Samsung Steals TSMC Trade Secrets?

[Arachnotronic]

Very interesting read:

TSMC is now lagging an Asian competitor in logic process technology for the first time in more than a decade and may lose A9 processor orders because of it.

A key force behind Samsung's edge is a middle-aged Taiwanese with bushy eyebrows who has been Samsung's System LSI division's chief technology officer for three and a half years and works at Samsung's R&D headquarters in Seoul. His name is Liang Mong-song, a former senior director of R&D at TSMC's Advanced Modules Technology Division.

To him, and the five former TSMC colleagues he took with him to Samsung, this constitutes a momentous victory.

"If TSMC didn't allow him (Liang) to leave at the time, it wouldn't be in such bad shape today," laments one semiconductor professor close to TSMC's executives.
This one individual, whose changing of jobs so profoundly affected the fates of two countries' major chip makers, has come to be known as Taiwan's No. 1 turncoat.

More here:http://english.cw.com.tw/article.do?action=show&id=14895

 

[witeken]

What incentive does Samsung actually have for a leading edge foundry? They produce lots of NAND, but not so much logic. Are all the investments just to keep Apple as customer?

 

[Arachnotronic]

What incentive does Samsung actually have for a leading edge foundry? They produce lots of NAND, but not so much logic. Are all the investments just to keep Apple as customer?

Why does any company seek to expand its addressable markets?

To deliver growth (and additional value) to its shareholders.

 

[witeken]

Why does any company seek to expand its addressable markets?

To deliver growth (and additional value) to its shareholders.

Sure, but besides Apple, I don't really know more about Samsung's semiconductor business. Guess they're still pretty serious about their Exynos, but just a year or so ago I would have though Samsung would be the next company to drop off, together with or slightly after Global Foundries and IBM, leaving only TSMC and Intel at the bleeding and leading edge, respectively.

 

[NTMBK]

Sure, but besides Apple, I don't really know more about Samsung's semiconductor business. Guess they're still pretty serious about their Exynos, but just a year or so ago I would have though Samsung would be the next company to drop off, together with or slightly after Global Foundries and IBM, leaving only TSMC and Intel at the bleeding and leading edge, respectively.

They could conceivably end up fabbing for Qualcomm too, if their process beats TSMC. Not to mention AMD in the (not unlikely) event that GF screw up again.

 

[Arachnotronic]

They could conceivably end up fabbing for Qualcomm too, if their process beats TSMC. Not to mention AMD in the (not unlikely) event that GF screw up again.

Qualcomm has designed Snapdragon chips on 14nm, so yes. Also, AMD is designing next generation dGPUs on Samsung's 14nm process as well.

 

[NTMBK]

Qualcomm has designed Snapdragon chips on 14nm, so yes. Also, AMD is designing next generation dGPUs on Samsung's 14nm process as well.

14nm Snapdragon is confirmed? Wow, impressive win for Samsung.

V. interesting article by the way; if Samsung is this cavalier with IP, one wonders how much this will affect their foundry business. Do you really trust your competitor with the plans to your next-gen SoC?

 

[Arachnotronic]

14nm Snapdragon is confirmed? Wow, impressive win for Samsung.

V. interesting article by the way; if Samsung is this cavalier with IP, one wonders how much this will affect their foundry business. Do you really trust your competitor with the plans to your next-gen SoC?

I know Qualcomm is designing at both (just look at LinkedIn profiles of Qualcomm engineers), but who knows what the production/sales mix ultimately turns out to be...

At any rate, Samsung has been building Qualcomm chips for a while. 28nm Qualcomm standalone modems for example have been multi-sourced from TSMC and Samsung, according to Chipworks.

 

[dawheat]

I do think an article from a Taiwanese paper does need to be taken with a grain of salt. The ending sentence does make it pretty clear which side they're on:

This is truly a national security crisis. Liang Mong-song's story has sounded the alarm to corporate leaders, executives and individuals.

If true and from the details in the article, Liang should return profit from his golden handcuff shares. I assume if he really was that valuable, Samsung would happily reimburse him. I didn't read anywhere in the article that he was forbidden from working for a competitor - but doing so would forfeit the stock he was given.

NT$100 million is a little more than 3 million USD - pocket change for these companies.

 

[Arachnotronic]

I do think an article from a Taiwanese paper does need to be taken with a grain of salt. The ending sentence does make it pretty clear which side they're on:



If true and from the details in the article, Liang should return profit from his golden handcuff shares. I assume if he really was that valuable, Samsung would happily reimburse him. I didn't read anywhere in the article that he was forbidden from working for a competitor - but doing so would forfeit the stock he was given.

NT$100 million is a little more than 3 million USD - pocket change for these companies.

