美版知乎论坛发问:美国能在芯片上卡中国脖子多久?荷兰芯片工程师给出答案

搬一篇微信公众号“互联网思想”的文章

“卡脖子”的表述很形象,借用到核心技术中来,那就说明这项技术非常关键,在一个项目或一个产品中,“卡脖子”技术的重要性不言而喻。我们常用的电脑核心芯片、高端手机核心芯片、存储设备、视频系统中显示驱动芯片、数字信号处理设备芯片,及可编程逻辑设备核心芯片等,都是中国自己造不出来的高端芯片,几乎完全依赖进口,在这些产品中,中国自产芯片在全球的市场占有率连1%都不到,甚至绝大部分低于0.5%。

这样的现状下,高端芯片自然就成了“卡脖子”问题,原因是极大的需求量和极低的自供给量,所以美国制裁华为,杀手锏就是芯片。近期在美版知乎论坛上就有这样的一个话题引起各国网友的热议:“美国向荷兰施压,阻止荷兰ASML公司向中芯国际出售EUV设备,最终影响了中国的技术梦”。美国能压制中国芯片发展多久?

让我们也来看看国外民间网友对这个问题的的观点吧!

500

话题内容

500

荷兰半导体工程师的观点

I
think I am the perfect person to answer this. I have been working in
the semiconductor industry for years, currently living in the
Netherlands, lived and worked in China for 3 years, I know a lot of
engineers from ASML (expecially after they hired hundreds of Turkish
engineers).

我认为我是回答这个问题的最佳人选。我在半导体行业工作多年,目前生活在荷兰,在中国生活和工作了3年,我认识很多来自ASML的工程师(特别是他们雇佣了数百名土耳其工程师后)。

Let
me tell you something: High-end semiconductor manufacturing is black
magic. Both the processes and tools used for it are very complex. ASML’s
EUV lithography machine is probably the most complex tool humankind
ever developed since it stopped jumping between trees. It took billions
of Euros and decades of experience to perfect it. Other experienced
lithography machine suppliers failed at it. China has no experience in
high-end semiconductor manufacturing tools with the exception of
one-off/few-off prototypes.

让我告诉你:高端半导体制造是一种黑魔法。它所使用的过程和工具都非常复杂。ASML的EUV光刻机可能是人类发明的最复杂的工具,因为它不再在树之间跳跃。它花费了数十亿欧元和几十年的经验来完善它。其他经验丰富的光刻机供应商在这方面失败了。中国在高端半导体制造工具方面没有经验,除了一次性或少量的原型。

ASML’s
EUV lithography machine. Needs 41 semi-trucks to get transported, costs
$150 million, has 100.000 major parts, has mirrors that need months of
grinding to reach needed smoothness, needs multiple people with PhD’s as
machine operators. Quite high-tech. Isn’t it?

ASML公司的EUV光刻机。运输需要41辆半卡车,耗资1.5亿美元,有10万个主要部件,镜子需要数月的打磨才能达到所需的光滑程度,需要多名拥有博士学位的人作为机器操作员。很高科技。不是吗?

500

EUV光刻机

Unfortunately,
ASML is a very convenient target for the USA. The company uses a lot of
critical parts from the USA but those parts don’t represent anything
significant in the US economy in terms of their monetary value. Chinese
electronics industry still depends on foreign chips so it can not
threaten fabs with banning the sale of chips in China that were
manufactured using ASML tools. Also, China isn’t a big customer of ASML
too. In short, China can not answer with reciprocal sanctions.

不幸的是,ASML对美国来说是一个非常方便的目标。该公司使用了许多来自美国的关键部件,但这些部件的货币价值在美国经济中并不代表任何重要的东西。中国电子行业仍依赖外国芯片,因此无法通过禁止在中国销售使用ASML工具生产的芯片来威胁晶片厂。此外,中国也不是阿斯麦的大客户。总之,中国不能以对等制裁作为回应。

Is China hopeless? No.

中国绝望了? 不!

1-
All of those tools are engineered and made by humans, and the laws of
physics are the same both in the Netherlands and China. If the
Netherlands could, then there is no reason for anybody else to fail with
the correct approach.

