As Hynix CEO Jong-Kap Kim says in his message to the shareholders, Hynix keeps seeing image sensors is its future growth platform and plans to strengthen its competitiveness in CMOS sensors.
Well, there is a logic here. When DRAMs were profitable, Hynix decided to concentrate on them and span off its non-memory businesses, including CIS. In the meantime, Hynix-Magnachip leading CIS designers left for Samsung. To regain its lost design capability, Magnachip hired Jason Hartlove and built a very fine team in the US. However, it did not work well, so Magnachip got rid of, first, Jason and then whole his team. On Hynix side things also deteriorated due to well known free fall of DRAM prices. So, Hynix decided to win back its former laurels of the world's biggest mobile sensor manufacturer, this time with help of Siliconfile. Hope it explains Hynix logic.
Also remember that after the creation of Magnachip (roughly fractured along the old LG-Semi and Hyundai-semi company lines) that Hynix was excluded from working in CIS until last year, at which point they made a big announcement of intent. Having CIS run in older memory fabs makes sense as evidenced by a certain company.
Image sensor .. I think you add too much weight to the SiliconFile acquisition.
> I think you add too much weight to the SiliconFile acquisition.
If not Siliconfile, what could be a better option for Hynix? Probably, the best solution would be to hire few key persons from Samsung and build a team around them. Actually, Hynix was able to attract few knowledgeable Samsung engineers, but too few for a critical mass.
Just as a mental exercise, given Hynix financial restrictions, what can be better than Siliconfile option?
Well, this isn't a either/or sort of thing. Getting SiliconFile probably was a leg up, that certainly was the stated intention when they were acquired. But SiliconFile didn't really have the history so at the time or subsequent to that perhaps an acquisition of people from other groups. For example Samsung's recent move with the Magnachip people and Fossum, could have been Hynix's move (maybe not, but it certainly is a counter point to your stated perception of Samsung strength). Even the acquisition of a hand full of key people is all that would be needed, critical mass be damned. Innovation does not arise from groups, but groups can benefit from a few innovative people.
I was surprised to learn that there is a law in Korea that prevents company H, say, from hiring engineers directly from company S. Company H can only hire engineers that have not worked in the field or for a competing Korean company for at least one year. This makes raiding engineers almost impossible, and it also makes it a challenge for engineers to job hop. This law does not apply in the US though, so raiding and job hopping between Korea subsidiaries in the US is possible. Meanwhile, I am still looking for a few good people for Samsung in San Jose, Pasadena and Korea. -EF
Indeed, the acquisition of Magnachip CIS group was an option for Hynix. My understanding is that Hynix perceived Magnachip as much weaker than Siliconfile in its CIS knowledge and skills.
> Innovation does not arise from groups, but groups can benefit from a few innovative people.
I don't think Hynix looks for big innovation here. I believe it just wants to be best in volume manufacturing - same intention it pursues in DRAMs.
> Company H can only hire engineers that have not worked in the field or for a competing Korean company for at least one year.
Still, indirect transitions are possible and happen.
I'm confused, Hynix spins out MagnaChip which fab'd CIS at Hynix.. Magna rids itself of CIS, now Mynix says CIS is its future growth platform
ReplyDeleteWell, there is a logic here. When DRAMs were profitable, Hynix decided to concentrate on them and span off its non-memory businesses, including CIS. In the meantime, Hynix-Magnachip leading CIS designers left for Samsung. To regain its lost design capability, Magnachip hired Jason Hartlove and built a very fine team in the US. However, it did not work well, so Magnachip got rid of, first, Jason and then whole his team.
ReplyDeleteOn Hynix side things also deteriorated due to well known free fall of DRAM prices. So, Hynix decided to win back its former laurels of the world's biggest mobile sensor manufacturer, this time with help of Siliconfile.
Hope it explains Hynix logic.
Also remember that after the creation of Magnachip (roughly fractured along the old LG-Semi and Hyundai-semi company lines) that Hynix was excluded from working in CIS until last year, at which point they made a big announcement of intent. Having CIS run in older memory fabs makes sense as evidenced by a certain company.
ReplyDeleteImage sensor .. I think you add too much weight to the SiliconFile acquisition.
MR
> I think you add too much weight to the SiliconFile acquisition.
ReplyDeleteIf not Siliconfile, what could be a better option for Hynix? Probably, the best solution would be to hire few key persons from Samsung and build a team around them. Actually, Hynix was able to attract few knowledgeable Samsung engineers, but too few for a critical mass.
Just as a mental exercise, given Hynix financial restrictions, what can be better than Siliconfile option?
Well, this isn't a either/or sort of thing. Getting SiliconFile probably was a leg up, that certainly was the stated intention when they were acquired. But SiliconFile didn't really have the history so at the time or subsequent to that perhaps an acquisition of people from other groups. For example Samsung's recent move with the Magnachip people and Fossum, could have been Hynix's move (maybe not, but it certainly is a counter point to your stated perception of Samsung strength). Even the acquisition of a hand full of key people is all that would be needed, critical mass be damned. Innovation does not arise from groups, but groups can benefit from a few innovative people.
ReplyDeleteMR
I was surprised to learn that there is a law in Korea that prevents company H, say, from hiring engineers directly from company S. Company H can only hire engineers that have not worked in the field or for a competing Korean company for at least one year.
ReplyDeleteThis makes raiding engineers almost impossible, and it also makes it a challenge for engineers to job hop.
This law does not apply in the US though, so raiding and job hopping between Korea subsidiaries in the US is possible.
Meanwhile, I am still looking for a few good people for Samsung in San Jose, Pasadena and Korea.
-EF
EF - that explains another element of attractiveness of SiliconFile for Hynix. They are all ex- Hynix people who went out on their own.
ReplyDeleteMR
Indeed, the acquisition of Magnachip CIS group was an option for Hynix. My understanding is that Hynix perceived Magnachip as much weaker than Siliconfile in its CIS knowledge and skills.
ReplyDelete> Innovation does not arise from groups, but groups can benefit from a few innovative people.
I don't think Hynix looks for big innovation here. I believe it just wants to be best in volume manufacturing - same intention it pursues in DRAMs.
> Company H can only hire engineers that have not worked in the field or for a competing Korean company for at least one year.
Still, indirect transitions are possible and happen.
Gee ... I wonder if Hynix found out about Omnivision's camera cube before or after they put out this announcement?
ReplyDeleteMaybe they wanted to reassure their customer base in China that they aren't going to just pick up and move back to Korea?