日志
黄汉森低估中国能力者成自己掘墓人美国已丧失竞争力
|
黄汉森低估中国能力者成自己掘墓人美国已丧失竞争力
美国对抗全球竞争者: 芯片大战的赢家是不是美国? | 华尔街现场
https://www.youtube.com/watch?v=NMKP3DgRoes&ab_channel=
台积电首席科学家语惊四座： 低估中国创新能力者，迟早成自己掘墓人！美国在这一方面还有优势，其余方面已丧失竞争力
https://www.youtube.com/watch?v=6-JUyh0FJdQ&t=68s&ab_channel=
斯坦福大学教授，前台积电首席科学家黄汉森 Philip Wong,自2004年進入史丹佛大學擔任電機工程系教授。1988至2004年間，曾在IBM TJ Watson研究中心工作，專責產品生產技術的前期研究。2018至2020年間，在全球最大的半導體公司台積電(TSMC)出任副總經理，負責技術研究。 自2020年起成為該公司的首席科學家，負責諮詢顧問工作。
黄汉森在接受美国媒体采访中表示，中国在芯片领域研究论文、学术报告及交流发言的水平已经从质量上，数量上都超过了美国，美国的优势在0到1的创意和想法，但一旦创意出来后，中国会迅速的做出来并且做得比美国更好，这是人才和政府支持的优势决定的，无可匹敌。我想要说这只是需要一个过程，中国人并不是没有创意，只是从掌握使用，到低级模仿，到改良优化，到工艺创新，到原始创新需要一个过程。人的创意不是凭空产生的，人无法想象自己不知道的事物，科技进步，人才水平不断向顶尖科技攀升后，原始创新的概率就会不断增加，最终在海量规模优势下胎熟落地
链接：https://xueqiu.com/3963746782/291350710
主持人是印尼前贸易部长Gita Wirjawan
台积电首席科学家 黄汉森
主持人
你是怎么长大的？你是在香港长大的？你是如何对现在从事的领域产生兴趣的？
好吧，我是在香港长大的。香港曾经是英国的殖民地，当时香港的大多数人都比较穷，香港是一个发展中国家，我的意思是，香港不是一个国家，而是一个殖民地，是一个发展中社会，当时大多数人的教育水平都不高。在我父母那一代，他们都因为墙和二战而受教育中断，许多人没有正式上学，我的父母都没有读完高中，但他们和许多亚洲父母一样，非常坚持认为所有的孩子都应该接受良好的教育，这就是他们所强调的，所以我经历了非常草根的教育。我们家住在非常拥挤的地区，我很小的时候，我们就从政府那里中了彩票，因为当时政府正在建造许多政府住房部门的高层公寓，这些公寓以非常低的价格出租给居民，所以我们就像中了彩票一样搬到了其中一栋，这真的改变了我们的人生轨迹，我们住在那里，我在那里的住房项目中长大，后来，随着我们家庭获得更多资源，我们搬到了私人住宅，但早年我们在一个住房项目中长大，这让我们真正融入了社会大众，这让我对社会有了很好的了解，在你的父母中，谁更重视教育，是你爸爸还是妈妈，他们两个，因为他们没有机会接受正规教育，所以他们意识到，每个人都需要谋生，所以他们没有太多机会接受正规教育，所以他们非常坚持，家庭中的每个人都应该受过良好的教育，你是怎么发现电气工程的，我的意思是你可以做其他事情，对吧，我可以做其他事情，嗯，我想那是当时的社会，呃，倾向于 STEM 类型的教育，因为这似乎是一张通往更好生活、更好职业的门票，而人文和艺术的定义不太明确，例如，如果你获得了成为工程师的执照，那么当然你就获得了执业资格，你有了职业等等，而其他职业似乎更加模糊，定义明确，所以当时很多在学校表现出色的孩子倾向于选择 STEM 领域之一，嗯，所以我也不例外，跟随人群，呃，进入一种类似科学类型的课程在科学类课程中，物理和化学部分
科学课程更加定量和正式，更容易量化，至少对我来说更容易理解，而不是例如生物学，对我来说，生物学更像是一种观察性科学，你需要记住很多东西，嗯，我不太擅长记忆东西，所以你很谦虚，所以物理和化学看起来更好，嗯，当我即将进入大学时，学习基础科学的人的职业生涯相当有限，因为它不像美国那样是基础科学研究，对吧，B 在香港，成为一名基础科学研究员并不是一项职业，所以大多数科学类职业都是战术工程，其中之一，所以我们许多同时代的人都成为了工程师，嗯，在所有工程中，电气工程最接近物理学，嗯，它使用许多相同的
物理学概念，你知道量子力学和电磁学等等
它们更接近物理学，所以这就是我被它吸引的原因，是的，你知道，我认为
可以肯定的是，每个国家或民族或政府或政治文化
如果非常重视 STEM，他们往往
会做得很好，而那些可能不太重视 STEM 的国家则相反，你会把这归因于你家里的政治文化，或者
有点像任何层面、任何层面的普遍政治文化，或者
你认为这可能是由该国、民族或政府的领导层推动的，我认为如果你谈论
我认为如果你看看许多亚洲国家，他们都非常重视 STEM，是的，
我的粗略理解是，从经济发展的角度来看，你确实需要这种技能，从经济发展的角度来看，如果你想建设基础设施，你需要工程师来建设桥梁和道路等等。如果你想拥有
工业，你需要能够从事工程机械工程、电气工程等的人。我认为这主要是从
经济发展的角度来看的。但我应该说，
经济发展本身只能让你走这么远。在某个时候，社会将需要经济发展以外的东西。这就是人文学科和艺术等学科的用武之地。
社会中的人们不仅仅满足于金钱，对吧？但除此之外，还有更多的东西，而不仅仅是谋生。一旦你对谋生感到满意，你就会寻找其他东西。我认为这是许多
亚洲国家可能没有意识到的部分。我只是感觉到
仍然有很多国家没有像我们可能看到的一些国家那样关注 STEM。为什么，为什么，所以不确定我的意思是呃，我
认为，我认为对 STEM 的更高程度的关注与你的能力相关。移动
边际产品，因为它与创新和所有那些好东西相关，对吧，并且能够
将其与文科或人文学科相结合，我认为这使它
成为一个更好的组合，但我认为有些国家往往不太关注 STEM，我认为这在一定程度上阻碍了发展，
而且我很高兴香港和其他一些亚洲国家，如韩国、中国、日本，
最近你知道你谈到了越南和东南亚，对吧，你可以看到越南人民对 STEM 的执着，我来自一个叫东南亚的地区，我认为未来需要更加强调 STEM，你去了香港大学，然后你决定搬到
美国，是的，为什么很简单，嗯，我对更高级的知识感兴趣，超越了学士学位类型的教育，我正在寻找一个可以
获得博士学位的地方，当时是
80 年代初期，香港没有一所大学可以称为
研究型大学，它们基本上是教学型大学，只是在殖民时期
香港的大学，负责为政府服务的公务员培训
官僚，他们对做研究不感兴趣，所以在香港真的没有
机会做任何研究，所以我来到了美国，到目前为止，我和你相处得很好，我
我没事，我想现在我想换个方向
到你的重点领域，我的意思是，我知道你在教电气工程，你是少数几个了解半导体的人之一，你知道为什么它被称为半
导体，而不是全导体，这很有趣，但在我讲到这个之前，让我先说完其他呃
你之前的问题，我认为我从香港到美国的旅程对于许多从事技术领域的移民来说是一个非常典型的旅程，他们在自己的祖国接受教育，来到美国，看到了接受更好教育的机会，然后很多人留下来，我认为这是美国非常重要的人才来源，呃，也许我们可以稍后再谈这个问题，但我认为这真的很重要，如果你看看学术界和工业界，他们中的许多人都是通过这条途径来到这里，获得高级学位，并留下来，这绝对是美国创新的主要来源，如果你看看大公司的高层管理人员首席执行官，他们中的许多人都是通过这条路线来的，是的，这非常重要，嗯，我的意思是，你是否认为这种情况可能会持续下去，因为我认为这处于危险之中，好吧，考虑到 Reon，政治算计似乎是是的，这似乎是钟摆
摆到了另一边，近年来，你会看到很多人
对移民或外国学生以及
研究的开放性等问题进行深刻反思，我认为这创造了一种环境，当一个来自其他国家的潜在学生看到美国时，他们会开始思考这是否是适合他们的地方，或者他们会去其他国家接受教育和获得经验，是的，当我刚从大学毕业时，这个答案非常明显，你会去美国哪里，当然是的，但现在你有了选择，我们现在有了选择，而且不仅仅是选择，从拉动的角度来看，从推动的角度来看，对的，很多人觉得美国可能不像以前那么受欢迎了，所以我认为眉毛或人流
在某种程度上是的，在另一个问题上为什么叫半导体哦为什么叫半导体我实际上我忘了我是听谁说的当我有人谈论半导体时有人不在这个领域问为什么半导体为什么全导体不是更好的杯子半满嗯呃什么是半导体a半导体是一种材料，顾名思义它不太导电不太绝缘并且介于两者之间为什么我想要介于两者之间的某种东西是因为全导体会像一块金属一样铜或银或那种东西对吧所以对于全导体现在不是食物而是真正的导体基本上你为什么半导体sem博士呃用于构建开关呃像电开关墙壁开关翻转开关打开灯和计算机实际上是一堆开关它们我们通过设计组织安排开关如何布线开关嗯在早期电话呃你有一个总机并且你有真正的人将一条线与另一条线连接呃呃那个在总机中，而所有这些都被今天的芯片取代了，所以计算机通过翻转开关进行计算，所以半导体可以打开以导电或关闭以不导电，这基本上就是开关的作用，所以半导体用于制造开关，这就是为什么一些导体是有用的，因为它可以打开或关闭。戈登·摩尔几十年前就提出了这个预测，你知道事情会以某种方式表现，通过容量增长，速度增长，你知道它非常非常准确，是的，每隔一年半到两年，容量和速度就会翻一番，摩尔定律会发生什么，好的，首先，你提到的摩尔定律，呃，加倍的东西，嗯，最初他只是建议将芯片上的晶体管数量加倍，呃，后来，人们对其进行了修改，将任何东西都加倍，这不是他真正的意思，但人们往往会联想到它有了 MOA，嗯，但嗯，那基本上是一个自我实现的预言
嗯，这实际上是一种技术或行业相当不寻常的情况，嗯，我相信我的一位同事 Jim prer 教授曾经说过，他不知道有任何其他行业具有这种自我实现的特性，也就是说，每个人都认为在 X 年内这样的事情会发生，对吧，因为
公司在竞争，因为有一个预测，公司试图要么满足要么超越这种保护，这样你就可以比你的竞争对手更好
因此，由于竞争激烈，达到了那个特定的目标，这个目标得到了满足，因为，而且不仅仅是
竞争的愿望，而且有一个目标也允许每个人上下
整个事件链，从材料到设备技术到电路设计和架构等等，在链条上上下下，有一个可预测的节奏，你说，假设我在设计我需要我设计的东西，我们可以在某种芯片上运行。现在，除非我知道芯片技术可以做什么，它的能力是什么，否则我无法开始设计芯片。这就像如果你试图跑马拉松，除非我知道我的能力是什么，否则我无法赢得马拉松。所以你需要知道我的能力是什么，然后才能设计你的芯片。现在，如果没有这个预测，你就不能说我将基于这个假设来设计这个芯片，芯片技术将为我提供这种性能。我不能那样做，但有了这种自我实现的预言，我知道两年后芯片技术将发展到我可以依靠这种特定性能的地步，因此我将基于该假设来设计我的芯片。这非常强大，非常强大，它在某种程度上框定了行业的行为。绝对正确，绝对一致。在这种情况下，我甚至给出了设计芯片的例子。甚至材料供应商，材料供应商也知道芯片制造商将需要这种材料或需要这种材料两年内达到这种性能需要的材料，因此我
在开发过程中一直使用手机，当我完成开发时，我会有一个
客户，风险降低非常重要，如果我开发某样东西，我花了
数百万美元，是的，开发某样东西，我不知道是否有一个不好的客户
是的，但在这种情况下，我知道你知道在 RIS 中引用你的话
呃在之前的对话中，所以以芯片占据我们生活（如果不是生活方式）的百分比的速度增长，无论是打开
车窗，打开烤面包机，打开微波炉
无论如何，是的，在你谈到更复杂的东西之前，莫尔定律会发生什么，它有可能吗它是否可能比我们在过去几十年中看到的任何情况都更具指数性？我认为它有两个维度，嗯，当然，答案是肯定的，它将更具指数性，是的，好吧，为什么我要这么说，呃，首先，在过去的几十年里，我们知道有一种方法可以提高芯片的性能，那就是让它变得更小，这涉及到很多物理原理，要解释为什么让它变得更小更好，对吧，所以我们知道这是一种方法，我们已经转动这个旋钮几十年了，对吧，这个旋钮有点饱和了，就像其他任何东西一样，你踩下油门，油门踩到底，一开始你加速，但很快你就踩到了底，对吧，这是肯定的，但这并不意味着进步不会继续，因为还有许多其他方法可以取得进步，我喜欢的比喻是，在过去的 50 年里，我们一直在里面行走一条隧道，我们只有一条路可以走，我们知道目标是什么，我们只想取得进展，取得进展就是制造设备，使晶体管和内存更智能，这将使我们能够在同一芯片上安装更多设备，从而让人们获得更多功能，更好的性能，因此，这种方法是我们知道的一种方法，而且已经饱和，但它更像是在隧道里行走，没有路可走，这是前进的唯一途径，但现在我们在隧道的出口，现在隧道的出口有两种解释方式，一种是你从悬崖上掉下来，你就死了，第二种是世界是敞开的，不再受到限制，你不能左转或右转，你可以去任何地方，我想我的解释是我们隧道的出口，有很多可能性，我们的能力将继续以指数级增长，我们现在的速度更快，为什么我对此如此乐观，是因为必要性是发明的模型，而我的乐观情绪由此产生，乐观情绪源于这一事实社会对这些热电联产技术的需求越来越大，社会对半导体技术和热电联产技术的需求将比过去几十年更多，因为需求创新将会出现，并以更快的速度推动技术发展，因为我们不再受隧道的限制，可能出什么问题，可能出什么问题，地缘政治出问题，我认为这是一个非常重要的方面，可能几年前还不存在，我认为这是一个重要的部分，因为研究和开发依靠合作而蓬勃发展，有很多例子，我们可以稍后再深入讨论，但它依靠合作而蓬勃发展，如果地缘政治导致合作无法发生，那么进展速度将受到严重限制，我认为这可能会出问题，你知道，我们正在经历这个我称之为多极化的时代，我之前说过，具有讽刺意味的是，在多极世界中，多边化成为更难，因为新兴中等强国或中等以上强国有修正主义倾向，而现有的超级大国不太愿意接受这种新的多极模式。我想把这个问题放在你所隶属的台积电的背景下，他们基本上创造了一个成功的秘诀，这涉及很多合作，从供应链的角度，从研发的角度，从你们有什么，不仅在企业之间，而且在学术界和政策之间，以及企业家精神之间，是什么让他们变成现在的样子，我认为你的观察非常敏锐，很多人认为台积电这样的公司是工厂和工具的集合，甚至只是工厂工人，是的，就是这样，但实际上并非如此，它是一家企业，涉及大量合作，这些合作在客户网络、供应商网络和供应商网络上蓬勃发展研发网络的合作者与他们所生活的更广泛的社会的合作，对吧，台湾社会，这是一个所有这些的网络，甚至更多，也许我没有深入研究，所以那部分，那关系的集合，就是这家公司现在的样子，呃，假设我们突然把这家公司移植到另一个国家或另一个没有相同政治经济体系的地方，他们是否继续以现在的方式运营是值得高度怀疑的，是的，那么我们如何让其他国家尝试复制台积电可能做到的任何一点，而且我不是单单指台积电，台积电还有其他传奇和伟大的公司，对吧，三星是世界，英特尔是世界，但我认为台积电已经证明了自己非常有弹性，非常不同呃，他们的成功似乎到目前为止都是可持续的，是的，回到历史，回到传奇人物张忠谋和台湾一些人之间的对话，他们让台湾变成了现在的样子，我认为有两个甚至更多非常重要的因素，一个是技术竞争，实际上是人才竞争，首先是人才，其次是社会对行业的广泛支持，不仅是公司，还有整个行业本身，为公司和供应商、客户等提供基础设施，所以让我们先谈谈人才，对吧，台湾在 60 年代、70 年代到 80 年代和 90 年代早期都派遣了人才，不是政府派遣的，而是那个时代的学生，他们大学毕业后都去了美国、欧洲，除了台湾，还有受过高等教育的人，在大学里取得了成就，接受教育，他们中的许多人留在了当地的东道国获得工作经验，随着时间的推移，许多人回到台湾继续从事这个职业，原因有很多，他们想离家更近，我想这也许是最大的吸引力之一，离家更近，或者在祖国有更好的职业机会，就像在家乡打球一样，所以人们有时会觉得更有优势，了解文化和人，所以人才流动是最重要的事情之一，如果你看看今天台湾科技行业的领导者，他们中的许多人都在美国接受教育，通过这种方式在大学接受教育，去其他国家，美国、欧洲等，然后回到台湾获得工作经验，回到台湾，把工作经验带回来，把知识带回来，这是非常重要的，我想说，这可能是成功的主要因素之一，它不仅限于台湾的职业生涯，是的，我们看看三星和海克斯等公司的许多高层管理人员，他们中的许多人都是在美国接受教育，嗯，这是其中一部分，但这只是一部分，而另一部分也很重要，那就是社会的支持，如果你想想台湾半导体产业是如何起步的，它确实得到了政府的支持，这是产业政策，基本上，台湾政府决定这是他们想要进入的产业，他们支持建设工业园区，启动工业技术研究院，该研究院剥离了 SMC 和许多其他公司，所以政府现在确实有一个非常强大的产业政策，当然，产业政策可能不会成功，也可能不会成功，如果你押注错误的事情，那就不好了，对吧，当然，如果你在美国，你知道美国觉得他们不需要产业政策，对吧，我们不谈这个，但在这种情况下，产业政策对他们很有效，政府真的支持它建设工业园区，找到其他拼图来填补台湾是 S
产业的基础设施，也是科技产业，在 80 年代、90 年代和 21 世纪初期，台湾基本上是制造业，但基本上是 PC 和自行车等所有产品的全球制造商，是的，所以政府真的通过 IND 和非常强大的产业政策支持它，这真是太棒了，现在你知道，在早期，许多与微芯片技术相关的知识产权都是在美国完成的，你认为事后看来，如果美国做出其他决定，他们会希望将制造能力转移到国内，而不是将制造能力转移到海外，情况会有所不同吗？这是一个非常复杂的问题，我最近一直在思考这个问题，但我还没有找到任何合理的解决方案或答案，我可以告诉你，呃，是什么样的脱节或问题，一个是在美国，另一个是我说过这是一场技术竞争，是人才竞争，是的，如果你看看亚洲国家和美国的人才成本，成本差异相当大，是的，所以嗯，现在这个差距越来越小了，但当时差距很大，是的，是的，是的，所以让教育工作者在台湾、新加坡或韩国从事心脏工程问题，成本比美国低得多，换句话说，他们可以部署许多高技能人才来解决工程问题，而美国在成本方面处于劣势，不是资本成本，而是人才成本，举个例子，嗯，我最近看了台积电 20 多年的年度报告，每年他们都会报告员工的教育水平，从 1999 年开始，大约 40% 的员工拥有学士及以上学位今天
80% 哇，现在
恰好与技术的稳步攀升相吻合，呃，领导权竞争
从一开始就稳步攀升，他们从 RCA 获得许可，从美国获得许可技术，
然后逐渐将其带回台湾，他们自己学习如何做事，而做到这一点的方法是获得人才，所以他们从 40% 的员工开始，拥有学士学位及以上学位，到今天
80%，所以这真的告诉你技术竞争是人才竞争，你好，在这种情况下，你认为哪些其他国家做得和台湾一样多或差不多，我的意思是，当你想到人才时，你会想到人力资本，你会想到那些在亚洲或世界范围内表现出正确能力来投射智力资本的国家，你认为哪些国家
可能能够效仿我们在台湾看到的那种成功，我见过像新加坡这样的国家，韩国取得了巨大的成功，对吧，韩国我
记得在 60 年代和 70 年代，韩国根本就不存在，然后
后来他们有了一种跟我一样的产品，我们把这些产品与
MEOS 联系起来，是的，开始创新和独特，然后他们慢慢地爬上
学习曲线，我记得几年前你看到三星说这到底是什么，你知道现在不可靠
这是世界上最好的东西，所以这些国家做得很好，那中国呢，我认为我们不能忽视
能力，他们让我谈谈我熟悉的东西，只是研究，对吧，学术界，我知道几年前我们参加
技术会议，我们看到中国的论文说啊算了吧
质量太差了，甚至没有竞争力，可能
在 80 年代和 90 年代，现在他们比我们好，他们比我们好，如果你看我看论文，出版物数据来自芯片领域的关键会议，例如国际电子设备会议，这是一个设备技术会议，也是我们领域最大的会议，国际电路和系统会议是我们领域的两个最大的会议，你基本上翻了个身，几年前美国拥有大多数论文，我记得大约有 40 60 40 40 到 50% 的论文来自美国和中国，也许 20 30 年前，它们今天无处可寻，中国和亚洲的论文超过 40%，几乎接近 50%，哦天哪，美国从 40 50% 稳步下降到 30 到 40%，世界其他国家，主要是欧洲和日本，基本上已经跌落悬崖，因此亚洲国家的研究和开发能力，中国、韩国、台湾、新加坡等，真的成为最强大的地区，就嗯，进行高质量的研究，我指的不仅仅是数量，还有质量，嗯，我唯一看到的是，美国在提出新想法方面仍然领先一点，中国人总是说从零到一，也就是从无到有，然后提出这个真正新的想法，如果我看看我所说的新想法，美国以前从未讨论过，如果你读过研究论文，你会发现美国仍然是这些新想法的主要来源，但是一旦这些新想法为人所知，你就会看到这些新想法在其他地方如雨后春笋般涌现，嗯，我在日常研究中感受到了这一点，我的学生，天啊，我们提出的任何新想法，我们提出的新想法，一旦它们为人所知，这是一个好主意，下周它将出现在中国，下周出现在中国，不仅如此，他们做得比你更好，是的，我不能再竞争了，他们有更好的资源，他们有更多的学生，他们有更多的政府资金，我无法竞争我必须离开这个领域，我想回到这个话题，但是呃
我想问你关于两个国家的问题，比如印度和以色列，我的意思是他们
拥有巨大的人力资本，你认为他们是
竞争者吗，嗯，能够复制这些成功具有
巨大的潜力，他们的人力资本
令人难以置信，嗯，他们不仅是人口众多的国家，当然他们也有
规模，是的，他们有规模，统计数据对他们来说是正确的，当你有规模时，就会有非常聪明和
非常创新的异常值，呃，等等，就像中国一样，所以他们有
这非常重要，嗯，我认为如果他们做对了事情，他们就有巨大的潜力，因为他们有规模，人们很聪明，就以色列而言，他们没有规模，一个小国
但他们是聪明的人，显然他们面临的困难是缺乏规模，有点像新加坡是的，有点像新加坡，对吧，新加坡有很多聪明人，550 万人，是的，那是有限的，是的，有限的，他们没有规模，回到中国，你知道，随着人工智能的出现，你认为新创意的产生速度会受到什么影响，我认为新创意的速度会让人工智能创造出一篇论文，对吧，是的，但我认为创意仍然来自人们，嗯，这是从创意到有用的东西的加速，人工智能将能够帮助，嗯，创意仍然来自人们，但今天有很多障碍，需要时间和努力才能将一些创意带到人们的面前，也许人工智能的出现会缩短这个生命周期，但创意仍然来自人们，我相信现在，嗯，人工智能真的会帮助美国和更先进的国家重新回到制造业的游戏中，让我解释一下为什么我之前提到，人力资本的成本是你不能忽视的，差异是因为生活水平等而存在的，现在有了人工智能和机器人，我们应该能够在先进制造业方面取得很大进展