1-所有这些工具都是由人类设计和制造的,而且物理定律在荷兰和中国都是相同的。如果荷兰可以,那么其他国家就没有理由用正确的方法失败。

2-
China is filthy rich compared to the Netherlands. Chinese economy is
17x of the Netherlands’, 9x of SK’s, 27x of Taiwan’s, 3+x of Japan’s.
With state support, Chinese fabs and tool makers can hire the top people
from the rest of the world with salaries ASML, LamResearch, AM,
Synopsys, TSMC, Samsung, … simply can not compete with. A significant
portion of these companies’ employees are expats anyway, most of them
are just after money.

2、中国比荷兰富得流油。中国经济是荷兰的17倍,SK的9倍,台湾的27倍,日本的3倍。在国家的支持下,中国的晶圆厂和工具制造商可以用ASML、LamResearch、AM、Synopsys、台积电、三星等公司无法与之竞争的薪水从世界其他地方雇佣顶尖人才。这些公司的员工中有很大一部分是外国人,他们中的大多数只是为了钱。

In
fact China is already doing this successfully with good results. For
example, it already has a working EUV lithography machine prototype,
already caught up with the rest in chip testing, packaging, wafer
production, also its first immersion lithography machine (good enough
for most things) is getting prepared for commercial use.

事实上,中国已经成功地做到了这一点,并取得了良好的成果。例如,它已经有了一台工作中的EUV光刻机原型,在芯片测试、封装、晶圆生产方面已经赶上了其他公司,同时它的第一台浸没式光刻机(对大多数事情来说已经足够好了)也准备投入商业使用。

3-
China is a scientific powerhouse on its own. It is the country with
most patent applications, most research output, graduates more STEM
students than any other country, 2nd largest R&D spender, has 11
universities in top 100. This leads us to my first point. If the
Netherlands could, so can China if given enough time.

中国本身就是一个科学强国。它是专利申请最多、研究产出最多的国家,STEM专业的毕业生比其他任何国家都多,研发支出排名第二,有11所大学进入前100名。这就引出了我的第一点。如果荷兰可以,那么只要给中国足够的时间,中国也可以。

4-
Catching up is much easier than innovating. Knowing something is
possible and having a general knowledge of how it works make things much
easier.

赶超比创新容易得多。知道某件事是可能的,并且对它的工作原理有一个大致的了解会让事情变得容易得多。

5-
Time is on the Chinese side. Technology of semiconductors is close to
maturity/stalling (choose the word depending on your view). If the
development slows (which it does) it gives China the opportunity to
catch-up. If a tech revolution happens, then the playing field evens out
anyway.

5-时间在中国这边。半导体技术接近成熟/停滞(根据你的观点选择这个词)。如果发展放缓(确实如此),中国就有机会迎头赶上。如果科技革命发生了,那么竞争环境无论如何都是公平的。

500

6-
You don’t need EUV for the most things. You don’t even need high-end
processes for the most things. There is more to semiconductors than the
latest smartphone processors, GPUs, and CPUs. Look at iPhone 12 teardown
videos. You will see a lot of chips. Only one of them needs EUV. An
average modern car has 250+ computers inside. That means thousands of
chips. All of them are manufactured using old processes. This is even
more true for military and space applications. Those use very old chips
that are known to be reliable and secure.

6-大多数事情都不需要EUV。大多数事情你甚至不需要高端的流程。半导体不仅仅是最新的智能手机处理器、图形处理器(gpu)和cpu。看看iPhone

12拆卸的视频。你会看到很多芯片。只有一个需要EUV。一辆普通的现代汽车内部有250台电脑。这意味着成千上万的芯片。所有这些都是用旧工艺制造的。在军事和太空应用方面更是如此。它们使用的是非常老旧的芯片,但已知是可靠和安全的。

Conclusion:
Blocking ASML from selling EUV machines to China can hurt Chinese
businesses for some time but in the grand scheme it is insignificant.
The USA needs to run faster rather than keep trying to block China if it
wants to preserve its dominance in tech.

结论:阻止ASML向中国销售EUV设备可能会在一段时间内伤害中国企业,但从长远来看,这是微不足道的。如果美国想要保持其在科技领域的主导地位,它需要跑得更快,而不是一直试图阻止中国。

An update on the Chinese EUV light source: New Options for Synchrotron Light Sources

中国极紫外光源的更新:同步加速器光源的新选择

It seems the basic research is complete and the method completely different than of ASML’s.

