China's 7nm Chips: SMIC N+2

there has been a lot of talk recently about smic I was hoping to spend my weekend working on a video about the bubble tea industry but I have gotten far more emails asking me to do this one Tech insights and semi-analysis recently broke news that the Chinese semiconductor National Champion has started producing chips made with their own seven nanometer process node own in quotations in this brief video I want to say a few thoughts about smic's n plus 2 node the seven nanometer Leap Forward now before we start I want to say that this is a current event more news will break and may invalidate or change what I have said in this video I do want to make clear that the only information I have here is what everyone else has and I'm not interested in American failure doomsayers China maximalists anti-china name calling or any general nastiness please be civil in the comments if you want to go back and dig up salt take a deep breath and go outside for a picnic or something life is too short and yes there are probably many analysts out there who know the technical side of this much better than I do I'm just a guy with the YouTube channel now that we have acknowledged all of this let us break it down together first we need to go through this whole song and dance about the meaning of seven nanometers here it is again process node numbers are marketing devices that try to convey a better chip in either power performance or area literally speaking the seven nanometers phrase means nothing it does not correspond to any physical Dimension it simply says that the chips are 70 better than the previous generation 10 nanometer chips which in turn are 70 better than the previous generation 14 nanometer chips and so on to avoid controversy tsmc calls their seven nanometer node and seven this smic node dubbed n plus 2 is a descendant of the N plus 1 process which was first taped out in October 2020 both nodes use duv immersion lithography with multi-patterning which is what many foundries had to resort to before euv finally came out smics n plus 2 is capable of using euv technology as well and smic tried to purchase such a device from asml to use it but the United States has placed an export ban on this technology blocking this part of the road map semi-analysis via Tech insights notes that n plus 2 is very similar to tsmc's N7 like as in 99 the same and generally what that means is that the physical measurements of its transistors fin shaped thingies called finfets and their placements are similar tsmc first shipped N7 chips in high volume in 2018 so an N7 equivalent node in high volume would put smic about four to six years behind the Leading Edge no question n plus two is one of the most advanced processes made available by any Foundry to its customers but is it economically competitive without euv note the wording here economically competitive we can still use 193 nanometer immersion technology and multi-patterning to Fab chips at even more advanced nodes than N7 I did a video about this a long time ago in January 2022 legendary tsmc R director Dr Bern Lin the father of 193 nanometer immersion lithography mentioned in an interview that smic confab even N5 equivalent chips with only multi-patterning no UV necessary but making a few Wafers isn't enough the issue is whether their node can yield in meaningful quantities Samsung has come out with a lot of really advanced process nodes but the chips have not hit their yields yield meaning that they can be used and also performed to spec and then there's Intel which famously struggled to go from their 10 nanometer equivalent node to a 7 nanometer equivalent node without UV but tsmc ramped up N7 without much reported trouble and N plus 2 is a node very similar to the tsmc N7 node so I would bet it follows the same yield optimization curve as N7 now I do note that the market has changed since 2018. competition is rougher a Foundry with euv is going to have economic advantages over those only with duv and N7 equivalent node with only multi-pattening has manufacturing limitations that place serious design restrictions on the chip designer it's a product but not an ideal one is n plus two economically viable in the broader marketplace right now in an open market I would say no there is a better performing N7 equivalent out there with tsmc's n6 an evolution of their N7 and one equipped with euv but then again maybe smic is already expecting n plus 2 to be a china-only product n plus 2 is such a close ringer for tsmc's N7 that it undoubtedly intrudes on some trade secrets in 2002 and 2006 tsmc successfully went to court in the United States to get product import bans on smic products made with tsmc Trade Secrets they can try it again if successful this would remove smic's n plus 2 viability as an export product leaving it for use only in domestic Chinese products and those for export to friendly countries smic does care about their export capability to countries like the United States they highlight it in their latest earnings call but I do suspect they no longer depend on it the Chinese chip design industry has advanced a great deal since the first tsmc smic lawsuit back in the mid-2000s it is much more sophisticated and the Chinese electronics Industry has greatly matured I imagine that high silicon huawei's chip design arm and once China's biggest Semiconductor Company would love to use n plus 2 to make high performing Chinese mobile processors for their very popular phones and server chips for their very popular cloud services it would not surprise me to see a very sophisticated High silicon n plus 2 chip in the near future they should be familiar with it high silicon was one of tsmc's Flagship customers alongside Apple fabing the kitty 980 985 and 990 series chips on N7 if some random Bitcoin miner can use n plus two then so can high silicon smic would be kicking the Hornet's Nest here but in some ways it is kind of worth it because High silicon and Huawei can bring