因为在先进制造业中，你需要很多高技能人才，因为这些制造业非常复杂，不像按照食谱一遍又一遍地重复做事情，你需要能够理解这个过程来改进它，而这些需要高技能人才，受过高等教育的人，现在发达国家的人力资本成本非常高，在这种情况下，你想要真正好好利用这些人，他们的生产力必须高得多，人工智能机器人的出现将使发达国家能够比以前更有效地部署这些高技能人才，因此使用人工智能机器人将使我们能够在未来10到20年内公平竞争，关于高技能劳动力的成本，你如何看待竞争格局，就国家而言，我的意思是，我认为中国肯定会在这方面投入更多的资金，美国似乎想要在这方面投入更多的资金，不仅仅是在知识产权领域，而且在制造业方面，韩国似乎非常坚持不懈对吧，你怎么看待这个格局？我的意思是，这有点像斯普特尼克号，是的，你知道从芯片的角度来看，不仅是芯片，还有许多技术，还有许多其他技术，你可以想到，比如电动汽车电池等等，它们都喜欢这样，所以世界正在变得有点制造业或技术的世界变得有点两极分化，所以你有点分成两个凸轮或两个半凸轮之类的东西，人们需要一些阵营，一些国家，我们需要两边玩，找到自己的出路，但说话时，我们可以把它看作两个阵营，现在既然如此，正如我之前提到的，为了技术进步，你真的需要开放科学，你需要合作，所以现在他们是两个阵营，他们不互相交谈，这意味着事情会进展得更慢，这意味着社会会受益更少，因为事情会进展得更慢，第二，供应链成本的分离将上升，你需要找到你的供应商，不是最低的出价，而是考虑地缘政治局势，所以成本现在会上涨，真正的问题是谁来承担这些成本，目前还不清楚谁来承担这些成本，今天，如果你看看为什么美国制造业今天的情况如此糟糕，部分原因是，我相信我不是经济学家，我只是猜测我是一名工程师，对吧，所以也许一个听这个播客的经济学家会说，等一下，这家伙在说什么，对吧，但我的猜测是，利润的分配在整个链条上并不公平，那些在食物链顶端运营的公司，他们攫取了大部分利润，我给你举个例子，前几天我刚去过西部数据，他们告诉我，如果你想买更大容量的 iPhone，他们会多收你 100 美元或 300 美元，但制造芯片的人只能得到几美元，所以你我们正在谈论公平竞争环境，那里没有太多的公平，那里有很多公平，所以这种情况需要纠正，否则美国的制造业就不会顺利了，我记得例如我们谈论人才，我曾经和一群行业公司进行过一次小组讨论，公司里的一个人哀叹道，哦，我们雇不到人，没人想加入我的公司，我们的职位随时都会失败，等等，缺乏人才，缺乏劳动力，所以我问他你付给他们多少钱，他沉默了，他无法回答我的问题，为什么，因为他们付不起钱让他们与其他人竞争，而其他人拥有更好的人，谁能获得利润，因为他们付不起钱无法获得足够的利润，所以他们无法支付工资，所以他们无法隐藏人才
所以人才的稀缺是真正的结构性问题，这是结构性问题
事实并非如此，很多人认为，哦，我们只需要告诉大学提供更多
课程，不，不会发生，不会发生，你需要有需求
我可以开设 10 门课程，但如果没有人选我的课程，这对你没有任何帮助
你需要有学生的需求，你从哪里得到学生的需求，学生需要
知道他们面前有光明的职业生涯，是的，如果他们选修这些课程，如果他们进入这个领域，
他们将拥有一份好工作，而今天许多制造业
相关的工作并非如此，所以你是说，如果我理解正确的话，竞争可能会减速，是的，通过两极分化，如果我们没有管理好它
对，希望我们能更好地管理它，对吧嗯，还有人才的稀缺，我认为
这适用于双方，是的 SC，是的，人才的稀缺，我认为
太平洋这边也有，太平洋那边也有，当然，考虑到规模，当然，这与利润如何公平分配有关，因为
制造和供应链下方所有东西的公司都没有获得足够的利润来提供良好的薪酬，为员工提供良好的工作环境，即使在供应链中，也有一些参与者
实际上比其他参与者赚得更多，哦，是的，好吧，哦，是的，这是一个结构性障碍
是的，这是你想要的速度，我认为
政策制定者和《经济学人》应该重新认识到这一点，否则它就是结构性的，在美国境内的愿望，你知道有一些对话讨论以及台积电在美国各地（包括亚利桑那州）建厂的计划，你认为进展顺利吗？还是像预期的那样顺利？或者我之前在 12 月的大型会议上告诉过你，现场演讲者会议上，我看到了来自世界各地的同事，来自欧洲和日本，所以我的日本朋友都对我们很着迷，你们是最慢的，好吧，我想这是众所周知的，哇，现在谈谈逻辑芯片、内存芯片和特殊芯片，我们称之为离散芯片或其他什么，这将如何公平地继续发展，就谁将在这三个类别中比其他类别更占主导地位而言，我认为这三个类别都很重要，是的，并不是说一个比另一个更重要，更重要的是，例如我们正在做这个播客，嗯，你需要电脑来处理数据，然后把东西流出来，把它讲出来，故事也讲出来，故事要存储起来，麦克风视频
摄像机他们都制造芯片那些是专业芯片是的所以它们很重要所以我们不能说哦我们可以没有其中一个不我们不能它们现在很重要就资源而言嗯今天截至今天逻辑芯片的开发需要大量的资源因为它们是推动我们这种行业合作伙伴的技术驱动因素要求推动新技术新设备新工艺新材料新设备新设计这些应用程序的发展推动呃呃人工智能它驱动大型计算机可以处理这些呃人工智能模型的数字它还推动计算芯片的发展非常节能你可以在他们称之为互联网边缘的地方做事情对吧你可以今天去买一个VR耳机你必须有一个电池组来搭配它这不是最终状态St of V耳机我很确定在最终分析中V不会那样运作但这有点像你需要经历一个中间阶段，就像卖手机一样，曾经是一个突破性的市场，我记得 90 年代初期，嗯，那些只能打电话，不能做其他事情，我给你讲个故事，我在 IBM 的时候，我曾经从事成像图像传感器的工作，在 90 年代中期，我有这个想法，我想提交发明披露，关于在手机上安装摄像头，这是在 90 年代，在 90 年代，IBM 的律师说谁想要一个带摄像头的手机，我们这次不会解雇，几个月后，我终于说服他解雇了，我们确实有一个 B，他们有很强的预测未来的能力，所以所有这些事情，嗯，我们希望芯片能够取得的所有进展，我们希望今天，芯片具备这些功能是由计算机计算驱动的，但这并不意味着这种情况会一直持续下去，也不意味着其他地方不需要创新，例如图像传感器，例如图像传感器的巨大创新，它们为 V 耳机提供了新功能，您需要在图像传感器方面有新的创新，它们接收信号的方式对其进行转换，然后进行处理，就在前面，而不是在中心端，也不在主机架计算机中，所以所有这些创新都需要发生，以满足我们社会对技术的总体需求，如果你看看联合国的 17 个可持续发展目标，我基本上会全部看完，然后说嘿，其中至少有 12 个 13 个需要计算技术和芯片技术的进步，所以如果没有这些，所有这些发展目标都将受到严重限制，所有这些的支持者都是设计师而制造商，是的，现在是高通的世界，是英伟达的世界，他们似乎满足于仅仅设计，你如何看待未来的发展或转变，他们是否可能，或者我们是否可能看到他们朝着制造业的方向发展，考虑到公司的情况，我们可能会在这种两极分化的背景下看到复杂性或潜在的复杂性，30 多年前，当张忠谋提出这种分裂时，即公司只设计，没有任何工厂来制造芯片，而代工厂专注于制造芯片，而不是设计任何东西，这就是分界线，这种分裂导致了设计活动的蓬勃发展，因为设计的人不需要拥有资本和拥有工厂的所有负担，所以他们可以专注于设计，同时拥有工厂的人可以专注于经营工厂，同时拥有无数的好处如果我制造芯片，我将芯片直接卖给客户，那么我只需要一位客户，那就是我自己，这对于制造商来说，规模经济是不够的，因此这种分裂在使设计理念比以前更快方面确实取得了成功，并且创造了许多以前不存在的商业机会，因为以前你必须拥有巨额资本投资才能设计自己的芯片，现在你不需要，你只需要设计，然后有人会制造它，现在这种分裂在过去 30 多年里进展得非常好，但今天，如果你看看 Fabulus 和 The Foundry，他们并没有真正分裂，他们实际上将苹果、高通、英伟达、AMD 等公司聚集在一起，他们每天都出现在 The Foundry，他们告诉 The Foundry，运行这个晶圆，尝试这个实验，给我看结果，给我晶圆，这样我就可以自己测试了，所以他们实际上在做芯片本身的大量开发工作，所以前沿技术基本上是我们今天共同开发的在代工厂和
领先用户之间，谁下了最大的订单，谁就对如何设计和优化技术有很大的发言权，但是对于游戏的皮肤，在制造能力方面，他们是否可能会看到更多的皮肤，似乎就是这样
已经有一些指标，例如索尼在日本投入资金，并与台积电合作在欧洲建立工厂，
汽车公司的集合，他们已经投入资金在那里建立工厂，所以你看到了其中的一些，以及它会走多远还不清楚，但你看到一些关于这家位于阿姆斯特丹或荷兰的公司 asml 的讨论，他们制造了一种非常奇特的技术，对吧，紫外线光刻
是的，这对于很多事情来说都是必要的，对吧，谈论
那是一项技术，如果你详细了解这台机器，euv 机器，极紫外，神话机器，
真的是一项工程奇迹，现在它也是一个声明两件事，一是
这是关于全球供应链的陈述
激光来自圣地亚哥
镜头是德国抛光的
机械运动来自
康涅狄格州
所以它不仅仅是delence
它不仅是机器零件本身的全球供应链
而且是知识
全球知识供应链
这是一个例子
它是开放科学的一个例子
第一篇欧盟论文发表于80年代末
由B组应用程序
然后是一系列会议
日本和欧洲的研究人员都参加了这些会议
然后美国国家实验室参与了电子部门早期原型的开发国家实验室等，然后公司开始进入英特尔、台积电等，开始投资，我们谈论的是全球公司，不只是一个国家，全球公司都在投资，几年后，30 多年后，也许 40 多年后，你就有了这台奇妙的机器，这一切都来自于全球研究人员的合作，他们创造知识，分享知识，自力更生，这是我们要学到的一个非常重要的教训，如果你把自己与世界其他地方隔离开来，你就无法实现这一目标，即使你能做到这一点，你也会花更长的时间，你会失去机会，这是一个非常重要的教训，在这个地缘政治紧张的时代，每个人都想把所有东西都留给自己，而不是彼此分享，如果你不分享，你就不会获得回报，你得到的回报比我们分享的更多，我认为这是基本上，我认为，我会如何描述它，你会通过多个国家不同利益相关者的交织来培养一种乐观的感觉，你似乎表明，这种结构是一种结构性的凝聚力，很难把它拆开，是的，是的，如果你真的把它拆开，你就会输，每个人都会输，是的，每个人都会输，但你知道，考虑到我们在世界许多地方都看到了如此多的政治神经官能症，有时人们不会理性地行动，是的，他们根据自己的心声而不是大脑的声音来行动，这不是风险，有风险，但我认为，在某些方面，我认为这是由于缺乏理解，而不是任何不良意图，这不是因为某人的不良意图或诸如此类的事情，但我认为这更像是缺乏理解，呃，我通过我参与的一项研究得出了这一结论大约两年前，胡佛研究所正在研究我们中国台湾和半导体，两年后他们发表了一份名为“硅三角”的报告，你可以去 tuber 网站下载，通过参与，我了解到我的观察是，政策圈中的许多人都不懂技术，是的，因此由于缺乏了解，他们会制定从技术角度来看毫无意义的政策，如果你确实有这项技术支撑，那么现在它就毫无意义了，这并不意味着制定政策的人意图不好或诸如此类，这只是缺乏了解，也不意味着制定政策的人很愚蠢或诸如此类，不是没有人了解世界上的一切，今天他们可能正在制定有关疫情的政策，明天他们可能正在制定 SEM 芯片的政策，下一个可能是核武器，所以任何人都不可能，任何一个人，甚至一群人，能够了解世界上的一切，但我认为如果有的话，政策制定真的会有所帮助有更多领域专家参与制定政策，而今天我认为我们只有很少的人了解至少我所熟悉的半导体领域，他们实质性地参与制定政策，我认为这是一个很大的漏洞，我们需要填补，你建议更高程度的文化互动吗？哦，是的，绝对的，我感觉人们没有花足够的时间在彼此的地方，是的，绝对的，就像我说过的，我参加了胡佛研究，他们中的许多人都是政策制定者，中国是专家，等等，在研究结束时，我们都对其他人所擅长的领域有了更多的了解，我以前对中国政策或产业政策一无所知，但现在我至少有一些入门级的了解，我的同事同样对半导体有一些入门级的了解，他们不会问为什么不再是一个完整的导体，我必须问你这个问题，我必须问你有什么区别你知道，石油和芯片之间，我知道有人问过你，石油和芯片之间有什么区别？石油和芯片，是的，没错，我的意思是人们一直在说芯片是一种新的石油，是的，是的，所以嗯，我参加了这个，参加了这个，嗯，胡佛研究所的研究，很多人说，哦，芯片，我们需要芯片，芯片无处不在，嗯，任何你需要打开电源的东西，里面都有电池，或者停车门，或者涉及某种芯片，你摇上车窗，那里有一个芯片，你听播客，你有一个芯片，一切都是芯片，所以它就像商品一样，就像石油一样，我们需要石油，我们需要化石燃料，对吧，所以人们把石油和芯片进行比较，这种比较是相当的在政策圈中很流行，但是现在它放错了地方，我认为它完全放错了地方，当我在呃研究中发表那条评论时，人们看着我说，哦，是的，好吧，为什么不同的芯片是 L，因为石油已经在地下埋了数百万年，它们从不改变，你今天把它拿走，你明天再拿走，是同一种石油，拿走的方法可能不同，但石油是一样的，芯片或没有，你有手机，你有手表，你有电视，你想用 20 年前制造的手机吗，你想看 20 年前制造的电视吗，你不想，因为芯片不断进步，它必须不断进步才能显示其价值，人们期望芯片在这种能力上不断进步，能力广义上定义为它们可以做的事情的数量，以及能源效率不要消耗太多的能量或功率不要消耗那么多电池那么多的电池电量等等，所以芯片对我们，因为
它对我们很有价值，因为它不断进步，不断自我更新
今天的芯片与明天的芯片不同，你想要明天的芯片，而不是昨天的芯片
石油不一样，你可以储存石油，毫无疑问，美国也是这样做的，许多
国家也这样做，你不会像史波克那样堆积芯片，你拥有的库存能力很快就没有用了，每隔几年它们就会过时，是的，每隔几年就会过时，是的，究竟是否可行，实现芯片主权或绝对芯片主权，考虑到我们刚才谈到的全球性质，我只是在想象，不可能，不可能，我认为，无论是谁在政策圈提出这一点，可能对事情的运作方式没有足够的了解，那么我必须问你以下问题，在最坏的情况下，我们看到一些意想不到的后果或事件发生在台湾海峡，什么时候会发生，让我引用两个消息来源，一个根据一个麦肯齐的报告说，从经济角度来看，疫情会比疫情更严重，其次，你可能已经在电视上看过台积电董事长李克强的采访，记者向他提问，我想引用他的回答，如果发生这样的事情，你要担心的不仅仅是 chps，让我们确保它不会发生，我会祈祷，让我们确保它不会发生，好吧，我们已经谈了很多，任何最后的消息，菲利，给东南亚或世界各地的人们，这样他们就可以感到更加乐观，他们正在变得更加积极和建设性地实现 STEM 能力，希望也拥有某种制造能力，能够与世界其他国家合作，所有这些都是好事之一，对那些对先进制造业感兴趣的发展中国家或 S 国来说，一般来说，不仅仅是芯片，而是先进制造业，这是因为更新了意识对供应链驻留等的敏感性，现在正在进行更多的多元化努力，这些国家过去的历史并没有真正参与这个领域，现在有一个很好的机会，这意味着他们需要准备好抓住这个向他们开放的机会，这意味着借鉴我之前提到的台湾经验，这意味着人才培养和政府政策来支持它，作为一般社会当地社会来支持它，顺便说一下，很多人谈论劳动力发展，我其实不太喜欢这个词，我喜欢称它为人才培养，因为它真的不仅仅是那些重复做同样事情的工人，而是真正能够继续学习和尝试做新事情的人，了解他们正在做的事情，并能够改进它，这些是人才应该做的事情，而不仅仅是传统意义上的劳动力，另一方面是一般社会支持关于这个行业，我再给你讲一个有趣的故事，台积电想在圣路易斯建立一个新的工厂，那里有很多工厂，政府为他们划了一块土地，不幸的是，那块土地上有一座历史悠久的寺庙，我在报纸上读到过这件事，所以，当然，寺庙是神圣的，没有人想移动这座寺庙，但是最后他们决定，好吧，我可以为你搬迁这座寺庙，这是关于社会如何支持经济发展的重要声明，他们知道优先事项在哪里，他们知道优先事项在哪里是正确的，这是支持社会的非常有力的声明，让我事后再问你，如果没有张忠谋和一位政府官员之间的会议，权威和远见，双方都有远见
对吧，台积电可能有，也许吧，也许吧
我的意思是历史的发展是因为某些事件和某些人
对吧，这是偶然的，是的，呃，你知道中国有句
中国谚语说，是形势造就了英雄，还是英雄
创造了形势，对吧，我认为两者都有一点，对吧
即使有官员与某人展开对话，如果该人不够精明，
也不会发生任何事情，或者有一个非常有远见的人，但没有相应的政府官员以同样的方式看待事情，
什么也不会发生，我认为两者都有一点，是的，我认为
两者都有一点，之前你问过关于亚洲国家的经验教训，我认为在美国也有经验教训，在美国我们谈论
制造业在岸化，对我来说听起来就像给别人一条鱼，是的，你还没有教他们如何钓鱼，是的，我同意，这是很重要的，嗯，要理解，另一件事是制造和创新或研发齐头并进，制造不仅仅是对食谱重复做同样的事情，制造车间有很多创新，如何改进制造过程，如何提高产量，如何让当前一代技术变得更好一点，这样你就可以提供基于当前技术的更新版本，嗯，只是轻微的变化，不是大的变化，而是为你的客户提供价值的轻微变化，你怎么做到的，台积电每一代都这样做，他们会想出制造 Flor 的创新，将其变成比上一代略好一点的产品，吸引客户，你可以告诉客户嘿，只要稍微提高一点价格，你就可以得到更好的性能或能源效率，或者更好的成本不需要你进行下一步
这对客户来说很有价值，所以制造、研究和
开发和创新真的齐头并进，这是硬币的一面，硬币的另一面是，研发必须非常接近
制造，否则你梦想的解决方案是不可制造的
你怎么知道什么是可制造的，什么是不可制造的，因为你离制造车间很近，换句话说，很难将研究开发与
制造分开，所以如果我们只是说嘿
公司X来俄亥俄州亚利桑那州德克萨斯州什么的，然后建立你的
FB，我认为基本上就是将
南加州的制造业与其他部分隔离开来，这是不会成功的，你必须意识到
这两个组成部分的不可分割性，是的，它们是同一枚硬币的两面，它们是同一枚硬币的两面，你不能只是说我只把制造业带到岸上，我不在研发部门做这些这非常重要，所以
换句话说，所有这些芯片补贴 520 亿美元，390 亿美元用于制造补贴，110 亿美元用于研发，应该翻转，好吧，可能不会翻转，因为你需要付出更多的努力和金钱来进行人工制造，但其中的重要性是
是的，是研发，最重要的是 110 亿美元，如果你这样做
P 对，是的，我这么说不是因为我是学者，我做研究，但
如果你不这样做，我甚至会为研发投入更大的数字，确切地说是 10 11 对，这必须做对，否则你会
得到一条鱼，你吃了那条鱼，然后就这样了，你必须教渔民如何捕鱼，绝对的，我认为这是最重要的事情，对于渔民来说，要想成功，你需要能够与其他渔民交谈，是的，你知道在哪里，所以他们知道鱼在哪里，是的，你必须有一条船，是的，有一条船，是的你知道，过去几年有很多关于 FR 岸上回流、离岸外包的言论，但如果你看一下经验证据，就会发现东南亚的资本形成并没有真正以有意义的方式增长，我推测甚至假设，这在很大程度上是因为缺乏对人力资本的投资，MH，对，而我们真的需要齐心协力，你知道，有回流的愿望，有搬迁的愿望，但如果我们没有准备好，是的，你知道，机会不会来，机会并没有像我们希望的那样到来，是的，Capt 抓住机会，你需要做好准备，人才培养真的需要做好准备，你看看中国，他们毕业的工程师比世界上任何国家都多，这就是他们成为世界制造业中心的原因，去 sunen如果你想制造某种硬件，你必须成为是的，他们不仅拥有人才，还有基础设施，而且
人才真的很棒，他们知道如何做事，所以你需要
投资
how how did you grow up uh you grew up in Hong Kong and how did you get interested in whatever ended up you know in your field right now all right I grew up in Hong Kong Hong Kong was uh a
British colony at a time and uh most people in Hong Kong were rather
relatively poor uh at the time it was like a a developing I mean it's not a
country it's a colony it's a developing society and uh most people were uh not
very well educated at the time in my I my parents generation they all get CL by
the wall and the World War II and the educations were interrupted and many of
them don't formally go to a school and and uh both my parents didn't did not
finish high school and um but they were just like many Asian parents they are
very adamant that all the children should be well educated and uh and that
that's what they put the emphasis on so I went through I very Grassroots type uh uh family we
live in very cramed departments and uh when I was at the very young age we got
we kind of won the lottery uh from the government because at the time the government was building plenty of
government housing department highrise apartments that have uh that they rent out to the the population at very low
prices low rent and so we sort of Wonder Lottery and moving to one of those and
that really changed our tra trajectory and we stay there I grew up there in this in the housing project and uh and
later on as our family get little bit more resources we moved out to to uh to
private housing but the early years we grew up in a housing project which really uh immersed us
into the general public of the society which give me a pretty good perspective
of what the society is about who would have put more premium on education amongst your parents was it your dad or