看来基础研究已经完成,方法和ASML的完全不同。


500

德国网友的观点

This
EUV thing has been adopted into China’s 14th 5-year plan which has been
started since this year. And for the record, China has never missed a
single deadline for the past 13 5-year plans. So it’s safe to say that
China will have this thing by 2026. So speaking of the US resisting
China’s development, it would be 4 more years for this particular item.

EUV项目已被纳入今年启动的中国第14个五年计划。根据记录,在过去的13个五年计划中,中国从未错过任何一个最后期限。所以可以肯定地说,中国将在2026年拥有它。所以说到美国抵制中国的发展,这个项目还需要4年。

And
the good thing for China having the EUV equipment is that the price of
the equipment will be significantly dropped. So I guess that would be a
benefit of the entire world.

中国拥有EUV设备的好处是,设备的价格将大幅下降。所以我想这对整个世界都有好处。

America
is ALREADY behind China in so MANY aspects and they are still denying
and avoiding the thought that America might one day second to China.
That frightens many Americans.

美国在很多方面已经落后于中国了,他们还在否认和避免美国有一天会落后于中国的想法。这让许多美国人感到害怕。

No
matter how ineffective many of the counter-China measures, e.g. trade
war, Americans have and are to put in place, they will keep doing the
same things and getting themselves hurt before China does.

无论许多繁华措施有多无效,例如贸易战,美国人已经并将付诸实施,他们将继续做同样的事情,他们会主动出击而不是等待中国超越他们。


The Chinese are the most industrious race on earth.

中国人是世界上最勤劳的民族。

Necessity is the mother of Invention.

需求是发明之母

The
Chinese will find a way. The more you suppress them and the more you
try to deny them the technology - they will begin to make it on their
own and they will slowly do better and better.

中国人会找到办法的。你越是压制他们,你越是试图否认他们的技术,他们就会开始自己创造,慢慢地做得越来越好。

Chips is the best example.

芯片就是最好的例子。

As long as Taiwan kept supplying them the chips - The Chinese were happy. They focused on other things

只要台积电继续向他们供应芯片——中国人就很高兴。然后会他们专注于其他事情。

500

江苏南京的台积电工厂

The
Minute Trump decided to threaten the Chinese - they decided to get
their act together and start making their own chips. They will start
with inferior ones but in 10 years - they will outmanufacture Taiwan at
1/3 the cost and take away the market.

特朗普决定威胁他们那一刻起,他们就会齐心协力开始制作自己的芯片。他们一开始会用劣质产品,但10年后,他们的生产成本将是台积电的三分之一,并夺走市场

And
the businessmen will say - “Uigyurs???? Who gives a damn about them. My
shareholders matter” and will migrate from Taiwan to China in 10
seconds.

商人会说:“维吾尔人????谁会在乎他们呢。我的股东才是最重要的”,并将在10秒内从台湾迁移到中国大陆

The US may try again and again but

美国人可能会一次又一次地尝试,但是:

(a) They waited too long. China is too rich today. They have too much money.

他们会等得很久。今天的中国太富裕了。他们有太多的钱。

(b) China has too many tentacles in foreign countries. Thousands of Chinese in various industries who are experts.

中国在国外的触角太多。数以千计的华人专家分布在各行各业。

In
Space alone - China was behind India until 2010 - but today - they have
their own Mapping System for their huge landmass as well as are in the
position of becoming the Third country in the globe to land on Mars -
having landed on the moon.

仅在太空方面——中国在2010年之前一直落后于印度——但今天——他们拥有自己的庞大陆地测绘系统,并处于成为全球第三个登陆火星的国家的位置——已经登陆了月球。

China
and Russia are on the verge of building their own International Space
Station having both the financial muscle and technology.

中国和俄罗斯即将建立自己的国际空间站,他们既有财力又有技术。

500

You cannot bully or intimidate or stifle Progress. Eventually Life finds a way.

你们不能靠欺凌、恐吓或扼杀进步。最终,生活找到了一条路。

US
should learn this lesson hard. The more they try - they may get 10
years more but in the end China will get there and take over.

