#
Despite our bias for bad news, it is hard to deny that we live in the best time in human
#
Two of the reasons for these are technology and globalization.
#
Technology has empowered individuals more than ever before and globalization has connected
#
us to the world with all the benefits that these deeper interconnections carry.
#
So it's kind of ironic that one of the major geopolitical crises of our times has technology
#
at its heart and is created partly by the unintended consequences of globalization.
#
Most of you might vaguely have heard that there's some problem with semiconductors.
#
You would have heard a mention of China, Taiwan, US, COVID and you might have thought that
#
it's some arcane economics or tech problem.
#
The geeks will figure it out.
#
Well, no, it isn't quite like that.
#
There's a crisis brewing that affects the whole world and that has semiconductors at
#
If you care about the state of the world, let me tell you, you should care about semiconductors.
#
Listen to this great conversation to figure out why they matter.
#
Welcome to The Scene and the Unseen, our weekly podcast on economics, politics and
#
Please welcome your host, Amit Verma.
#
Welcome to The Scene and the Unseen.
#
My guests today are Pranay Kutasane and Amira Manchi, who had just written a wonderful book
#
called When the Chips Are Down.
#
Both of them have worked in the semiconductor industry and care deeply about public policy.
#
This is Pranay's 12th appearance on the show and Abhiram makes his debut.
#
Now, what's a funda with semiconductors?
#
Here are two things to note.
#
One, you need semiconductors for almost every piece of technology you use.
#
One modern car can have up to 3000 semiconductor chips in it.
#
Two, semiconductors are deeply complex and the supply chain of a semiconductor chip involves
#
many processes, many countries.
#
An iPhone, for example, has more than 200 suppliers from 43 countries across six continents.
#
And for some of these parts and processes, just one country or even one company can give
#
So if any part of the supply chain breaks down, everything comes to a halt.
#
Well, during COVID, multiple parts of it did break down.
#
And even though COVID is over, we are in a world where the geopolitics is getting kind
#
of risky and many parts of that chain are vulnerable.
#
This is crazy complicated.
#
There's a lot at stake and all the big battles of our time, starting with USA versus China,
#
have a strong semiconductor element to it.
#
Pranay and Abhiram break this down beautifully in the superb book and in this conversation
#
But before we get to it, let's take a quick commercial break.
#
Hey, the music started and this sounds like a commercial, but it isn't.
#
It's a plea from me to check out my latest labor of love, a YouTube show I am co-hosting
#
with my good friend, the brilliant Ajay Shah.
#
We've called it Everything is Everything.
#
Every week, we'll speak for about an hour on things we care about, from the profound
#
to the profane, from the exalted to the everyday.
#
We range widely across subjects and we bring multiple frames with which we try to understand
#
Please join us on our journey and please support us by subscribing to our YouTube channel at
#
youtube.com slash Amit Verma, A-M-I-T-V-A-R-M-A.
#
The show is called Everything is Everything.
#
Please do check it out.
#
Pranay and Abhiram, welcome to The Scene and The Unseen.
#
Great to be back on TSTU.
#
Thank you for having me here.
#
So, you know, I've done many episodes with Pranay and my loyal listeners will be very
#
familiar with his life story and his thinking and his absolutely clear mind, which everyone
#
But Abhiram, the first time I am coming across you in any form, we've never met, etc.
#
So, I'd like to, you know, start by knowing a little bit about you, tell us a bit about
#
your background and so on and so forth.
#
Okay, okay, so I am Abhiram Manchi.
#
I am from Andhra Pradesh, particularly from, I was born in Hyderabad, but my father had
#
So I've been to most parts of India, all over India we've stayed, but primarily my education
#
has been in Tamil Nadu.
#
So right since kindergarten within Tamil Nadu, we've been roaming around, mostly in Chennai
#
and there's another place called Karur, where I stayed for a few years.
#
And my engineering background, I did in electronics and that was from Kastri University, which
#
was again in Tamil Nadu.
#
But meanwhile, my parents have been roaming around and we've stayed in Chennai, Hyderabad,
#
some place near Varanasi, Vishakhapatnam and right now everyone is in Bangalore.