I don't think it's legal to give trade secrets of a former employer to a current one.

 

[witeken]

I don't think it's legal to give trade secrets of a former employer to a current one.

Idontcare once made a nice comment about this.

 

[positivedoppler]

What incentive does Samsung actually have for a leading edge foundry? They produce lots of NAND, but not so much logic. Are all the investments just to keep Apple as customer?

You need to understand the difference between Asian vs Western philosophy. Western business philosophy is what can I do now to make more money next quarter or a few quarters down the road. Asians thinks more in terms of a few years or even a few decades down the road.

Samsung is one of the (if not the largest) company in South Korea. Their products extend beyond home electronics. They build ships, cars, and who knows what else. It's silly to think that their ambition will not grow. When you own 1/2 the country and is one of the largest exporter on the global stage, it's worth whatever price to be the best at anything you can compete in.

 

[Tuna-Fish]

I don't think it's legal to give trade secrets of a former employer to a current one.

Regardless of whether it happened or not, the law is irrelevant. Any Taiwanese court cannot really enforce any laws against Samsung, and the odds of any Korean court ruling in favor of a foreign company against Samsung are pretty much nil.

 

[Arachnotronic]

Regardless of whether it happened or not, the law is irrelevant. Any Taiwanese court cannot really enforce any laws against Samsung, and the odds of any Korean court ruling in favor of a foreign company against Samsung are pretty much nil.

Yeah, pretty much...:/

 

[dawheat]

I don't think it's legal to give trade secrets of a former employer to a current one.

I've always been a bit hazy on the lines between skills, even if acquired on the job, vs. trade secrets. Especially in specialized industries, assuming he didn't take schematics and formulae, but knowledge of successful approaches and applied them to his new work, where do you draw the line.

I assumed most companies protect themselves through non competes, but his seemed loose as long as he was willing to forfeit stock.

 

[Abwx]

According to the quotes below TSMC seems stupid enough to use arguments that are actualy counter arguments to their claims...

As evidence, TSMC presented a report by outside experts offering a structural analysis of key manufacturing processes of TSMC, Samsung and IBM products. The process involved analyzing transistors 1/10,000th the size of a hair with the most advanced electron microscopes to compare the main structural characteristics and constituent materials of the three companies' last four generations of chips.

But what shocked TSMC was that in the following years, the characteristics of Samsung's 45nm, 32nm and 28nm generations increasingly mirrored those of TSMC's chips. The report listed seven key process areas that were nearly identical for the two companies.

So obscure experts have the mean to analyze several processes but Samsung enginers and scientists are not supposed to be up to thoses decidely undeservedly unknown specialists..?..

Worse :

But what shocked TSMC was that in the following years, the characteristics of Samsung's 45nm, 32nm and 28nm generations increasingly mirrored those of TSMC's chips. The report listed seven key process areas that were nearly identical for the two companies.

So while at Samsung the scientist still had knowledge of TSMC s process that were released years after his departure from this firm..?.

What is obvious is that TSMC process were deeply influenced by this scientist, once he worked at Samsung his technical views didnt change overnight, he kept materializing his inventivity at Samsung, TSMC just aknowledged that their current process is based on this scientist ideas and that once he went away they didnt had people as smart to keep on advancing theses ideas.

 

[Headfoot]

Yeah, pretty much...:/

This is a myopic view. They may not have jurisdiction but their laws likely align in many respects. I don't know what IP treaties South Korea and Taiwan have, but they're both part of the WIPO TRIPS intellectual property treaty which has provisions for trade secret misappropriation.

It's not so simple as forum warriors and journalists with an angle make it out to be...

I've always been a bit hazy on the lines between skills, even if acquired on the job, vs. trade secrets. Especially in specialized industries, assuming he didn't take schematics and formulae, but knowledge of successful approaches and applied them to his new work, where do you draw the line.

This is the perennial problem in trade secret law. Even in the US, this is in the domain of the States, so the answer of the exact line is different from State to State. It's likely much different in each country. Typically, developing economies favor the employee's mobility so that high tech employees can move to their country. Developed economies favor somewhere in between. A few States in the US even have a doctrine which holds that you don't even have to have an identified use of the secret knowledge if you are a sufficiently high-level executive -- it is imputed to you because (paraphrasing) you can't help but use that knowledge.

 

[turtile]

If this guy was at the forefront of FinFET, I don't think he really needed to take any IP from TSMC since he has the knowledge.

Unless he stole documents from TSMC, I don't think he did anything wrong. He's not TSMC's slave. If he doesn't like working there, he can work somewhere else and use his knowledge there.

Maybe his departure is the reason for the 20nm failure...