce.
I predict China is embarking on a dual path on catching up and researching and developing different path to manufacturing chips. They are working on photonic, carbon type chips, etc. So just like china dominate the EV sphere. They maybe on the same journey to leapfrog the legacy companies. Just like an article said.. They are developing disruptive technology that will shake up the traditional semiconductor industry.

Are there many examples of China developing breakthrough technologies? Generally they excel at taking technologies pioneered in other countries and developing methods to reproduce them at scale for lower cost.

Semi-conductor manufacturing is different then EVs or pretty much any other manufacturing - leading node manufacturing at scale is the most difficult production on the planet. It's not easy.

Sure, Chinese companies may develop ways to produce sub 7nm class nodes with current equipment, but each successive node generation that comes, without improved tooling, is going to suffer from yield issues as more layers and patterning is required.

 

[kwohlt]

Pat talks to investors, TSMC N7 was released in 2018, Samsung N7 in 2019 and Intel N7 in 2021, the gap is 5 years not 10, it's a hell of a long time in tech development but N7 is still viable for consumption of mass products for office, schools and embedded systems.
The bleeding edge might not be in risk to be overtaken by China but the mass market is.
https://en.wikichip.org/wiki/7_nm_lithography_process

Missing the forest for the trees. Current Chinese production capabilities are due to pre-sanction tooling and inputs. Further node development is essentially "paused" until these capabilities can be domestically reproduced. TSMC N3 launched in 2023. Intel is launching GAA-FET and BSPD later this year. Densities are surpassing 200MTr.

Comparing 7nm class nodes is missing the point - the question is when will post-sanction China be able to reproduce the nodes that are coming out post-sanction from Intel, TSMC, and Samsung?

As it stands, China can definitely pose a risk to other industries like Global Foundries, but that's not really the concern of the sanctions - the Sanctions are specifically to limit leading node capabilities and their relationship to AI hardware.

 

[jp7189]

Missing the forest for the trees. Current Chinese production capabilities are due to pre-sanction tooling and inputs. Further node development is essentially "paused" until these capabilities can be domestically reproduced. TSMC N3 launched in 2023. Intel is launching GAA-FET and BSPD later this year. Densities are surpassing 200MTr.

Comparing 7nm class nodes is missing the point - the question is when will post-sanction China be able to reproduce the nodes that are coming out post-sanction from Intel, TSMC, and Samsung?

As it stands, China can definitely pose a risk to other industries like Global Foundries, but that's not really the concern of the sanctions - the Sanctions are specifically to limit leading node capabilities and their relationship to AI hardware.

Furthering your thoughts, can China even maintain the equipment they have now? Presumably sanctions cause them to lose access to replacement parts for many of the machines.

 

[Avro Arrow]

I feel so much better knowing that the master of putting his foot in his mouth says that China will remain ten years behind. 🙄

I'm not sure how anyone can take this joker seriously at this point.

 

[Order 66]

I feel so much better knowing that the master of putting his foot in his mouth says that China will remain ten years behind. 🙄

I'm not sure how anyone can take this joker seriously at this point.

I would like to believe that China was 10 years behind.

 

[Blastomonas]

I suspect that he might be eating his words in the near future.

 

[Avro Arrow]

I would like to believe that China was 10 years behind.

You're right. I'm willing to bet that China's a lot closer than we think they are. It's like I often say with regard to China:

"Underestimate them at your own risk!"

 

[phead128]

Mr. Pat "AMD is in the rearview mirror" Geisinger. The Jim Cramer of predictions.

 

[Neilbob]

It almost seems as if there are almost as many 'Gelsinger says' articles as there AI ones lately.

Whether there is any accuracy there or not, I couldn't possibly say.

 

[tanon]

This is the same China that graduates 1.4 million engineers every year - this compared to the United States' 200,000 or so.

Oh and let's not forget the fact that many of those engineers will actually find their way into industrial/engineering jobs after finishing university, due to China's enormous manufacturing base - as opposed to, say, joining an investment bank or tech publication, as in the case of those engineers graduating in Western service-based economies.

Even taking Gelsigner's 10 year figure at face value, I wouldn't be surprised if China was able to achieve parity with the West by 2030, at least.

 

[AndrewJacksonZA]

OK, so they're actually 2 - 4 years behind. Got ya.

 

[kwohlt]

This is the same China that graduates 1.4 million engineers every year - this compared to the United States' 200,000 or so.

Oh and let's not forget the fact that many of those engineers will actually find their way into industrial/engineering jobs after finishing university, due to China's enormous manufacturing base - as opposed to, say, joining an investment bank or tech publication, as in the case of those engineers graduating in Western service-based economies.

Even taking Gelsigner's 10 year figure at face value, I wouldn't be surprised if China was able to achieve parity with the West by 2030, at least.

Parity? How so? While the west moves on to High-NA EUV. Has China even begun the process of reproducing all of the pre-requisite upstream industries that are being sanctioned? No company on earth has been able to match ASML's equipment, but you think not only will China be able to catch up to current ASML machines, but to match TSMC's adoption of High-NA EUV? Do you have any idea how monumental of an undertaking this is?

You and most commenters here are missing the key point. While it was certainly a (slim) possibility that pre-sanction China could catch up to Intel and TSMC by 2030, they now have to also build out an entire domestic upstream industry before they can even begin to attempt this.

Are you really suggesting that China will have mass produced GAAFET, BSPD 200MTR+ chips within the next few years??

 

[kwohlt]

OK, so they're actually 2 - 4 years behind. Got ya.

I'm going to use a made up car analogy here because many people here don't seem to understand what's going on.

Imagine Mexico starts manufacturing cars. They're not quite as good as Hondas, Fords, or BMWs, but they're rapidly improving and will likely be on par by 2030.

However, this Mexican company buys their leather from Canada, their tires from Italy, their transmissions from Germany, their engines from the US, the factory machinery from South Korea, and their in-dash electronics from Japan, and then assembles that into a car in their factory.

Now imagine all of these countries cut off this trade. Now, before they can advance and improve their cars, they have to build and design their own domestic engine factory, they need to design and build new transmissions, they need to assemble their own domestic tire production, their own domestic electronics, their own domestic factory machinery, and build relationships and shift to domestic tanneries. Then they could start back where they left off and try competing.

Your "2 - 4 years behind" statement might've made sense in an alternate reality where there were no sanctions, but you forgot to include these sanctions in your formula.

 

[artk2219]

Not if China gets Taiwan before that.

You better believe there are some pretty horrible scorched earth contingency plans if it comes to that, lets hope that it never does.

 

[hotaru251]

Gelsinger & badmouthing competitors....name a more famous duo.

 

[SSGBryan]

Not if China gets Taiwan before that.

Get back with us when China has the capability to get more than a brigade of troops across the straits.

 

[gg83]

Pat talks to investors, TSMC N7 was released in 2018, Samsung N7 in 2019 and Intel N7 in 2021, the gap is 5 years not 10, it's a hell of a long time in tech development but N7 is still viable for consumption of mass products for office, schools and embedded systems.
The bleeding edge might not be in risk to be overtaken by China but the mass market is.
https://en.wikichip.org/wiki/7_nm_lithography_process

Excellent point. I believe China is really struggling with yields on their "7nm class" node. I think They will have to subsidize the extra cost to take over mass market. And that's not sustainable either.