Did chip espionage, IP theft give Samsung its 14nm manufacturing lead?
extremetech.com
extremetech.com
Which apparently he was, as TSMC argues they are now "losing" millions in revenue because this guy helped Samsung get their 14nm node up. And yet at the time they were unwilling to give him a promotion and/or pay raise. What this argues is that TSMC could have doubled the guys salary and come out ahead by millions. Think about that employers when you start "nickel and diming"[1] the engineers that work for you :-)
[1] American slang for charging (or being charged) in many small and ways which add up to a much larger total later on.
When corporations have no recourse about their knowledge workers leaving and working for a competitor and giving up their IP/Trade Secrets, well companies are going to start to take more drastic measures. It's like the cold war, with scientists defecting back and forth, at some point these companies are going to start have spy/counterintel services if they can't rely on governments to deal with it.
Companies do have recourse. It's called treating your employees well enough that they don't leave.
If your company isn't smart enough to know who is important and how to keep them happy, then your company is going to have problems. Technical staff, especially, tend not to jump for small amounts of money or minor problems.
As for technical workers, you simply can't prevent knowledge from leaving when they walk out the door. Especially for things like semiconductor manufacturing research, the knowledge of what was tried and didn't work is often more valuable than the knowledge as to what did work. How in the world would you stop that?
Finally, he left in 2011. You're going to have a pretty difficult time convincing me that Samsung's current 14nm node is solely base upon TSMC manufacturing that was only at 28nm at the time.
Overall, it sounds like TSMC had an internal political battle over what direction the semiconductor process research should go. Liang lost the battle and left. It seems that Liang was correct from an engineering standpoint, and the folks who stayed behind at TSMC were wrong. Sucks to be TSMC.
Greater note: I've been convinced for some time that the non-enforcablility of non-competes is "Silicon Valley's" unique advantage. Given the steadily increasing difficulty of doing business in the state, it's almost certainly been critical in keeping that Golden Goose alive and somewhat healthy.
Except one person is part of a team that produced the IP. So they would always have access to IP worth more than their salary (since salary x NTeamMembers must be <= 100% of the IP value).
So even if the team members were paid every dime of the value they've created, a single person would always be able to earn more by stealing IP/trade secrets.
First wrong assumption: that one person can carry 100% if the IP with them. That's simply not possible for anything remotely complex. Companies regularly have trouble managing software when they own the people and the source code. Any single individual only can carry a small portion of the knowledge.
Second wrong assumption: technology is complete when "stolen". Something like a semiconductor process is always live and being debugged. In 4 years, that knowledge will move a VAST amount and what you "stole" has no value.
Third wrong assumption: a company can incorporate the "stolen" knowledge into its business. Semiconductor processes rely on many things which are specific to a company. Back when Digital Equipment Corporation was working with Samsung to second-source the 21264, it took dozens of colocated engineers almost 24 months to work out and paper over the differences between the semiconductor processes.
As I pointed out, this sounds like there was a political struggle for technology direction at TSMC and Liang lost. Really, at that point his ONLY option is to leave. His employer has said "we don't believe in your assessment".
The fact that he turned out to be right and led Samsung to leapfrog TSMC is TSMC's own fault.
In addition, people are discounting the amount of money Samsung spent to develop that process WAY too much. Samsung is a $150 billion a year company and can allocate vast resources when it decides to. TSMC is a lot smaller.
Not a lawyer, but I gathered that when trade secrets are properly handled, you have legal recourse against an employee (or ex-employee) who discloses them.
they tried to have the cake (don't patent, don't publish) and eat it too (not pay people and processes enough to protect your trade secrets)
"We have essentially no patents in SpaceX. Our primary long-term competition is in China," said Musk in the interview. "If we published patents, it would be farcical, because the Chinese would just use them as a recipe book."
This is a copyright vs. patent sort of thing. What Phoenix copied using the "Chinese Wall" technique is the sort of thing that might be covered by patents, that is, ideas. The IBM BIOS was a creative implementation/expression of those ideas, and that code was copyright protected, as was Phoenix's independent version of the code that implemented the same ideas.
(genuinely curious if this style of blatant espionage still goes on today -- I have some stories from the 80's)
Don't the various semiconductor companies have massive stocks of patents, for just this sort of situation?
I have no idea how that plays out in patent-land, but patents are per-country. E.g., "U.S. Patent #123"
Trade secrets are essential or else nobody would innovate -- they'd just steal each others' designs until nobody could afford to invent anything. Fortunately a short non-compete (during which engineers should be paid compensation for taking time off between employers) is enough to exhaust a lot of cutting edge insider knowledge. The accusation here is that an engineer cheated his non-compete.
Competitors suits are only a good idea when you have a clear patent advantage. That is why Apple and Google got blitzed during the smartphone wars. They didn't have a phone patent portfolio like everyone else, since they were new players. It's the reason Google bought Motorola.
Also, semiconductor processing is hard to reverse engineer so there are legit trade secrets that are kept.
To be fair, Apple pretty much started that one on their own. It was just business as usual and didn't turn into war before Apple decided they owned usability and the critical rounded corner IP.
If you're going to make some strong claim like "Samsung stole IP", phrase it as a damn declarative statement. Phrasing it as a question instead only suggests to me that the claim resulted from rectal extraction and is entirely unfounded.