美国应该好好吸取教训。他们尝试得越多——他们可能最晚在十年内达到同等的技术并接管整个市场。


It might be quite awhile, but there are many variables that’s in play.

这可能需要一段时间,但有很多变数在发挥作用。

It
should be noted that NO country at this point can produce the ASML EUV
equipment on its own (or even the older generation DUV is questionable),
ASML relies heavily on suppliers from Germany / UK / US / Japan etc as
well to make their equipment, Japan might be the closest country that
can produce a similar (but older generation) equipment relying only on
company / tech inside their border.

需要注意的是,目前没有哪个国家可以自己生产ASML的EUV设备(甚至老一代的DUV也值得怀疑),ASML严重依赖德国/英国/美国/日本等国的供应商来生产他们的设备,日本可能是最接近能够生产类似(但老一代)设备的国家,只依赖本国的公司/技术。

As
of right now the ASML not selling highest-end EUV to China isn’t even
really a thing, because they can’t produce enough of them anyway, and
the whole project managed to get off the ground because Intel / TSMC /
Samsung funded the project SPECIFICALLY so that ASML only really sells
to them that machine. it’s not just SMIC of China that can’t buy it,
even global foundery of the world can’t either.

就目前而言,ASML不向中国销售最高规格的EUV还算不上什么,因为他们无论如何都无法生产足够多的EUV,而且整个项目成功地启动了,因为英特尔/台积电/三星专门资助了这个项目,所以ASML只向他们出售那台机器。不只是中国的中芯国际买不到它,就连全球铸造厂也买不到。

500

There
are many layers of problems to China building it’s own extreme
ultraviolet lithography machines, but the most difficult is the fact
that EUV and the way TSMC & Samsung uses them are the extreme edge
of precision engineering and manufacturing. something China has never
been particularly great at, for example, they still struggle so far to
build their own internal combustion engines to the same level as Japan /
Germany (the two key powerhouse players in doing such things.)

中国制造自己的极紫外光刻机存在许多层面的问题,但最困难的是EUV以及台积电和三星使用它们的方式是精密工程和制造的极端优势。中国在这方面从来都不是特别擅长,例如,到目前为止,他们仍然在努力制造与日本/德国(这两个关键的大国在做这类事情)相同的内燃机。

There
are certainly potential ways for China to reengineer the process and
come up with something that has a relatively similar end result but
relies on different technologies, however that would require them to
completely reinvent the wheel of something that was basically the
collaboration of the whole world over decades on their own. Again,
that’s not IMPOSSIBLE, but it’s certainly extremely difficult and
inefficient.

中国当然有可能重新设计这一过程,并提出一些具有相对相似的最终结果,但依赖不同技术的东西,但这将需要他们完全重新发明轮子的东西,这基本上是整个世界在几十年的合作,靠自己。同样,这不是不可能的,但这肯定是极其困难和低效的。

Not
to mention that this is not a purely scientific / engineering project,
the difference with TSMC winning over other competitors is primarily
COST. i.e that it’s not that Intel CAN’T do 7nm, they can, it’s that
their failure rate is much higher than TSMC that if say, Apple relies
only on Intel for their 7nm chip, they’ll have to sell their iphone at a
much higher price and/or one of Intel or Apple (or both) go bankrupt.

且不说这不是一个纯粹的科学/工程项目,台积电与其他竞争对手的区别主要在于成本。也就是说,不是英特尔不能做7nm,而是他们可以,而是他们的失败率比台积电高得多,如果苹果只依赖英特尔做他们的7nm芯片,他们将不得不以更高的价格出售他们的iphone和/或英特尔或苹果(或两者)之一破产。

So
even if China COULD reverse engineer or re-engineer the EUV machine or
subsititute, if it doesn’t actually win out on a cost basis it’s
basically useless commercially, an insanely expensive academic project
(we’re talking about NASA level funding.) most importantly, the US could
simply shut out said machine using standards, and for local chinese
companies like SMIC to adapt a different system would be extremely
problematic.

