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!
Perhaps, too much OT for me. I do not try to figure out Elon's brain and motivation.....putting 10-12 kids in the world, and seeing how he deals with his trans-child, is enough for me to value how he values/deals with people that do not follow his lane..... https://www.nbcnews.com/tech/tech-n...r-daughter-vivian-wilson-interview-rcna163665
I still remember reading about the difficult relation of SteveJ with his Lisa-daughter, very sad for all to only realize/reflect on this when you are on your death-bed. Too many kids suffer from their ego-troubled father......
Perhaps, too much OT for me. I do not try to figure out Elon's brain and motivation.....putting 10-12 kids in the world, and seeing how he deals with his trans-child, is enough for me to value how he values/deals with people that do not follow his lane..... https://www.nbcnews.com/tech/tech-n...r-daughter-vivian-wilson-interview-rcna163665
I still remember reading about the difficult relation of SteveJ with his Lisa-daughter, very sad for all to only realize/reflect on this when you are on your death-bed. Too many kids suffer from their ego-troubled father......
Elon Musk has been known to father 14 children, 13 of whom are alive, according to an article published by People magazine in June 2025. The exact number of children he has fathered remains unknown.
According to South Korean media outlet The Korea Economic Daily, Samsung is reportedly considering further investment in its Taylor, Texas plant. Spec...
www.trendforce.com
According to South Korean media outlet The Korea Economic Daily, Samsung is reportedly considering further investment in its Taylor, Texas plant. Specifically, the company is said to be reviewing the addition of an advanced packaging line alongside its foundry and R&D facilities.
The report notes that Samsung recently secured a 22.8 trillion KRW (approximately USD 16.5 billion) contract to produce Tesla’s AI6 chips at the site. As these chips require advanced packaging to integrate AI memory like HBM, the need to establish such capabilities in Taylor has become more pressing.
In H2 2025, the Foundry Business will ramp up mass production of a new mobile SoC with the 2nm GAA process. It also aims to improve factory utilization and profitability through expanded sales to major customers.
Why do you need a translation? From the picture, you can clearly see that Samsung transports their wafers on carts like any other commercial kitchen. Also, they skipped 450mm wafers and have upgraded to 600mm already.
Nothing you haven't already read: Tesla CEO Musk Samsung agreed to let Tesla provide support
To make the largest impact to production efficiency
I will inspect the lines myself
... to make faster improvements
Could this be one of the key reasons Tesla shifted from TSMC to Samsung? TSMC is well known for its accounts receivable operations that enforced contract payments from its clients. This rigid structure may not align well with Tesla’s payment habits.
"In one of the lien cases, Tesla’s own outside counsel agreed that Tesla is not great at paying on time. He said: “I don’t disagree that it does take Tesla some time to pay, that goes for legal bills, too … I know it full well.”
Yes, when this story first appeared I thought, 'hope Samsung gets the money up front, like Yaccarino should have'. But they probably didn't since they need the positive press of a 'whale customer ' for their Taylor fab more than any guaranteed moneys for the parts supplied.
Plus I think Elon is grossly inflating this deal's value, and Samsung for their own reasons won't publicly discount it; will be surprised if it actually amounts to more than 10 percent lifetime.
Yes, when this story first appeared I thought, 'hope Samsung gets the money up front, like Yaccarino should have'. But they probably didn't since they need the positive press of a 'whale customer ' for their Taylor fab more than any guaranteed moneys for the parts supplied.
Plus I think Elon is grossly inflating this deal's value, and Samsung for their own reasons won't publicly discount it; will be surprised if it actually amounts to more than 10 percent lifetime.
Didn’t the article specifically say the real value was less than half of the $16.B and that the full number was some potential “lifetime deal value”?
Elon literally always does this. He’s the king of over promising and over-confidence. Only when he has very capable lieutenants actually doing the work does anything come out close to what was promised (Space X, for instance). Samsung beware.
The first fab is completed with utilities and infrastructure, and the second fab is maybe 50% completed.
Initial tools that were to be installed in 2025 have been pushed into 2026 (or longer), so there is very little in the way of fab activity going on.
The tool orders will have to be reactivated and delivery dates set, along with which process will be running there - the initial aim of 4nm process appears to have been updated to either 3/2nm.
In addition to @Sherlock's observation above about Samsung being able to supply memory for Tesla's (and X's) needs, there's another pillar of the supplier relationship:
In addition to @Sherlock's observation above about Samsung being able to supply memory for Tesla's (and X's) needs, there's another pillar of the supplier relationship:
This is another example of why even a mighty company like Samsung has struggled in the foundry business. The world’s top image sensor companies, such as Sony (which has its own fabs) and OmniVision, have no problem outsourcing to pure play foundries. If Samsung is serious about succeeding in the foundry business, it must first resolve this conflict of interest problem.
This is another example of why even a mighty company like Samsung has struggled in the foundry business. The world’s top image sensor companies, such as Sony (which has its own fabs) and OmniVision, have no problem outsourcing to pure play foundries. If Samsung is serious about succeeding in the foundry business, it must first resolve this conflict of interest problem.
I'm not quite sure I agree 100% -- the memory side of the business can be a positive differentiator - "guarentee supply", "ensure quality integration", etc. A counter to TSMC's packaging differentiator, for example.
Your point is definitely correct in other scenarios though - Apple buying fab capacity from them would present a COI with Samsung being all-in on the Android ecosystem.
I suspect the politics (and powerplay by the wealthy class that are invested in Samsung) in Korea will prevent Samsung from ever breaking up sufficiently to avoid these Foundry conflicts.
I'm not quite sure I agree 100% -- the memory side of the business can be a positive differentiator - "guarentee supply", "ensure quality integration", etc. A counter to TSMC's packaging differentiator, for example.
Your point is definitely correct in other scenarios though - Apple buying fab capacity from them would present a COI with Samsung being all-in on the Android ecosystem.
I suspect the politics (and powerplay by the wealthy class that are invested in Samsung) in Korea will prevent Samsung from ever breaking up sufficiently to avoid these Foundry conflicts.
Conflicts of interest can take many forms. In my opinion, Samsung’s product capabilities, such as in image sensors and memory, may have helped it win the Tesla–Samsung Foundry deal. But at the same time, this recurring theme is one of the reasons why competitors to Samsung’s product divisions often avoid using Samsung Foundry.
For example, memory manufacturing seems to be a strength in the Tesla–Samsung Foundry collaboration. However, this hasn’t been a serious issue for UMC, TSMC, GlobalFoundries, or Intel Foundry, since none of them manufacture memory. On the other hand, developing advanced packaging technologies at foundries requires close collaboration with memory makers. The successful partnerships between SK hynix-TSMC and Micron-TSMC, in supporting Nvidia and AMD via TSMC's CoWoS are vivid examples. Samsung’s memory division was essentially forced to join the TSMC ecosystem later on.
So my question is: Are Samsung’s product divisions (memory, image sensors, smartphones, etc.) helping or hurting Samsung Foundry’s ability to secure more external orders?