You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
The logo of Rapidus Corp. is displayed at the company headquarters in Tokyo, Japan February 2, 2023. REUTERS/Issei Kato/File
TOKYO, April 2 (Reuters) - Japan's industry ministry said on Tuesday it has approved subsidies worth up to 590 billion yen ($3.9 billion) for chip foundry venture Rapidus as Tokyo pushes forward with plans to rebuild the country's chip manufacturing base.
Japan previously agreed to provide subsidies worth some 330 billion yen to Rapidus. The newly approved aid includes 53.5 billion yen for advanced packaging, which is becoming increasingly important for driving chip performance improvement.
Rapidus is headed by industry veterans and is targeting mass production of cutting-edge chips on the northern island of Hokkaido from 2027 in partnership with IBM and Belgium-based research organization Imec.
Countries around the world are looking to strengthen their control over chip supply chains after global shocks including the COVID-19 pandemic and trade tensions between the United States and China.
Taiwan Semiconductor Manufacturing Co (TSMC) (2330.TW), opens new tab, the world's largest contract chipmaker, has spent decades honing its processes, and many in the industry are therefore skeptical about the prospects for success by Rapidus.
They lost me at "in partnership with IBM". I'm glad to see all money invested in the semiconductor ecosystem but I do not believe this joint venture will succeed. Good thing Japan is also working with TSMC. I highly doubt TSMC is losing sleep over this one.
To once again reprise my role as the devil's advocate, what is the alternative if you want a national champion for advanced CMOS logic? SMIC is lagging even intel 10nm by 4+ years to market even with rampant TSMC/Samsung poaching and an effectively infinite budget. I can't think of any other way to jump start other than licensing IBM tech. Now if we want to have the discussion of this is a pointless endeavor and Japan shouldn't even try, I think that is a different can of worms. Time will tell I suppose. I might just be over optimistic/trying to hope something into existence, but I think it is a project that is worth perusing. But it is not for the faint of heart. All one needs to do is see the strain Samsung and intel are under to know that Japan/JV partners need to slam 100s of billions into Rapidus over the next 20 odd years before they can even hope for the possibility of having a sustainable foundry coming out the other end.
A few year later, shareholders will be hesitating to throw more money into Rapidus. Just like LCD and DRAM history in Japan.
Japan should just use $3.9 billion to invest in Solid-State Battery and help Japan car industry to another level.
To once again reprise my role as the devil's advocate, what is the alternative if you want a national champion for advanced CMOS logic? SMIC is lagging even intel 10nm by 4+ years to market even with rampant TSMC/Samsung poaching and an effectively infinite budget. I can't think of any other way to jump start other than licensing IBM tech. Now if we want to have the discussion of this is a pointless endeavor and Japan shouldn't even try, I think that is a different can of worms. Time will tell I suppose. I might just be over optimistic/trying to hope something into existence, but I think it is a project that is worth perusing. But it is not for the faint of heart. All one needs to do is see the strain Samsung and intel are under to know that Japan/JV partners need to slam 100s of billions into Rapidus over the next 20 odd years before they can even hope for the possibility of having a sustainable foundry coming out the other end.
Essentially my thoughts. They are taking the logical steps for a fruitless endeavour basically. I don’t see any scenario where Rapidus is a remotely significant manufacturer of leading logic. It’s just a very strange initiative in all respects for me.
It's indeed possible the poaching was countereffective. If you compare against the memory companies, CXMT and YMTC, arguably they're doing better since they did proper licensing.