your mom both of them because they they didn't have a chance to be to go through formal education and so they realized
and right after the more everybody needs to make a living so to speak and uh so
there wasn't much of opportunity for them to get formal education so they are
very very insisting that everybody should uh in the family should be you well
educated how did you discover electrical engineering I mean you could have done other things right yeah I could have
done other things um that was I guess the society at a time
uh gravitated towards of stem type education because that seems to be a
kind of like a ticket to a better life better career um and the humanities and
the Arts are less well defined for example if you got if you're licensed to
become an engineer then of course you're licensed to practice you got a career and so on whereas the other profession
seems to be more nebulous and that's well defined and so
a lot of kids at the time who who did well in school tend to pick
one of the stem fields to go into and uh so no exception for me follow follow the
crowd and uh enter kind of into like a science type uh curriculum and out of
the science type curriculum the the physics and the chemistry part of the of
the science curriculum is more more kind of quantitative and formal and makes it
easier to quantify and at least to me it easier to understand and rather than for
example in biology for example is more a to me a time it looks like a more
observational science you need to remember a lot of things right and uh
I'm not very good at remembering things so you're being modest so the physics
and the chemistry looks better um now where when I was about to enter
college or university at a time the careers for people who study basic
science is rather limited because it's not a like a basic science research as
in the US it is today right B being a basic science researcher is not a career
in Hong Kong at a time right so most of the science type careers are more
tactical engineering is one of them so many of our contemporaries becomes
engineers and uh out of all the engineering electrical engineering is
closest to physics and uh it use many of the same
Concepts in physics you know quantum mechanics and uh electromagnetics and so
on they are more closer to physics so that that's why I get attracted to it yeah you know it's it's I think it's
probably safe to assume that countries or nations or governments or political cultures within
each that would have put strong emphasis on stem they tend to
have done well as opposed to those that probably put less emphasis on stem right
would would you attribute this to the political culture in your household or
sort of like the pervasive political culture at any Dimension at any level or
you think this would have been something that would have been driven by the leadership
of that country that Nation or the government I think U if you talk about I
think if you look at many of the Asian countries they have the strong emphasis on them yeah and
my crud understanding is that from Economic Development point of view you do need this skills uh from Economic
Development point of view if you want to build infrastructure you need Engineers to build the bridges and Roads and so on and so forth if you want to have
industry you need people who can do the engineering mechanical engineering electrical engineering and so on so forth that's I think mostly coming from
the point of view of economic development but I should say that
economic development in itself only takes you so far at some
point the society will need need things other than Economic Development and that
and that's where the humanities and and the and the Arts and so on comes in that
people in society are not just simply satisfied with the money right but
beyond there's a lot more than making a living once you are comfortable about
making a living then you look for something else I think that is the part that a lot of the
Asian countries may not have recognized I'm I'm just you know sensing
that there's quite still quite a lot of countries that have not focused on stem
as much as some countries that we might have seen why why so not sure I mean uh I I'm
in the camp that believes that I think a higher degree of focus on stem is correlated with your ability to move
marginal product it because that's correlated with Innovation and all that good stuff right and and being able to
combine that with liberal arts or Humanities I think that's that makes it
even more a good combination U but some countries I think tend to focus a lot
less on stem and that that I think has hampered development uh to some extent
and and I'm just happy with with how Hong Kong you
know uh has has put a lot of emphasis on stem along
with some other Asian countries you know such as Korea China Japan and you know
now recently you know you've talked about Vietnam and Southeast Asia right U you can see that in in the tenacity with
which the people of Vietnam I think are focusing on stem uh I come from a region you know
called southeast Asia where I think the focus on stem needs to be you know more
emphasized going forward uh you went to Hong Kong University yeah then you decided to move
to the US yes why well very simple um I
am interested in kind of more advanced uh knowledge
beyond the bachelor's degrees type uh education I was looking for place where
you could where I could get a PhD uh at a time it was uh in early
8s uh none of the universities in Hong Kong are what you you would call a
research University they're basically teaching universities just in the colonial days
the universities in Hong Kong uh man to manag to train uh bureaucrats for the uh
civil servants in service of the government and they're not interested in doing research so there's really no
opportunity to do any research uh in Hong Kong so I came to the US and you've it's worked out okay with
you so far I I I'm okay I guess now I want to I want to shift gear
to your your area of focus I mean you I know you're teaching electrical engineering and you're one of the few
people that understand semiconductor you know sector why why is it called semi
not full conductor that's interesting but before I get to that let me just finish by other uh uh
your previous uh uh question I think my kind of Journey from the from Hong Kong
to the US is a very typical Journey for many uh immigrants who are in the
technology field right uh they came they get educated in their home count home
country and came to the US saw opportunity to get better education and
then many of them to stay and that's I think it's a very important source of talent for the US and uh maybe we could
come back to that later but I think is really important if you look around in
Academia in Industry many of them do come through this route of coming
getting an advanced degree and and up staying it's been the main source of innovation in the US absolutely if you
look at top management CEOs of big companies right they all many of them
are come through this route yeah and that's very important yeah well while on this I mean do you
see that as something that's likely to continue given I think that's in Jeopardy okay uh given the Reon the
political calculus seems to be yeah this seems to be the the the pendulum has
swung yeah on to to the other side uh in the in recent years and you see a lot of
uh soul searching about about uh immigrants or foreign students and
openness in research and things like that and that I think that creates an environment in which when for a St
potential student from other countries looking at the US and they would begin to think about whether this is the right
place for them or rather they would go to some other countries to get get their education and experience yeah when I was
in just graduating from college that that answer is very obvious
where do you go the US of course yeah but now now you've got option alties now we got options and also not only options
from a pull point of view but from a push point of view correct a lot of people felt that the US may not be so
welcoming as before and so that I think that brow or that flow of people is in
in somewhat in japy on on the other question why is it called semi oh why is it called semi I actually I forgot who I
heard it from when I somebody talk about semiconductors and somebody who's not in
the field ask why semiconductor why how come full conductor is not
better cup half full well uh what is a semiconductor a a
semiconductor is a material that it as it names implies uh not very conducting
not very insulating and in somewhere in between and why do I want something in
between is because a full conductor would be something like a piece of metal
like copper or silver or that kind of thing right so for full conductor now
not a food but real conductor basically U why semiconductor sem doctor
uh is used to build switches uh like electrical switches wall switches flip
on the switch turn on the lights and computers are actually a bunch of
switches they we comp built by designing the organization arranging switches how
to route the switches around um in the early days the the the telephone uh you
have a switchboard and you have real people who connect uh
one line with another uh the uh that in a switchboard and all these are replaced
by chips today of course and so computers do computation by flipping
switches basically so a semiconductor can be turned on to be conducting or be
turned off to be not conducting and that basically is what a switch is about so
semiconductor is used to build switches and so why that's why some conductors are useful because it can be turned on
or turn off Gordon Moore he came up with this prognosis
