Samsung to invest $115bn in logic chip biz over next 11 years as volatile memory mart cools Samsung Electronics will plough Won133tn (£116bn, $155bn) over the next 11 years to develop its logic chip business and boost manufacturing lines to produce those non-memory processors to order for other corporations. With Intense DRAM and NAND competition - China is entering the business and smartphone growth is rapidly …
Around $10 billion a year which has been their level of investment lately, and the same as TSMC's even though they are logic only and don't also have NAND & DRAM facilities like Samsung does.
This is like saying you will be driving 600 miles in the next 11 hours, thinking it will sound more impressive than "I will continue driving 55 mph for the next 11 hours like I already was the last few hours".
Yeah but.....isn’t the real point that Sammy are claiming “business as usual” in R&D for logic chips for the next 11 years.
Don’t we all think that probably 5nm will be the last silicon node, or at most 3nm? Either way, the R&D for that should be completed already well inside 11 years from now.
So the question is, what is their vision for silicon logic chips. And what (if any) is their vision for post silicon investment.
No one's roadmap extends beyond 3nm today, but a decade ago no one's roadmap extended beyond 7nm. Besides, memory (especially flash) uses less aggressive processes, so if even if Moore's Law did halt at 3nm for logic, there would still be further progress for RAM and NAND.
While no one's road map extends too far into the future, continuing to invest rather than sitting on what you have (i.e. Global Foundries) is still a significant statement at this point in the semi manufacturing game. ie. Eking out more revenue from your existing investments versus investing significant money into the latest generation that are difficult to develop and optimize for high yields.
If Intel hadn't stumbled at 10nm, I wonder if Samsung would have continued to invest or slowed down it's pace of investment as Intels potential manufacturing capacity at 10nm for CPU's and 14nm++ for memory would have put a lot more pressure on Samsungs flash business.
Intel doesn't use their 10nm or 14nm processes for memory, those troubles don't impact that market. I believe they sold out their joint venture in flash to Micron, the only kind of memory they are making now is Optane.
I doubt Intel had anything to do with Samsung's decision - if anyone's did it was probably Global Foundries because now there are only two commercial foundries following the bleeding edge, TSMC and Samsung. Had GF decided to keep trying, it would have made it more difficult for Samsung to succeed. They are doing great in memory, but are a tiny fraction of TSMC's size on the logic side.
Intel doesn't use 10nm for memory because it's not working, which is the same reason they are still using 14nm++ for CPU's.
In an alternative universe where Intel had a working 10nm process for CPU's, they would be utilising the previous gen for other products.
GF never had the money to invest in next gen - while Samsung doesn't have the high performance process knowledge, the challenge is from TSMC/Intel needing to re-coup the cost of their investments by moving into other areas.
Message to UK: this is how you build an industry.
Not close off the borders, market access, and have preening dilettante buffoons like Boris Johnson and David Davis parade around pretending they offer a "Global Britain".
Not selling ARM to SoftBank and declaring "Brexit Britain is open for business" (T. May).
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
