TSMC Founder Says China's Semiconductor Industry Still Five Years Behind (scmp.com) 95
An anonymous reader quotes a report from South China Morning Post: In a rare public appearance since retiring nearly three years ago, Morris Chang, the 89-year-old founder of the world's largest contract chip maker, said China is not yet a competitor in chipmaking and that Taiwan should defend its leadership in semiconductor manufacturing. "Mainland China has given out subsidies to the tune of tens of billions of US dollars over the past 20 years but it is still five years behind TSMC," Chang said. "Its logic chip design capability is still one to two years behind the US and Taiwan. The mainland is still not yet a competitor."
In his speech, Chang also took a swipe at US chip giant Intel, describing its recent decision to enter the contract chip making market as "very ironic" because it turned down an opportunity to invest in TSMC more than three decades ago. Contract chip makers like TSMC typically take orders from so-called fabless chip makers like Qualcomm, which design their products but outsource the manufacturing. Chang said he was rejected by Intel when he approached it for funding in 1985. "In the past, Intel was the alpha sneering at us and thought that we would never get big," he said. "They never thought the business of [outsourced] wafer fabrication would become so important today."
Chang said the US is also at a disadvantage compared with Taiwan because it lacks engineers dedicated to the semiconductor manufacturing sector, adding that the "US level of dedication to manufacturing was absolutely no match for that of Taiwan." "What I need right now are capable and dedicated engineers, technicians and operators. And they have to be willing to throw themselves into manufacturing," he said. "In the US, doing manufacturing isn't popular. It hasn't been popular for decades."
In his speech, Chang also took a swipe at US chip giant Intel, describing its recent decision to enter the contract chip making market as "very ironic" because it turned down an opportunity to invest in TSMC more than three decades ago. Contract chip makers like TSMC typically take orders from so-called fabless chip makers like Qualcomm, which design their products but outsource the manufacturing. Chang said he was rejected by Intel when he approached it for funding in 1985. "In the past, Intel was the alpha sneering at us and thought that we would never get big," he said. "They never thought the business of [outsourced] wafer fabrication would become so important today."
Chang said the US is also at a disadvantage compared with Taiwan because it lacks engineers dedicated to the semiconductor manufacturing sector, adding that the "US level of dedication to manufacturing was absolutely no match for that of Taiwan." "What I need right now are capable and dedicated engineers, technicians and operators. And they have to be willing to throw themselves into manufacturing," he said. "In the US, doing manufacturing isn't popular. It hasn't been popular for decades."
They'll catch up (Score:5, Insightful)
What America needs to do is have a government that grows local manufacturing for National Security reasons. The trick is grifters. We only need to look at what happened with Foxconn and that syringe factory [nbcnews.com] to know we need to avoid "America First" rhetoric.
People who lean heavy on nationalism are likely to be grifters and we'll never get anything out of them. We need competent administrators. No more actors, you're literally voting for somebody who's an expert in making you think they're something they're not. Administrators. We still need to watch out for corporate shills, but a little shilling is acceptable in service to the goal of self sufficiency.
Re: They'll catch up (Score:1)
Re: (Score:2, Interesting)
I was thinking about this today. It seems that it takes an extraordinary personality to make it to the highest levels of leadership in the U.S. such as megacorp CEO, billionaire, President, Speaker of the House, etc. I'm not a history expert but it seems there are Great Men (could be a woman, that's just the stereotype) and Psychopaths. Reading history perhaps they are all psychopaths, but the ones I call Psychopaths here are the pathological liars, grifters, and destroyers. What we need is a major public
Re: (Score:2)
"Drama Queen" is a term that is more fitting than "mental case".
Not you are entitled to your own opinion but in my opinion, you haven't a clue about Donald Trump.
A narcissist can function at a very high level, as Donald Trump does, and most professionals that I know and read propose that the basic feeling, sense that a narcissist have is Fear. Much more than than the average person. From this,
Re: (Score:1)
Re: (Score:2)
A personal disorder is a mental case.
Re: (Score:2)
Indeed. Nearly everyone just enabled his behaviors and helped him feed that narcissism. I'm not aware of any constructive way of dealing with a person like this, other than to be manipulative and narcissistic in return, which doesn't sound fun to me. Can you really talk someone out of being narcissistic with psychotherapy?
