Samsung slashes chip production after 96% profit nosedive

The situation might lead to good deals for PC owners.

Samsung chip division has been suffering since Apple dumped them and they no longer get a free tip off, look inside at what Apple is doing before hand.

danox said:

Samsung chip division has been suffering since Apple dumped them and they no longer get a free tip off, look inside at what Apple is doing before hand.

What are you talking about? Apple doesn’t have anything special aside from high volume. There’s no tip-off that Samsung could get from Apple that would be worth anything. Designing chips is child’s play and intellectually easy compared to fabbing, as it’s mostly an economics game of selecting a fab’s cost-yield trade-off. At most, Samsung would have some idea about core configuration that Apple is intending on going with, but this is hardly any competitive advantage, as again, design doesn’t matter. All Apple is doing is making some trade-off with the fab (IE. increasing core size will lower yields and raise costs). There’s no trade secret or hidden technology that Apple has which is what I think you’re incorrectly implying.

Apple moving away from Samsung fab has nothing to do with trade secrets, as Apple doesn’t have any technology worth stealing. It has to do with Samsung’s misteps with FinFET and applying EUV. Until recently, their FinFET nodes had poor yields, which is why Apple moved away from them. There’s no way to accommodate Apple’s volume demands if yields are subpar. If (And it’s a big if) Samsung makes a comeback with GAAFET, Apple will come back along with other fabless companies. 

That said, Samsung has never been a strong logic fab company. Their strength lies in memory fab, and aside from this downturn, they were posting $10B quarterly profit. The semiconductor industry has cycles of booms and bust. We’re currently in a bust cycle so all memory manufacturers including micron and Hynix posted billions in losses. At the very least, at least Samsung’s other divisions saved them from posting a loss.

TeslaDomination said:

danox said:

Samsung chip division has been suffering since Apple dumped them and they no longer get a free tip off, look inside at what Apple is doing before hand.

What are you talking about? Apple doesn’t have anything special aside from high volume. There’s no tip-off that Samsung could get from Apple that would be worth anything. Designing chips is child’s play and intellectually easy compared to fabbing, as it’s mostly an economics game of selecting a fab’s cost-yield trade-off. At most, Samsung would have some idea about core configuration that Apple is intending on going with, but this is hardly any competitive advantage, as again, design doesn’t matter. All Apple is doing is making some trade-off with the fab (IE. increasing core size will lower yields and raise costs). There’s no trade secret or hidden technology that Apple has which is what I think you’re incorrectly implying.

Apple moving away from Samsung fab has nothing to do with trade secrets, as Apple doesn’t have any technology worth stealing. It has to do with Samsung’s misteps with FinFET and applying EUV. Until recently, their FinFET nodes had poor yields, which is why Apple moved away from them. There’s no way to accommodate Apple’s volume demands if yields are subpar. If (And it’s a big if) Samsung makes a comeback with GAAFET, Apple will come back along with other fabless companies. 

That said, Samsung has never been a strong logic fab company. Their strength lies in memory fab, and aside from this downturn, they were posting $10B quarterly profit. The semiconductor industry has cycles of booms and bust. We’re currently in a bust cycle so all memory manufacturers including micron and Hynix posted billions in losses. At the very least, at least Samsung’s other divisions saved them from posting a loss.

Somebody is very, very high on their own supply. If chip design were “easy”, “child’s play” and “not worth stealing” then these clowns like Qualcomm, Samsung, Intel, or the Chinese knockoffs would have done it already. They didn’t. Fact remains only Apple’s silicon design has made the massive compute power-per-watt gains that has shepherded in 20-hour laptops. Once again, Apple has led the way the rest of the industry follows. That’s hard for a lot of people to accept. Oh well. 

StrangeDays said. 

Somebody is very, very high on their own supply. If chip design were “easy”, “child’s play” and “not worth stealing” then these clowns like Qualcomm, Samsung, Intel, or the Chinese knockoffs would have done it already. They didn’t. Fact remains only Apple’s silicon design has made the massive compute power-per-watt gains that has shepherded in 20-hour laptops. Once again, Apple has led the way the rest of the industry follows. That’s hard for a lot of people to accept. Oh well. 

LOL... I see you're another follower of the Apple "custom design" cult. What are your credentials? I have worked as a chip designer from 2000-2014 and have experience with TSMC, Samsung and Global Foundries fabs. I retired around the time Samsung was heavily pushing us to adopt their new V-NAND into our AP, though we turned them down and stuck to their planar NANDs due to the high cost of the former. 

Chip design is literally just a budgeting and cost-optimization procedure much like how you budget for parts when you build a PC. Given the same budget and same goals, 2 chips from different companies will perform equally on the same fab. There’s nothing intellectually difficult about chip design despite what Apple wants you to think, nor does Apple possess any worthwhile chip technology trade secrets.

Performance is all based on fab, not design. Apple's disappointing M2 performance gains is proof (Notice how the +20% performance is the exact same increase in TSMC's transistor density). Intel’s performance stagnation at 14nm despite numerous design changes is proof of this. They didn’t see any improvements until they sorted out their 10nm issues. Qualcomm, Intel and Samsung were using inferior nodes for their APs while Apple was given TSMC’s best nodes due to the volume Apple brings. 

Furthermore, Qualcomm chose to prioritize their GPUs over their CPUs, which is why the S8G2’s GPU completely dumpsters the A16. If Qualcomm wanted, they could have easily sacrificed their GPU performance for more cpu performance and matched the A16. It’s not hard. It’s like changing your order from a pepperoni pizza to a sausage pizza.

There is nothing to steal from chip designers because they literally have no technology… chip designers bring volume to a fab to keep costs down and utilization up. They should be thought of as salesmen more than engineers or scientists. I keep bringing up pizza in this analogy, because it's such an apt comparison. You can think of TSMC and Samsung as Dominos and Papa John's while Apple, Qualcomm and other fabless companies are salesmen. The salesmen have their own unique client base - Apple's customers might like pepperoni while Qualcomm's customers might like ham. Apple and Qualcomm go to TSMC and Samsung and asks for specs for a pepperoni pizza and a ham pizza. Based on the fab's technology, the quality of the pizza will differ. The fabs will have their own technological trade secrets that determines the quality of the pizza.
This is what "designing" is - it's just ordering a configuration that suits your needs. There's no tech secret. There's no hidden ultra technology that only Apple has. If Qualcomm wanted, they could make what Apple is making, but it wouldn't be worth it, as their customer base might want something different.

This is also a reason why Tim Cook is worried about lower sales and is rushing out a defective garbage VR headset - they need to provide enough volume to continue receiving favorable treatment from suppliers. Apple’s strength is in providing volume and buying the best tech from their suppliers. Apple provides so much volume that they receive exclusive first access to cutting-edge tech. Just look at how Samsung gives Apple their best screens first before they sell it to others. Their entire business model is predicated on sales volume, not tech development. Apple has never been great at engineering and hard sciences.

If Apple's sales start falling, they lose that edge and the myth of Apple “custom design” nonsense will break.