decades ago that you know things were going to behave in a certain way by way of the capacity growth the speed growth
you know it's been very very accurate yes right every year and a half to two the doubling of the capacity and the
speed what's going to happen with Moore's Law okay first of all that uh
Mo's law which you refer to uh uh doubling of something um initially he
was just suggesting doubling the number of transistors on on a chip and uh later
on people modified it and double
anything which is not really what he meant but people can tend to associate
it with MOA that way uh but um that one is basically a self-fulfilling prophecy
uh and it's actually a rather unusual situation for a technology or for
industry uh I believe I one of my uh colleagues of Professor Jim prer said
once said that he's not aware of any other
industry that has this kind of selfing propery and namely that everybody thinks
that in X number of years such things will happen right and so because
companies are competing because there's a prediction out there companies are trying to either
meat or exceed that protection so that you can be better than your competitor
so as a result of that competition to gets to that particular goal that goal is satisfied because of
the intense compet because and also that also not only just
the desire to compete but also having a goal allows everybody up and down the
chain the entire chain of events from materials to device technology to Circuit design and architecture and and
so on so forth up and down the chain to have a predictable Cadence you say let's
say I'm designing something I need what I designed to we run on some kind of
chip now I cannot start designing the chip unless I know what the chip
technology can do what the capability is just like if you try
to run run run a marathon I cannot win the marathon if I unless I know what my
capability is so you need to know what the capability is before you can design your chip now without that prediction
you wouldn't be able to say I'm going to design this based on this assumption
that the chip technology will give me this kind of performance I cannot do that but with this kind of
self-fulfilling prophecy I know in two years the chip technology will advance
to a point where I could count on that particular performance and so therefore I will Tater my design based on that
assumption and that's very powerful very powerful it's sort of
framed the behavior of the industries absolutely right absolutely consistently
consistently for that case uh even that I gave the example of Designing chips
even down to the material supplier the material supplier knows that the chip manufacturer will need this kind of
material or will need this kind of material to have that kind of performance in two years so therefore I
work on developing it with the phone on is that when I get done I will have a
customer that's risk mitigation is very important if I develop something I spent
Millions multi-million dollars yeah developing something I didn't know whether there's a customer that's bad
yeah but in this case I know you know at the RIS of quoting you
uh in in previous conversations so at the rate that the percentage of how the chips
occupy Our Lives if not Lifestyles at an increasing rate whether it's opening
your window in the car turning on your toaster turning on your microwave oven
whatever yeah before you get to the more complex stuff what's what's going to happen to
Mor's law is it likely be is it likely to be more
exponential than whatever we have seen in the last few decades I think there are two dimensions of it um
well for sure the answer is yes it's going to be more exponential yeah okay why do I say that uh first for the last
several decades we know that there's one way to improve the performance of the
chips which is to make it smaller and there a lot of physics involved and to
to explain why making it smaller is better right so we know that is one way
and we've been kind of turning that knob for several decades now right and
that knob has kind of saturated a little bit just like anything else you put the
gas the gas pedal it hits the floor right in the beginning you accelerating but then pretty soon you
hit the floor right so that's for sure but that doesn't mean that progress will
not continue to be made because there are many other kns there many other ways
to make progress it's just the what the analogy I like to have is for the past 50 years we've been walking inside a
tunnel we there's only one way ahead we know what the goal is and we just want
to make progress and to make progress that's one making devices making transistors and memory smarter will
allow us to pack more devices on the same chip and get people therefore get more functionality better performance so
on that method is the one method that we know and that is saturating but it's
more as like walking inside a tunnel there's no way to go it's only way forward but now we at the exit of the
tunnel now the exit of the tunnel there are two ways to interpret it one you fall off the cliff you're just
dead second is the world is open wide
open no more confining you cannot go left or right right you can go anywhere
you want I think my interpretation is we the exit of tunnel there's so many
possibilities and we will continue to grow in in capability in a exponential
way we been faster now why am I so so optimistic about that is
because necessity is the model of inventions and that arise my optimism
optimism arise from the fact that Society demand more from these chps not
less the society will demand more and more from semiconductor Technologies of
CHP technology in the coming decades than we have done in the past few
decades so because coming from the demands Innovation
will come up and bring the the technology up in a much faster pace
before because we are not confined by the tunnel anymore what could go wrong what could go
wrong geopolitics go wrong I think that's a very important aspect that
wasn't there uh just maybe like a handful years ago I think that is an
important part because research and development thrive
on collaboration and there are many many examples we can go into that a little
bit more later but it thrives on collaboration and if
geopolitics cause collaboration to not happen then the rate of progress will be
severely curtailed I think that could is something that could go wrong you know
with we're going through this era
of what I call multipolarity right and and I've I've said it before that
Ironically in a multi-polar world multilateralization becomes more
difficult because there's this tendency of revisionism by the new middle powers
or greater than middle Powers right and the pre-existing superpowers are not
comfortable in embracing this new mode of multipolarity I want to put this in a
context of what you've been affiliated with
tsmc right they've they've
basically created a recipe for Success that involved a lot of collaborative
undertakings right from a supply chain standpoint from a research and development standpoint from what have
you not just amongst Enterprises but also between Academia and policym
and Entrepreneurship right what what made them the
way they were or they are I think your observation is extremely astute in the
sense that a lot of people think company like tsmc is a collection of factories
and tools right or even just uh uh Factory workers yeah and that's about it
but really it is not it is a Enterprise
that involves a lot of collaborations that thrive on the
network of customers network of
suppliers network of collaborators in research and development network
of collaboration with the broader society that which they lives in right
the Taiwan Society it is a network of all these
things and even more perhaps I didn't um go into so that piece that collection of
relationship is what make this company the way it is uh let's say we all of a sudden
transplant this company to another country or
another place that doesn't have the same political economic
system it is Highly Questionable whether they continue to operate yeah the way it
is yeah so how do we how do other
countries try to replicate just any bit of what tsmc might have been able to do
and and I'm not single singling TSM tsmc out here there have been others that
have been legendary and great right Samsung's of the world the Intel's of world absolutely uh but but I think tsmc
has proven itself to be very resilient and very different MH uh and
their success seems to be sustainable so far so far yes right and go back to the
history of how it came about M go back to that conversation between Morris
Chang the legendary Morris Chang with you know somebody or some people in
Taiwan that made it the way it has become I think there are two if not more
at least two very important uh elements one technology competition is really a
talent competition first talent second the G
the broader society's support of the industry not just the company but the
industry itself the whole lead a infrastructure for the company to exist
and suppliers customers and so on so let's F let's talk about Talent first
right Taiwan in the early days of the 60s 70s through the 80s and
90s have sent well it's not that the government sent them but students from
that era all after they graduate from college they all went out to the US to
Europe outside of Taiwan who has more advanced education uh achievements in
the universities to get education and many of them stayed in the Local Host
countries to gain work experience and in time many of them go back to
Taiwan to pursue the continue the career for variety of
reasons want to be closer to home and uh I guess that's maybe perhaps one of the