It doesn't take an extrodinary personality (Score:3, Insightful)
Clinton seems to have come from honest to God white trash, but he got the connections he needed for politics from his advanced degree and Ivy League education.
Oh, and look into "Hollywood Royalty" sometime. Every wonder how Super Star Actor (and Actual Cannibal) Shia LeBouf gets to be in movies when he kills every scene he's in?
Basically, we have a rul
Re: (Score:1)
Re: They'll catch up (Score:4, Insightful)
No, nations need to be independent on technology and medicine. Collaboration is ok, but no sane nation depends on other nations on these two issues because it is the recipe for a disaster.
When shit hits the fan, say a war or a pandemic, every nation will think first and foremost to their own citizens, that's the whole point of having nations after all, which means that is your nation can't make vaccines/surgical masks/chips/etc you will be left alone exactly as it happened during this pandemic.
Stop thinking that the world is a happy collaborative place, it isn't, you are living in a dangerous delusion and reality doesn't give a fuck about your fantasy world.
Re: They'll catch up (Score:5, Insightful)
"Independence" really is only possible for large nations. The majority of nations just don't have the populations to have that kind of "independence".
Which I -as an American- think is a really good argument for consolidation of smaller nations and Federalizing governments. Like the EU. It really seems to me like they would benefit A LOT by Federalizing and consolidating their armed forces, for example. I do not know what benefit they get from having 20+ separate armies that don't really work together very well at all. If they federalized they could become a real formidable National Security force in the world and easily rival the hard power of the US and China.
Re: (Score:2)
I do not know what benefit they get from having 20+ separate armies that don't really work together very well at all.
It prevents them from getting involved in foreign wars that, in hindsight, are usually regarded as stupid.
Re: (Score:2)
They also turn their military against their own people like in Myanmar.
There's no perfect system. But building on a foundation of cooperation between governments and guaranteed universal rights is a good start.
Re: They'll catch up (Score:1)
>EU
>Myanmar
Wat..
Re: (Score:2)
I don't know why you are upset. I volunteered that it was an opinion derived from my own American point of view.
Get over yourself.
Re: (Score:1)
Re: They'll catch up (Score:3)
>Limitations on individual rights in some EU nations have expanded to become limitations on individual rights in all EU nations.
This statement is completely contrary to how EU law works. EU directives start at the EU level and are transposed into national legislation. regulation have direct effect. There is no ability, from being in the EU, for civil rights restrictions in one member state to affect another member state. There is simply no legal basis for this. It has to be done on an EU level.
Re: They'll catch up (Score:5, Informative)
That is complete BS. All three American vaccines are being manufactured and shipped to nations all over the world. Only the US stockpile that the US Federal Government purchased has rules against export. The manufacturers themselves are not limited by those rules.
https://www.bbc.com/news/world... [bbc.com]
pure BS from dmay34 (Score:2, Insightful)
That link says nothing about America exporting doses at all. In fact America made a big fuss about not even exporting the doses it wasn't even going to use itself, preferring to just hoard them.
Of course your lies will be modded up though.
Show where America has exported vaccines...
Up until the end of March, zero American vaccines were exported. [statista.com]
The United States, which has been devastated by Covid-19, is one such example where its vaccine policy seems to be an example of "America First". Data compiled by Airfinity shows that out of the 164 million Covid-19 vaccines the U.S. manufactured up to March, none were exported. The UK's situation is similar with 16 million doses produced and zero exported.
By contrast, the EU has exported 42 percent of the 110 million doses it has produced so far while China has also sent just over half of its 229 million doses overseas. India is another example with 44 percent of its 125 million doses going abroad.
Re: (Score:3)
I can't believe a post so wrong was modded +5 informative.
The only exports from the USA that I am aware of are a very small lot of AstraZeneca vaccine exported to Canada and Mexico.
Since AstraZeneca is not yet approved in the USA, they have no use for it. And it's not a sale or a gift: it's a loan. Canada and Mexico will have to provide back the same amount of doses at some point to the USA.
Re: They'll catch up (Score:1)
No doses are being exported from the USA.