serious volume foundries rely on massive production volumes to scale and yield optimize their process nodes High silicon and smic together will be a Duo to be reckoned with so if they can get those volumes then we should expect n plus 2 to get better over time back in 2018 gsmc shipped millions of N7 Wafers for companies like apple before euv arrived and they got really good at it too we should expect smic to do the same China has made Leading Edge semiconductor manufacturing a high policy priority for a very long time basically since the 2000s so what is different now I would point to smic co-ceo liangbong song A tsmc alumni Liang is known for his technical Brilliance difficult personality and his dogged Chase for the Leading Edge I profiled him in a video that may or may not be released by now I highly recommend watching it if you can when he joined smic in 2017 from Samsung he brought with him his own 200 person R D team of Taiwanese and Koreans and immediately reoriented the company from being a boring Second Source manufacturer to a Leading Edge monster so long as Liang remains smic's co-ceo not exactly a given considering his past Shenanigans the company should continue pushing towards the Leading Edge we should all know and respect his name if I were Intel I'd give this guy and his team a Steph Curry sized contract to come to the United States and make magic happen and what about the Gulf States the Saudis are spending hundreds of Millions on a golf tournament Abu Dhabi put 10 billion dollars into Global foundries what's a few billion more having access to the latest lithography Technologies is crucial to smic's Long March to the Leading Edge recently news has come out about the United States government pushing the Netherlands asml and Japan's Nikon to prevent export of their immersion duv lithography machines to China the timing of this push and the N plus 2 shipment is probably not a coincidence the United States has long blocked the export and sale of the newest most advanced euv machines to China but to expand the block to immersion lithography is interesting and represents another step forward immersion machines use water and other fluids to improve the resolution of an older 193 nanometer ARF Laser Technology referred to as dry they are the workhorses of modern semiconductor manufacturing and quite mature prototypes first entered the market in 2003. today the only two suppliers in the market are asml and Nikon dry immersion 193 nanometer ARF machines can be modified into immersion lithography machines and it can be done relatively quickly as in like less than five years though not without substantial technical effort and expertise the semiconductor industry in the west transitioned from dry to first generation immersion lithography in about two to three years an N7 equivalent node will require more sophisticated immersion Technologies than that so that adds to the timeline as of this writing Japan and the Netherlands have not decided on any ban the Chinese government has nevertheless responded with exclamations of unfairness Foul Play and so on it's a very emotional topic it doesn't feel Fair what about free markets I get it but every government also protects their nationally important industries Europe protects their automobile Industries China protects their Telecom and internet Industries and America has been protecting their Computing technology Industries since the days of the Cold War export control agreements to prevent American Technologies from reaching unfriendly countries have been widely debated for decades now think tanks and internet forums all around the world continue to debate whether or not they work my best guess is that an export ban of 193 nanometer immersion lithography technology without a corresponding ban on dry ARF technology which to me does not seem feasible adds from four to six years to smic's March towards the Leading Edge a ban is also likely to stimulate substantial investment in chinese-made lithography probably first an equipment that can maintain existing immersion equipment and second anything to facilitate the dry to immersion transition if mastered then the expertise can be leveraged into a more comprehensive system can such companies ever ship a complete system on par with asml or even Nikon it is debatable personally I feel that the Gap there is greater several years perhaps but insurmountable nothing is a legitimate seven nanometer node at high volume and good yield is a Triumph for China's Advanced manufacturing industry only three other companies have ever progressed to this level of scale and sophistication China and the Chinese should be proud of this accomplishment whatever the asterisks so what is America to do any successful immersion ban if it comes would not be a permanently damaging blow it can only maintain and perhaps slightly expand the sino-american semiconductor Gap the administration of course is probably aware of this and thus values those few years very highly as they should I want to hearken back to the video I made about the American response to the Japanese semiconductor challenge over 30 years ago yes there were politics and yes the American government got involved but all that can do was buy a little time had the American Semiconductor companies did nothing else but whine to the government to come save them then they would have rightfully drowned with that preciously bought time the industry built out new markets improved their manufacturing weak spots adopted Innovative new suppliers and more nearly 40 years later I reckon they need to gather themselves up and do it again to quote my favorite Netflix show Castlevania don't be sorry just be better all right everyone that's it for tonight thanks for watching subscribe to the Channel Sign up for the newsletter and I'll see you guys next time maybe with a boba video

Channel: Asianometry

Views: 477,437

Keywords:

Id: dQGnwKBxAKk

Length: 13min 39sec (819 seconds)

Published: Sun Jul 31 2022