biggest draw be closer to home or perhaps a perceived better career
opportunity in the Home Country so like playing in home field right so which
people sometimes feel more vantages and being knowing the culture
and the people well so Talent the flow of talent is one of the most important
thing if you look at the leaders of the tech industry in Taiwan
today many of them were educated in the US and through this route getting
educated in in college go to other countries the US Europe and so on and go
back to Taiwan have work experience go back to Taiwan bring the work experience
bring the knowledge with them and that is very important that is I would
say I wouldn't perhaps is the one of the
major ingredients for the success and it's not just limited taian career as
well yeah we look at the many top Management in Samsung and heix and so on
many of them were educated in the US um so that's one part of it but
that's only that's one piece but the other piece which is important is support of the society if you think
about how the semor industry was started in Taiwan it was really supported by the
government it's industrial policy basically the T government decided that
this is an industry they want to get into and they supported with building
industrial parks with starting it the industrial uh technology Research
Institute which spun off uh SMC and many other companies as well so the
government really have a really robust industrial policy now of course
industrial policy may not may or may not succeed if you bet on the wrong thing that's bad right and that's really
of course if you talk if you're in the US you know that the US feel that they don't need an industrial policy right
and we don't go into that but in this but the case industrial policy worked out for them and the government really
supported it building industrial parks uh finding
other pieces of the puzzle to fill in the uh infrastructure for this for the S
industry and also basically the technology industry uh back in the 80s
and 90s and early 2000s Taiwan basically man ufacturer but basically it's the
world's manufacturer for PCs and everything even bicycles bicycles bicycles well yeah and so the government
really supported it with IND with really strong industrial policy that's amazing
now you know in in early days a lot of the intellectual property with respect to the
microchip technology was done in the US MH do you think that with the benefit of
hindsight of the US were to decide otherwise mhm meaning they would have
wanted to onshore the manufacturing capabilities as opposed to Offshore the manufacturing
capabilities would things have been different um this is a very complex
problem that I i' I've been thinking about more recently and I haven't come
to any reasonable resolution or answer yet and I I can tell you what the
uh uh what the kind of Disconnect or problems are one is in the US this as I
said this is a technology competition is a talent competition yeah if you look at the cost of
talents in Asian countries and in the US the cost differential is pretty high
yeah right so um for a it's getting smaller but back then it was huge was
huge was huge right so to get t Ed people working on heart engineering
problems in let say in Taiwan or in Singapore or in Korea yeah is way less
costly than it is in the US so in other words they could deploy many high skill
talents to solve engineering problem whereas the US is at a disadvantage in
terms of cost not Capital cost but just Talent cost right give an
example um I recently went through more than 20 years of tsmc annual report every year
they report on the uh education level of the
employees uh starting from 1999 about uh 40% of the employees are
have a bachelor's degree and above today
80% wow now that
coincides precisely with the steady climb in the
technology uh letter uh the the leadership race the
steady climb from beginning they were licensing it from RCA licensing technology from the US and
bring it back to Taiwan gradually they learn how to do things themselves and
the way to do that is to get talented people so they started from 40% people
employees who has uh bachelor's degree and above to today
80% and so that really tells you technology competition is a talent
competition you okay in this context um which other
countries do you think have have done as much or just about the same
amount that Taiwan has done and I mean you know when you think about Talent you
think about human capital you think about countries that have shown right abilities to project intellectual
Capital right within Asia or within the world which which countries do you think
are likely to be able to emulate the kinds of successes that we've seen in Taiwan well I've seen like
countries like Singapore Korea have been tremendously successful right Korea I
remember back in the 60s and' 70s Korea was nowhere to be found right and then
later on they have kind of a me too type products right and we associate those us
start out with meos yeah start start being Innovative and unique exactly and then slowly they climb up the uh the the
the learning curve and I remember years ago when you see Samsung say what the heck is this you know not reliable now
it's the best thing the world so so these countries have done quite
well what about China China is I don't think we could ignore the
capability uh they have let me talk about something that I'm I'm familiar
with just research right Academia I know research years ago we go we have
technical conferences and we see papers from China said ah forget it
just quality is so bad it's not even competitive different that was probably
in the 80s and the 90s now they're better than us they're better than us if
you look at I look at uh papers uh Publications data from Key conferences
in the chip spaces uh for example in the international electron devices meeting
uh which is a device technology conference the back the biggest one in our field International cir sze circus
conference the circus and system conference in our field the two biggest conference in our
field um you basically basically flipped uh years ago the US had the
majority uh of the papers I remember there were roughly about 40 60 40 40 to
50% sent of the papers from the US and China maybe 20 30 years ago they were
nowhere to be found today uh China and Asia the papers are more than 40% almost
close to 50% oh my gosh and the US has steadily declined from 40 50% to 30 to
40% and the rest of the world principally Europe and Japan has basically fallen off the cliff
so the research and development the research capability in Asian countries
China Korea Taiwan Singapore and so on have
really become the strongest region uh in in terms of uh producing
good quality research I'm not just talking about quantity just Quality quality and uh the only thing that I see
that the US is still little bit ahead is
in coming up with the new ideas what the Chinese always say going
from zero to one namely starting from nowhere nothing
and come up with this really new idea and if I look at the what I would
call new idea that is not been discussed
before the US still if you read the research papers the US still is the
principal place where these new ideas come from but once these new ideas
become known then you you see mushrooming of this new
ideas in other places exactly and um I
feel it in in my everyday research my students my gosh any new ideas that we
come up new ideas that we come up with once they become known this is a good
idea next week it will show up in China next week show up in China not only that they do it better than you yeah I can't
cannot compete anymore they have better resources they have more students they have more more funding from the
government I cannot compete anymore I have to get out of that field I want to get back to this but uh
I want to ask you about two countries like like India and Israel I mean they
have tremendous human capital right do you see them as
contenders um and being able to you know replicate the kinds of successes has
tremendous potential it's tremendous potential their human capital is
unbelievable um not only they are very populous country of course they got
scale yeah they got scale right and the statistics work out for them right when
you have scale then there are going to be outliers who are really smart and
very Innovative and uh and so on just like China right so that that they have
that's very important um I think they have tremendous potential if they do things
right because they got a they got scale the people are smart right and in terms
of Israel they don't have scale a small country
but they are smart people obviously the difficulty they have is lack of scale a
bit like Singapore then yeah a bit like Singapore right Singapore lot of smart people million people 5.5 yeah that's
finite yeah finite they they don't have scale going back to China you know with
the Advent of AI how do you think the speed of new ideas creation is
going to be impacted I think speed of new ideas get AI to create a paper right pretty much yeah
but I think ideas still come from people um is
the is the acceleration from an idea to something useful that AI will be able to
help with um the idea still come from people but today there are a lot
of barriers and times and efforts required to bring some idea to fr
perhaps the Advent of AI would shorten that life cycle but the idea still come
from people I believe now um AI would really help
with the US and more advanced countries to get back into the game of
manufacturing let me explain why I mentioned earlier
that the cost of human capital is you
cannot ignore it the differential is there because of the standards of living and so on so forth it's dead now with AI