Why do you call these manufacturers American anyways? They are global corporations. BioNTech developed the vaccine in Germany. Janssen in the Netherlands. Only Moderna was developed in the USA.
Re: (Score:2)
AFAIK, the US banned the export of all Covid vaccines, and also vaccine ingredients.
Vaccines available for export are not manufactured in the US. For example, Moderna vaccines supplied to Canada are made in Europe.
In other words, US companies are allowed to make vaccines elsewhere, with materials sourced elsewhere, but everything made in the US stays in the US, so it is still an export ban. A ban that other countries don't have. The justification for it is another debate.
Re: (Score:1)
Re: They'll catch up (Score:2)
You need to look at history, your view has resulted in far more deaths and wars than what we have today. Comparatively, we have lived in a "luxury of peace". It may not seem as such, because we live in the Information Age where every little detail is permanently recorded and every single person has massive amounts of access and free time to dabble in the art of discussion with everyone else in the world.
Historically, those with armies took from those without. Those who focused on economies and people could
Re: (Score:2)
because their government is actively trying to promote national interests.
There's too much bad research in China halting progress. Like that one "researcher" who claimed to have developed a home grown chip, only to be found out just literally scratching off the logo of commercially available ones. https://en.wikipedia.org/wiki/... [wikipedia.org]
And I suspect officially sanctioned industrial espionage to steal details from manufacturers and designers of chips is too alluring, and so they'll take longer to actually figure out how to do things on their own. https://www.wired.com/story/ch... [wired.com]
Re: (Score:2)
For advanced R&D where the ideal motivation for workers is not aligned with pure self enrichment, their cultural immorality is a huge drag on business. Trying to impose morality on the population through big brother, but having it be "whatever the CCP says it is today" isn't going to work.
Re: (Score:2)
their cultural immorality
What cultural immorality? It's capitalism. Plenty of American companies do a lot of immoral stuff - and get praised for it as "freedom". It's not cultural. It's economics.
You get that kind of scam work everywhere (Score:2)
Re: (Score:1)
This always happens when development programs as accelerated, fraud goes on, gets exposed and scammers get long sentences.
There is a error in the article, they are not so much 3 to 5 years behind as more like they will take 3 to 5 years to catch to where it will be in 3 to 5 years. There is a difference in the statements, one implies they are not catching up and will remain behind and the other reflect reality, they will catch up in 3 to 5 years and likely accelerate past the west.
In some segments of the ma
Re: (Score:2)
Re: (Score:2, Insightful)
The politicians are only as good as the electorate. Let's face it. The American voting population is a bunch of idiots. On the left, they are a religious cult. On the right, they are also a religious cult, though maybe not as zealous as the left. Meanwhile, everyone has been told that they are as smart as anyone else and "their voice counts." No it doesn't. And it shouldn't, because they are idiots.
Unfortunately, this is a ratcheting, positive feedback process, and there is little hope that it can be
Re: (Score:2)
What America needs to do is have a government that grows local manufacturing for National Security reasons.
The US has done this in the past, e.g. the Heavy Press program [wikipedia.org]. This ran 1950-1957 and built the largest forging presses and extrusion presses in the world. These are still in use today and a vital part of the aerospace supply chain.
Re: (Score:1)
Re: (Score:1)
Re: (Score:1)
Innovation is a diminishing asset. (Score:2)
Innovation is a diminishing short-term asset. For the most part in desktop and laptop computers chip design and processing power available from chips designed 10 years ago is still adequate for most people. Outside of processors most chips (controllers, etc.) really don't need much of any improvement at all and haven't for a long time.
Unless AI comes and really truly changes the development game and we start seeing crazy leaps in power and efficiency, I can imagine that in the next 5-10 years the cost to ad
Re: (Score:2)
Only if you stay in the same business, making the same things.
E.g. making oil lamps is not gonna get much better anymore.
But quantum dot LEDs are quite a different thing.
When we get the first publicly sold neuron-compatible bionic photonic quantum computer, there'll be so much innovation headroom, it won't matter if you're Intel, TSMC or somebody completely new.
Re: (Score:2)
Innovation is a diminishing short-term asset. For the most part in desktop and laptop computers chip design and processing power available from chips designed 10 years ago is still adequate for most people. Outside of processors most chips (controllers, etc.) really don't need much of any improvement at all and haven't for a long time.