因此,即使中国可以对EUV机器进行逆向工程或重新设计或替代,如果它在成本上没有真正赢得胜利,它在商业上基本上是无用的,一个疯狂昂贵的学术项目(我们说的是NASA水平的资金。)最重要的是,美国可以简单地排除使用标准的机器,而对中国本土公司,如中芯国际适应一个不同的系统将是非常有问题的。

Is
it POSSIBLE if China just like nationalize SMIC and stuff? yeah, but it
would in the grander scheme of things, be pretty close to cutting off
your own nose to spit your enemy. China would effectively be abandoning
the world market for decades in such a scenario. and probably move
slower than their competitors.

如果中国把中芯国际国有化,可能吗?没错,但从更大的角度来说,这就像是割下自己的鼻子来唾弃你的敌人。在这种情况下,中国实际上是在几十年内放弃世界市场。而且可能比他们的竞争对手行动更慢。

What
China SHOULD do, is what they have been doing most of the time (some
major exceptions of terrible mistakes and frauds non-withstanding.)
which is to try and gradually move up the chain and establish more and
more companies in this field and hope that some of them eventually hold a
major niche that becomes almost indispensable in this field. I’d say
that most of the actual workers / players in this field in China
understands this and is more or less trying to do that.

中国应该做的,是他们在大多数时候一直在做的(尽管存在一些严重的错误和欺诈),那就是尝试并逐步向上移动链条,在这个领域建立越来越多的公司,并希望其中一些公司最终在这个领域占据一个几乎不可或缺的主要利基。我想说的是,中国这个领域的大多数实际工作者/参与者都明白这一点,并或多或少地试图这样做。

But just to remind people, semi conductor’s main raw material input is Silicon

但提醒大家一下,半导体的主要原材料是硅。

So just to give an idea what shutting out China REALLY looks like.

所以,让我们看看把中国拒之门外到底是什么样子。

Now
that’s the part on why it’s a major problem for China, the other side
of the coin is that the US is increasingly playing a dangerous game as
well and could blow up its own advantages or worse.

这就是为什么它是中国的一个主要问题,硬币的另一面是,美国也在越来越多地玩一个危险的游戏,可能会摧毁自己的优势,甚至更糟。

If
you’re ASML and the EU in general, you’re not exactly thrilled to
basically have the US policy forced upon you as well. and would probably
now actively think about having replacement tech for the US/UK stuff if
possible so that you have at least political and economic independence
(i.e even if they don’t sell to China, it’s on their terms and not
because the US commanded them to do so.) and once that happens the next
potential moves by US and others could become quite scary.

如果你是ASML和欧盟,你不会因为美国的政策强加给你而感到兴奋。如果可能的话,现在可能会积极考虑拥有替代美国/英国产品的技术,这样你至少有政治和经济独立(也就是说,即使他们不卖给中国,这是根据他们的条件,而不是因为美国命令他们这么做),一旦发生这种情况,美国和其他国家的下一个潜在举动可能会变得非常可怕。

It
should be generally noted that the EU and other US allies is actually
paying almost all the price for the trade war/ geopolitical struggle
with China / Russia while the US is benefitting at their expense.

需要指出的是,欧盟和美国的其他盟友实际上为与中国/俄罗斯的贸易付出了几乎所有的代价,而美国却从他们的损失中受益。

(
One easy prediction, there will be a major famine crisis in Africa /
Middle East next few years, and will result in another refugee crisis
that might make the last one look mundane. )

(一个简单的预测是,未来几年在非洲/中东将会有一场严重的饥荒危机,并将导致另一场难民危机,这可能会让上一场危机显得平淡无奇。)

So
far if one look at the score board it’s pretty much the US wins and
everyone else (friend or foe) loses with a few small exceptions.

到目前为止,如果你看一下记分牌,基本上美国赢嘛了,其他所有人(朋友或敌人)都输了,只有一些小的例外。

500

This
is clearly unsustainable over the long term, and it’s not like this is
just happening as a new thing, the general trend has been on-going since
at least the early 2000s and not only have the likes of China and
Russia been making obvious contingency plans, one would be naive to
think that the EU is just a pure puppet with no independent thought or
interest apart from the US.

从长期来看,这显然是不可持续的,这并不是一件新鲜事,至少从21世纪初开始,这种大趋势就一直在持续,不仅中国和俄罗斯等国一直在制定明显的应急计划,人们会天真地认为欧盟只是一个纯粹的傀儡,除了美国之外没有独立的思想或利益。

来源:世界先进制造技术论坛

全部专栏