and Robotics we should be able to make a lot of progress in advanced manufacturing
because in man Advanced manufacturing you need many high skilled people because those manufacturing are so
complex it's not like doing repeated doing things over and over again
according to a recipe you need to be able to understand the process improve
on it and those require High skilled people Highly Educated people now the
developed world have very high cost for human capital that being the case you want to
make really good use of those people those productivity has to be much higher and the advant of AI robotics would
allow developed countries to deploy these highly skilled people much more
effectively than before and so using that would allow us to level the playing
field a little bit about the cost of high skill labor in in the next 10 to 20 years how
do you see the landscape of the competition you know in terms of
countries I mean I I think China is definitely pouring a lot more money
right onto this the United States seems to be wanting to put more money onto this not just on the IP site but also on
the manufacturing s Korea seems to be very right tenacious right how how do
you see the landscape it's I mean it's a bit like a Sputnik
moment kind of yes you know from from a chip standpoint and not only chip but
but many technology as well many other Technologies you can think about like you know electric vehicles batteries and
so on they all like that right so the world is becoming little bit the world of manufacturing or technology becomes a
little bit more polarized right so you kind of separate into two cams or two
and a half cams and things like that and people need to some camps will some
countries we need to play both sides and find their way around but the speaking we may think of it as two camps right
now that being the case then as I mentioned
earlier for technology to advance you really need open science you need
collaboration so now being that they're two camps they don't talk to each other that means things are going to go slower
what does that mean that means the society going to benefit less because things are going to go slow right second
is the separation of supply chain cost will go up and you need to find your
suppliers not the lowest bids but thinking about geopolitical situation so
cost is going to go up now the real the question is who's going to absorb those
cost and it's not clear who's going to absorb those cost and today if you look
at in part why Manufacturing in the US
is such a bad situation today is in part I
believe I'm no Economist I'm just guessing I'm an engineer right so maybe
an economist listening to this podcast would say wait a minute what's this guy start talking about right but my guess
is that the distri ution of profits is not Equitable across the entire
chain the companies who are who are operating
at the top of this of the food chain the
they they glean most of the profits I'll give you an example I was just at Western Digital the other day and they
told me that if you want to buy iPhone with more more storage they charge you
$100 or $300 more but the people who make the chips to make it happen get a
few dollars so you were talking about Level
Playing Field right there's not a lot of equity there
there's a lot of equity there right so this is this situation needs to be
corrected otherwise it will the manufacturing in the US is not going to
be uh go well I remember for example we talk about Talent right I was once in a
panel discussion with a bunch of sctor companies and one person from company
lamented that oh we couldn't hire people nobody wants to join my company we have
positions ready to fail and so on so forth lack of talent lack of Workforce
so I asked him how much do you pay them and he went silent he couldn't answer my
question why because they couldn't pay them to compete with the other people
who have better who reaps the profit right and because they couldn't pay them they
couldn't get the enough profits so that's why they couldn't pay them and so they don't they couldn't hide the people
so this this scarcity of talent is real really structural it is structur true
it's not so much A lot of people thought that oh we just need to tell the universities to offer more
courses no not going to happen not going to happen you need to have the demand I
could offer 10 courses but if nobody take my course it wouldn't help you at all right
you need to have the demand from the students and where do you get the demand from the students the students need to
know that they have a bright career in front of them yeah if they take this courses if they go into this field they
will have a good career today is not the case for many of these manufacturing
related jobs so you're you're suggesting if I get you correctly there's likely to be a
deceleration yes of the competition by way of the polarization if we haven't manage it
right right well hopefully we'll manage it better right uh and also the scarcity of talent a scarcity tent I I I think it
it applies on both sides yeah SC yeah exactly the scarcity of tent it's not just on this side of the Pacific I think
it's on the other side of the Pacific too exactly given the scale of course exactly exactly and that has to do with
how the profits of Fairly shared right because the
companies who are doing the manufacturing and all the things Below in the supply chain are not getting
enough profits to offer good compensation to offer good working
environments to the to the employees even within the supply chain there's there's some players that are
actually making disproportionately more than the other oh yeah okay oh yeah that's that's a structural impediment
yeah it is to to the desired speed with which you want to move it is I think uh
the policy makers and The Economist should re has to recognize that
otherwise it is structural the the aspiration to
onshore in the US you know there's been some conversations discussions and
initiatives for tsmc to build the plant here and there in the US including in
Arizona do you see that moving smoothly or as smoothly as aspired or well I was
uh at this big conference in December I told you earlier field speakers conference I saw all my colleagues from
around the world from Europe and Japan and so what my Japanese friends are all AF at us you guys are the
slowest okay that's public knowledge I guess
wow now talk about you know there's the logic chips there's the memory chips and
there's the special chips right we call it discreet or whatever how will that fair going
forward in terms of who's going to be more dominant than the others within each one of those three
categories I think those three categories they're all important yeah
it's not like one is more important than the other and more important like for example we're doing this podcast right
uh you need computers to Crunch the data and right stream the thing out going to
story it too somewhere yeah up stories to store right and the microphones the video
cameras they all made the chips those are specialty chips yeah right so they are important and so we couldn't say oh
we can we can do do without one of them no we cannot they are important now as
far as resources goes um the today as of today the development
of logic chips requires a lot of the most of the resources because they are
the ones who are driving in our kind of Industries Partners is the technology driver the
requirements Drive the development of new technology new equipment new
processes new materials new devices new design those applications Drive uh the
uh with AI it drives big computers that can crunch this uh
numbers for the AI models and it also drives development of computing chips
that are so energy efficient you can do things on the edge what they call the edge of the internet right you could you
go buy a VR headset today you have to have a battery pack to come with it and
that's is not the end state St of V headset I'm pretty sure about
that in the in the final analysis V is not going to operate like that but this
is kind of like the Intermediate stage that you need to go through just like sell phones used to be like size of a
break right and I remember those you remember those right in the early 90s in the early 90s right and um
those can only make a call and cannot do anything else right and I tell your
story when I was at IBM uh I used to work on Imaging image sensors and in the
in the mid90s I had that idea I want to file I want I filed the invention disclosure about putting a camera on a
phone on a cell phone this was in the '90s right it was in the 90s and the IBM lawyers tell who wants a camera with a
phone we're not firing this time say no more
after like months I finally convinced him to fire it and we do have a B on that they had a great ability of
prognosticating about the future right so all these things um uh
the all the progress that we desire the
chips to have to we desire the chips to have these capability today is driven by
comp Computing Computing means but it doesn't mean that it it will continue to be the case doesn't mean that it and
also it doesn't mean that there's no uh real need for innovation in others uh
other places for example just take image sensor for example tremendous innovation in image sensor and and they offer new
capability in V headsets you need to have new Innovations in image image
sensor and the way they receive the signal convert it and uh do processing
right up front not right at this centor end and not back in the main main frame
computer so all these Innovation needs to happen in order to kind of satisfy
our society's demand for technology in general if you look at for example the