Unless AI comes and really truly changes the development game and we start seeing crazy leaps in power and efficiency, I can imagine that in the next 5-10 years the cost to add more processing power into a chip will cross the threshold of value added, and device makers will just opt for a cheaper chip that is perfectly adequate. And that chip will likely be made in China.
this is unlikely to be true at all. Chips have gotten massively faster since 2011 and we have seen many applications that were not possible. If you look purely at gaming. GPUs have gotten massively faster which enable much better graphics at a much higher resolution. You would not have been able to drive VR at the latency you see today. In HPC, the problems that were considered mid size in 2011 are essentially considered small today.
But you also forget that better chips is not only faster chips but also low
Re: (Score:1)
Gate speed has not gone up much in the last 10 years. Density has increased greatly, resulting in more computing power on a single chip at lower energy per computation.
Just looking at Intel processors, single thread performance is a little more than 2X as fast as it was 10 years ago, but multithread performance is 10X to 20X as fast.
Re: (Score:2)
You are misunderstanding what innovation means.
You are thinking of incremental improvements. Important, but not innovative.
Innovation is coming up with new ways of doing things, and even new things to do. It's a fresh start in a new direction.
Only 5 years? (Score:5, Insightful)
Re: (Score:2)
If it's 5 years and they can keep up the rate at which they were going so far, it won't take long until a break even point is reached. China has so much resources that it's difficult to compete with them on a national basis alone.
If it wasn't for the CCP, which I simply can't bring myself to trust, I'd get rid of my shares in some of the Western high tech corporations and invest in Chinese semiconductors instead.
Re: (Score:3)
Well except everyone else isn't going to cease development of their own just so China can catch up. The break-even point is a moving target.
They don't have to catch up (Score:2)
5 year old semi-conductors are pretty powerful. More than enough for practical purposes.
They may not be competitive on an international market. But they sure stop any threat of an embargo being dangerous.
China's vulnerability is not technology but food. They import a lot, and the people have gotten used to eating meat. There is a limit to how much food can be grown on a given area of land of a given quality.
Re: (Score:2)
They're all just bottlenecked by how fast ASML can ship out fabrication equipment. China is fucked because the Dutch government won't let ASML export anything capable of 16 nm or lower.
Re: (Score:2)
They're all just bottlenecked by how fast ASML can ship out fabrication equipment. China is fucked because the Dutch government won't let ASML export anything capable of 16 nm or lower.
Absolutely true.
However, remember when the US banned China from joining the ISS? Now China is about the send out their own space station up.
Forcing China to develop their own by refusing to sell them only works for a while, then when they eventually caught up, you will find that not only did you lose a huge customer, in its place you get a huge new competitor in the market. In 5-10 years time, we will be reading similar articles about how China is only 5 years behind ASML. Worse, they will drain all pote
Same level as Intel (Score:2)
So in fact, they are at the same level, and just overtaking Intel.
Of course they are (Score:1)
All they need to do is assume ownership of plants such as those of TSMC
Re: (Score:2)
You know TSMC's plants are in Taiwan and that's not China, right? Not to anyone but China, and the rest of the world is making sure of that, anyway.
Oh, and just because you manage to take over an alien spaceship doesn't mean you can fly it.
Re: Of course they are (Score:1)
The Nationalists in Taiwan consider themselves China. In fact the basis for their government is that they are the legitimate Chinese government in exile.
Re: (Score:1)
Not to be pedantic here, but this is not an "in exile" situation. Exile means they are pushed outside of their territory, while Taiwan was returned to the Republic of China in 1945 after WW2 (and retreated to Taiwan in 1949, four years later). The PRC/ROC situation is basically the same as North/South Korea, except highly asymmetric.
Re: (Score:1)
The main point is, the Nationalists in Taiwan consider themselves the legitimate government of China.
They could probably do a pretty good job taking over. How they would convince a taxidermist to preserve Xi for public appearances remains a problem.
Re:Of course they are (Score:5, Insightful)
> and the rest of the world is making sure of that, anyway.
That's what they said about Hong Kong. Meanwhile China is engaged in military exercises to surround Taiwan with warships:
https://www.newsx.com/national... [newsx.com]
The USS John S. McCain is in the Taiwan Strait, but if China invades, NO other nation has the appetite to come to Taiwan's defense.
Meanwhile, the US is entirely reliant on both China and Taiwan for its infrastructure, due to forty years of government policy that sold out the American Citizens for privatized profits.
The smart thing to do is to become self-sufficient among the Fifty States.
Re: (Score:2)
Hong Kong is an apple/brick comparison. You can go to Wikipedia and read the historic section on Hong Kong and Great Britain.
And yes it's in every countries interest that's using TMSC silicon to support the status quo. Not that an invasion will be anything but a pyrrhic victory for China.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Missing the point with that joking.
We think it's... *stupid*.
We prefer working smart, and get a better product done in 2 hours a day, 3 days a week, with little effort, because we've been smart before and created something that lets us do that... like automation that isn't just creating wealth for a few leeches but for *us*... over working your ass off like a troglodyte and a slave.
Of course just because you *can* achieve in with that little, doesn't mean we're going to settle for that. We can still work 8
Re: (Score:2)
No one's calling is IC manufacturing (Score:2)
ASML gets buy without people with a culture of suicidal work ethic and it's not like IC manufacturing is some kinds of fashionable industry like say game development which attracts kids who let themselves be abused.
Create the demand and the engineers, technicians and operators will come. It's a job and they want a pay-cheque.
When the CCP takes over Taiwan... (Score:2, Interesting)
The desire to conquer and rule the World is a driving force of all Totalitarian Regimes -- reading history leads to this conclusion!
Re: (Score:2)
Yeah, I'll be flying in on my pig, straight from a ski resort in hell.
Re: When the CCP takes over Taiwan... (Score:1)
When China takes over Taiwan, the gap will extend to 7 years or more.
Re: (Score:2)
No it will not. BUT if the infrastructure is destroyed there will be a gap FOR US to recover from.
TSMC is only part (Score:4, Interesting)
most of the engineering is done elsewhere
Netherlands and Japan own most of the production lines equipment
the only near fully integrated manufacturer is Samsung, thats the real powerhouse
Re: (Score:2)
I always wonder why the world doesn't consider ASML's quasi-monopoly a threat.
It's pretty much a single point of failure to almost all industries and everything there is by now.
Does anyone even know any competitors with equal capabilities? (Say... EUV...)
Re: (Score:2)
ASML has strong competitors though, especially AMAT.. also Lam Research and Tokyo Electron could step up.
Re: (Score:1)
Re: (Score:2)
TSMC works VERY closely on the engineering with their suppliers. You don't just buy an EUV system like buying a printer from Walmart.
https://www.tsmc.com/english/d... [tsmc.com]
what does it matter (Score:2)
China can still manufacture pretty good chips for their own use. as they're less powerfull, china will become an expert at optimizing their code...
last 5 years of what? (Score:2)
Oh my, they're 5 years behind? What advances have we had in microchips in the last 5 years? How to add hardware wireless KVM back-doors to every processor? A processor from 2011 can do almost everything a processor does today and nearly at the same clock frequencies. The only reason hardware can't run say Windows 10 is because of forced obsolescence through refusal to write updated drivers (and purposely adding driver incompatibilities in the first place).
5 years is very short in a stagnant pool.
Re: (Score:2)
During those 5 years, TSMC process technology went from 14 nm to 5 nm.
Essentially, mainland China is even further behind Taiwan (TSMC) than the US (Intel): Intel can't really volume produce bettern than 14 nm, but at least they can make some working chips.
For comparison: 5 years is more than the East was behind the West in semiconductor manufacturing at the end of the cold war: AFAIK Japan (Fujitsu) produced the first 1 MBit DRAM chip in 1985, and Eastern Germany made its first such chip in 1988 (the U61000
last 5 years of perception. (Score:2)
From the article. [csmonitor.com]
“A year ago it was all about how America breaks the rules ... while we are a ‘play by rules crowd,’” says Michael Adams, the president of the Environics Institute, which measures Canadian attitudes. Now the narrative centers around just how much of a global leader in science, manufacturing, and distribution the U.S. is while Canada waits.