17 uh sustainable development goals of the United Nations I I basically go
through all of them and say hey at least 12 13 of them require advances in
Computing technology and chips technology in general so without those all those development goals will be
severely curtailed the the backers of all these are consisting of the designers and the
manufacturers right yes now the qualcom of world the nvidias of the world they
they seem to be content with just designing right how do you you see the
progression going forward or the shifting going forward are they do you are you likely or are we likely to see
them moving in the direction of manufacturing given the comp you know
the complications or potential complications we might see in the context of this polarization wonderful
observation 30 some years ago when Morris Chang came up with this uh split
between fabuless namely companies only design and do not any have any f Factory
for building chips and foundaries and Foundry focus on building chips and not designing anything that that's
demarcation that split caused a blossoming of design
activities because you don't people who design do not need to own the capital
and all the baggage of owning of factory and so they could then specialize on
doing design at the same time people who owns the factory can specialize on running the factory and also having the
benefit of having zillions of customers instead of one customer if I manufacture
my chip I sell my chips directly to the customer then I need I only one customer
that's myself that's not enough to have scale to have this economy of scale for
the manufacturer so this split has really blossom in terms of making the
design ideas go a lot faster than before and
having a lot created a lot of business opportunity that did not exist before because previously you have to have
tremendous capital investment before you can design your own chip now you don't you just have to to design and somebody
will make it now that split has gone on really well for the last 30 some years
but today if you look at the fabulus and The Foundry they not really split they
actually together companies like apple Qualcomm Nvidia AMD they show up at the found
every day they tell The Foundry hey run this wafer try this experiment show me
the results give me the Wafers so I can test it myself so they are actually
doing a lot of the development work of the chip itself so the the Leading Edge
technology basically would our today co-developed between The Foundry and the
leading user whoever place the largest order have a big say in how the
technology is going to be designed and optimized but what about skin on a game though in terms of manufacturing
capabilities are they likely to see put more skin on it it seems to be the case
the there are indicators in there already for example Sony putting money in Japan and
in kind of work with uh collaborate with tsmc to build a factory in Europe the
collection of automobile companies having uh putting their money down to
build a factory there so you're seeing some of that and where how far it will
go is not clear but you see beginning to see some of
that talk about this company in Amsterdam or in in Netherlands asml mhm
they they they make a pretty fancy technology right ultraviolet litography
yeah that's well needed for a bunch of things right talk about
that a technology marel um if you go into the details of
how this machine the euv machine extreme ultraviolet uh mythography machine is
really an engineering Marvel now it is also so a statement about two things one
it's a statement about a global supply chain the laser come from San
Diego the lens is Polish in Germany the mechanical movement came from
Connecticut so it's not just the delence it's a global supply chain of not only
the machine parts itself but also a knowledge
Global Supply Chain of knowledge that's one the other one is
the is an examp it's an example of open
science the first EU papers were published in the late 80s
by Group B apps and Then followed by a bunch of conferences in which
researchers in Japan and in Europe all participated and those formed the and
then the US National Labs participated in the development of the early
prototypes in the department of energy the National Labs all these and then
companies start coming in Intel tsmc and so on and start investing and we're talking about global
companies right not just one country global companies investing in it
and years later 30 some years later or maybe 40 some years later you have this
wonderful machine and that's all coming from a global collaboration of
researchers generating knowledge sharing the knowledge and bootstrapping from
themselves and that's a very important lesson for us to learn is that if you
you cing yourself off from the rest of the world you wouldn't be able to
achieve that or even if you are able to do that you'll take a much longer time
and you you lose the window of opportunity that's a very important uh
lesson to learn in this day of age day an age of
geopolitical tension and everybody wants to kind of have everything to themselves
and not sharing with each other if you don't share you don't
gain you g you gain back more than we share I think that's basically what I
think what how I would characterize it you breed a sense of
optimism just by way of how the intertwining of different stakeholders
in multiple countries you you seems to suggest that that sort of that serves as
a structural cohesive
force for that to be very it's it's very difficult to rip it
apart yes right yes and if you do rip it apart you you
lose everybody loses everybody lose yeah yeah everybody lose but you know given
the fact that we're seeing so much political Neurosis in many parts of the world
right sometimes people don't act rationally yes they act based on what
their heart tells you not what their brain tells you is not a risk an there's a risk but
I think in in some parts I think is due
to the lack of understanding rather than any bad
intent it's not because somebody's bad intention or anything
like that but I think it's more like a lack of understanding uh uh I came to that
that um observation um by my through my
participation in a in study from the Hoover institution about two years ago they are study about us China Taiwan and
semic conductors and they publish after two years they published a report called silicon triangle you can go to tuber
website and download it and through that participation I
understand my observation is that many of the people in the policy Circle
they don't understand the technology yeah therefore because of this lack of
understanding they would make policy that really doesn't make sense from a
technology if you do have this Tech techology underpinning it doesn't make sense now it doesn't mean that the
people making policy meant bad or anything like that it's
just a lack of understanding it and it also doesn't mean that the people making
policy are dumb or anything like that is not nobody understands everything under the world right today they may be making
policy about the pandemic and tomorrow they may be making policy for sem chips and the next they may be nuclear arms
right so there's no way anybody any single person or even a group of person
can understand everything under the world right but I think policymaking would
really be helped if there are more people who have domain expertise are participating in the
formulation of the policy and today I think the we have very few people who
are who understand at least in the semiconduct space that I'm familiar with who are materially participating in the
formulation of the policies and I think that is a a big hole that we need to fail do do
you suggest a a higher degree of cultural interaction oh yeah absolutely
I I get the sense that people are not spending enough time in each other's yes places absolutely yeah like I said I was
participating this Hoover study many of them are policy people China to experts
and so on and at the end of the study we all understand a little bit more about
what the other people are expert about I didn't know anything about China policy
or industrial policy before but I have some at least some introductory
understanding about it by now and my colleagues likewise have some introductory understanding about
semiconductors they wouldn't ask a question why not a full conductor
anymore I have to ask you that and and I got to ask you what's
what's the difference between oil and chips you know I know you've been asked that what's what's the difference
between oil and chips what oil and chips yeah that's right I mean people have been saying chips is a new oil yeah yeah
yeah yeah so yeah so um I participate in this participated in this uh Hoover
institution study and a lot of people say oh chips we need chips and chips are
everywhere uh anything that you need to turn on the power have B have a battery
in it or parking door or involve some kind of a chips you roll up your car window there's a chip you listen to
podcast you a chip everything is a chip so it's like commodity it's like oil we need oil we need fossil fuel right so
people make a comparison between oil and chips and that has that comparison is
pretty prevalent in a policy Circle right okay now but it's misplaced I
think it's totally misplaced and when I made that comment at the uh study and
people look at me and say oh yeah okay why is it that different chips are L
like oil because oil has been in the ground for millions of years they never
change you take it up today you take it up tomorrow is the same oil the method
of taking may be different but the oil is the same chips or not you have phones you have watches you
have your TVs would you like to use a phone that was made 20 years
ago would you like to watch a TV that was made 20 years ago you don't because
chips constantly advance and it has to constant Advan to show its value people
expect chips to continuously Advan in this
capability capability broadly defined as in amount of things the kind of things that they can do and also Energy
Efficiency don't burn up to so much energy or power don't use up at that many batter that much battery power and
so on so the chips is useful to us because
it's valuable to us because it advances constantly it constantly renew itself
chips today is different from chips tomorrow you want the chips tomorrow not the chips of yesterday now
oil it's not the same you can stockpile oil no doubt and America does that many
countries do that too you don't Spock pile chips a stock power you have is not
useful very soon every couple of years they're going to be obsolete yeah every couple of
years obsolete yeah exactly is it feasible to attain chip
sovereignty or an absolute chip sovereignty given the nature that we
just talked about the global nature I'm just I'm imag not possible it's not
possible I think the whoever suggests that in put in policy circle is probably
not having enough understanding about how this how the thing works then I got to ask you the following question in in
the worst case scenario where we're seeing some
unintended consequence or episode you know happening in the Taiwan
Straits when would happen well let me quote
two sources one according to a McKenzie report it would worse than the pandemic
economically second you probably have seen the interview uh on TV of uh the tsmc
chairman Mark Li questioners post to him by the rep the reporter and I would like to quote his
response such a thing happen you have more things to worry about than just
chps let's make sure that it doesn't happen I'll pray let make sure it doesn't
happen well we've talked a lot any any final messages philli
to people in Southeast Asia or anywhere around the world so that they can feel more optimistic
about they're becoming much more you
know positive and constructive about attaining stem capable AB ities
hopefully some sort of manufacturing capabilities too and being able to be
collaborative with the rest of the world and all that well one of the things that
turn out to be positive thing for kind of developing countries or S countries
who are interested in uh in advanced manufacturing generally
speaking not just chips but you know Advanced Manufacturing is that because
of the renewed sense of sensitization to supply chain
residencies and so on there is more of a diversification effort going on right
now that these countries who
are historical in the past haven't really participated in this space Have a
great opportunity right now and that means a that means they need to be
prepared to take this opportunity that is opening up to them and that means
taking from the Taiwan experience that I mentioned to you that we discussed earlier that means Talent
cultivation and government policy to support it as
general Society local Society to support it uh the in terms of T by the way uh a
lot of people talk about Workforce Development I actually don't want like
that word that much I like to call it Talent cultivation because it's really
Talent not just workers who repeatedly do the same thing but really people who
can continue with learn and try do new things and understand the things what they're doing and be able to improve on
it those are what talent would do right not just generally Workforce in a
traditional sense of the word and the other aspect is uh this
General Society support of uh the industry I'll give you
another interesting story tsmc wanted to build a new Factory in sju in the place
where they have a lot of factories and the government zoned a piece of land for
them unfortunately there is a temple historical temple in that piece
of land I read about this in the newspaper so
and of course Temple is sacred and nobody wants to move the
temple uh but then at the end of the day they decided okay I can relocate the
temple for you that's an important statement about
how the general Society support the the economic development
they know where the priorities are they know where the priorities are right and
that's a really strong statement to support the
society let me ask you with the benefit of hindsight if there if there wasn't a
meeting between Morris Chang and a public
official who had the authority and also the vision well both of had a vision
right would there be in tsmc perhaps yeah perhaps I
mean history developed the way it is because of certain events certain people
right it's serendipitous yeah uh it's really you know the CH there's a Chinese
saying that whether it's the situation that created a hero or the hero that
created a situation right yeah I think it's a little bit of both yeah right
even if there is an official who started a conversation with a certain person if that person is not astute enough and
there's nothing's going to happen or is there's a very visionary person but
there's no corresponding government official who sees things the same way
nothing's going to happen I think it's a little bit of both and U yeah I think
it's a little bit of both and earlier you asked about any less lessons from
about selfe Asian countries I think there's also a lesson to be learned in the US as well in the US we talk about
onshoring manufacturing to me it sound like giving
a fish to somebody yeah you haven't taught them how to fish yeah I agree right and that's important um to
understand and the other thing is manufacturing and Innovation or R&D
research and development go hand in hand manufacturing is not just doing
repeatedly the same thing to a recipe there's a lot of innovation on the manufacturing floor how how to improve
on the manufacturing process how do you get better uh yield how do you turn the
current generation of Technology a little bit better so that you can offer a newer version of this
Technology based on this current one uh just slight change not big change but
slight change that provide value to your customer how do you do that and tsmc does that a lot every generation they
would come up with Innovation on a manufacturing Flor that turn it into a product that's slightly better than
previous one that brings the customer in you can tell the customer hey for a
small price increase you could get much better uh uh uh performance or Energy
Efficiency or better cost that doesn't require you to go to the next jump
that's value to the customer so the manufacturing and research and
development and Innovation really go hand in hand that's one piece one side
of the coin the other side of the coin is that research and development development has to be really close to
manufacturing otherwise you dreaming up with solutions that not manufacturable
and how do you know what is manufacturable or Not by very being very close to the manufacturing floor so in
other words it is very difficult to have the research developments divorced from
the manufacturing and so if we just say hey
company X just come to Ohio Arizona Texas whatever and build your
FB that's I think just basically isolate
that So-Cal manufacturing from a rest of it it's not going to be successful you got to be cognizant of the
inseparability yes of the two components those are two sides of the same coin those are two sides of the same coin you
cannot just say I only bring manufacturing on Shore and I I don't do them at R&D and that's very important so
in other words all these chips subsidies of 52 billion doll of
chips saxs right 39 billion goes to the manufacturing subsidies and 11 billion go to R&D should be flipped well it may
not be flipped because you takes more eort more money to do the man manufacturing but the importance of
which yeah is the R&D the most important one is the 11 billion if you if you do
that P right yeah I'm saying that not because I'm academic I do research but
if you don't I would even put a greater number for R&D exactly 10 11 right that has to be done right otherwise you'll
get a fish and you ate that fish and then that's it you got to teach the fishermen how to fish absolutely and I
think that is the most important thing and for the fishermen to be successful you need to be able to talk to other
fishermen yeah you know where so they know where the fish is yeah and you got to have a boat yes have a boat yes you
know there has been a lot of rhetoric about FR Shoring reshoring yeah
offshoring right in the last few years but if you take a look at the empirical
evidence call it southeast Asia Capital formation hasn't really
moved up in a meaningful Manner and and I I would speculate or even hypothesize
that a lot of that is because of the lack of investment on a human capital MH
right and and the ownness is really upon us to get our act together you know
there is a desire to reshore there is a desire to relocate but if we're not
ready yes you know yeah it's not going to come yeah opport it hasn't come as
much as we would have hoped for yeah to Capt to capture the opportunity you need
to be prepared and this Talent cultivation is really to be prepared you look at China as an example they
graduate more Engineers than anybody in the world yeah and that's how they got becomes a manufacturing uh Hub of the
world and go to sunen it's unbelievable uh if you want to build something some Hardware you got to be
there yeah they have not only the people but also the infrastructure yeah and the
people is really amazing they they know how to do things and so you need that
investment
[Music]