“You need a balanced view,” he says. “We – the world and Canada at the head of the list – are benefiting from American innovation and an American can-do philosophy. You can’t just look at America through all the problems they have.”
Competitor in what? (Score:2)
The biggest world shortages currently are not in the cutting edge of semiconductor manufacture, but rather the world relies on processes that are 10+ years old. I'm happy TSMC is a world leader in producing ultra high performance CPUs and GPUs but that's not really what the world needs right now.
Asia belongs to Asians, not NeoCons. (Score:2)
The US is not Asia. It is not the natural master of Asia, and who is that master will be decided by Asians, not elderly white rustics who cannot even speak Mandarin.
It's time the US was freed of its proxy empire by losing it. All we get from Asia is devastating economic competition. Asian nations do nothing for the US worthy of mention while we pay for the entire Chinese military we also built the Pacific Fleet to contain. (The US trade deficit exceeds the Chinese military budget.) We don't know why we de
Marketing claims (Score:1)
Five Years Behind TSMC? Then Equal to Intel?!? (Score:2)
Subject says it all. If China is 5 years behind TSMC, then does that mean that they're at par with Intel? ;-)
Re: (Score:2)
No joke. That's exactly what it means!
Zeno's paradox? (Score:3)
Intel has a little catching up to do, but their version of behind is a little special. On Intel's process, 8.5nm would probably be about the same as 7nm on TSMC or Samsung because Intel has more stringent requirements for first generations of a new node process than the others. Of course, since Intel is at 10nm, TSMC and Samsung are quite far out front.
That said, Intel will take bigger risks now. The main reason Intel fell so far behind the others was simply that they screwed up.
Intel was supposed to deliver 10nm far sooner, but they had an entire generational failure which cost the company $12 billion (if I recall correctly). This not only delayed their moves to the new node process they missed, but also made management a bit hesitant about trying to move too quickly in the future. After all, there wasn't much value in rushing since at the time, they had no real competition in the CPU market.
But now, Intel is receiving huge investments from the US government which, if I understand correctly will give Intel the ability to make 2 huge mistakes like their earlier one without a hit to their purse. This should allow them to roll out 3 new processes in rapid succession. And make no mistake, Intel generally has the next several node processes in development at a time. This is because they want to release a new node process every 2 years but it probably takes 5 years to actually develop and deploy a new process. This is also why the were so far behind. When they failed at one generation, it meant the next 1-3 generations in development wouldn't work either.
I expect Intel to be only a little behind the others soon.
Then there's the issue of quantum tunneling. I'm not a quantum physicist, but I've done some googling about this since I first learned of it. I'm not entirely sure I understood it though since quantum physics is a bit difficult to swallow in large bites.
The idea if I understand it (and I'm sure there are others here that understand it better than me) is that electrons travelling along conductors that encounter insulation, rather than penetrating a porous barrier such as you would expect from water passing through a filter, once we reach a small enough size that quantum physics takes over, the electrons leak through barriers because somehow they stop existing on one side the barrier and start existing on the other. It's like warping through walls.
Well, measuring on Intel's definition of a process, this probably should happen quite a bit at 3nm. To move to node improvements following this barrier, it would require redefining the term insulation... or moving to a totally different technology than "legacy semiconductors" and possibly use light. From what I've read over the years, there are many technologies of this sort that are viable alternatives technically, but so far as I know, they've never been scaled to production capacity and no one would consider investing in the technology until such time as making the switch would provide a clear competitive advantage over "legacy". So, we can't start at something approximately equal to 500nm tech and work from there. It would have to be a meaningful improvement over 3nm silicon.
There's also the return on investment issue.
TSMC and Samsung could slow down advancements and rather invest in capacity at current scale instead. The big problem with the current world's production capacity has nothing to do with growth of demand. It's just that more and more companies have moved to new processes. TSMC, Samsung and others have all made attractive deals with their customers that would convince those customer's to produce at 22nm or smaller. This allows them to retool or decommission older facilities. China is able to produce at 22nm now and without the embargo, they have the capacity to do 14nm.
When the US cut China off from using American technology, which so far as I understand includes silicon wafers, it set them back quite a bit and forced them to find either alt
Semiconductor (Score:1)