#
My father is retired right now, he's just working as a kind of like a banking consultant,
#
as an informal, I mean, insolvency resolution professional, exactly.
#
And I have one sibling, he is also working in Bangalore, he is working in Aditya Bela
#
Yeah, so that is wonderful.
#
And right now you're in Boston University.
#
Yes, I am currently pursuing my MBA, primarily focused on strategy in Boston University.
#
I just started, I think, two months ago, till then I was also in Bangalore.
#
I was working for a power semiconductor company, which focused on automotive.
#
I worked there for around five years and I was thinking of just continuing that with
#
Tell me a little bit about how you got into semiconductors in the sense of through your
#
education pathway, was it just that, was it kind of serendipitous or were you particularly
#
Tell me a little bit about, you know, how that journey was for you.
#
I would say it was more of luck rather than anything else.
#
So I did my electronics and communication from Shastra and we had multiple different
#
avenues to get into when you're in electronics and one of the advantage of being in electronics
#
is you can actually jump into computer science as well.
#
So that is what most people do.
#
I was kind of passionate about being in electronics, but I couldn't get any role while in college.
#
So on campus placements, I couldn't get anything, so I went into a data analytics company, it's
#
While I was working there, a few of my friends who had actually gotten into semiconductors,
#
they referred me back here.
#
And then I think I stayed in the data company for like four or five months and once I got
#
the chance, I moved back here.
#
And from then, I stayed in the same company.
#
It was Cypress Semiconductors back then, it was a US based company, Silicon Valley based,
#
but then it got acquired by Infineon Semiconductors, that's a German based company.
#
So last three years from 2020 onwards, I had been working for Infineon, but basically the
#
same role, that's the company name changed.
#
And you know, my next question actually ties in with a quote I just read today.
#
I saw this great interview of Ilya Satskever from a few years back.
#
He is of course the original Brutus in this whole open AI rebellion which started, great
#
And he was talking about how he worked in Google, in deep learning.
#
And at one point, in circa 2014, he got an email, he doesn't say from home, but probably
#
from Elon Musk, inviting him to dinner with Elon, Sam Altman and Greg Brockman.
#
And that's where he went and he met them.
#
And he found it really exciting and his reason particularly interested me.
#
And his reason was that, you know, when you work in research, very often you are solving
#
really small problems, you are solving the next small problem and then the next small
#
And you don't get the chance to sit back and to take kind of a broader view of your field
#
or the work or whatever.
#
And he realized that with these guys, he would be able to sit back and take that broader
#
Do not just think about, you know, the next specific narrow problem that AI could solve,
#
but think about AGI and the implications and so on and so forth.
#
And it strikes me that maybe up till a hundred years ago, within science, within any of the
#
academic fields or technical fields, you could have done that, where every specialist could
#
also have taken the broader view.
#
But today, more and more, you get pushed into silos and so on and so forth.
#
And especially, you know, the beautiful way in which you guys have described the semiconductor
#
industry here is how it got more and more specialized and more and more complex and
#
And therefore, it strikes me that if you're actually working, making semiconductors or
#
designing them, you are actually a specialist in a very narrow kind of path and you don't
#
get a chance to take the broader view.
#
And I want to ask you, you know, first you Abhiram, but then Praneya also about, you
#
know, whether the broader view was part of, you know, how you sort of looked at the field
#
because, you know, Abhiram, after all, you did do the GCCP at Takshashila as well.
#
So I'm guessing from that, that your interests are wider.
#
You want to kind of make a difference.
#
So tell me a little bit about how you're thinking about semiconductors sort of evolve.
#
Like, were you a narrow domain specialist to begin with?
#
But then you began to see the bigger picture or were you always a big picture kind of guy
#
just doing things which interested you?
#
Yeah, so specifically talking about the kind of work that I did, it was extremely narrow domain.
#
So it was my work essentially handled supporting third party software to be used by internally in my company.
#
It involved creating like multiple different flows and creating software.
#
So very, very specific work.
#
And I mean, while I was working there, I knew what I was doing.
#
But the issue was that I didn't know where exactly it fit into the overall grander picture.
#
I mean, I was just trying to optimize overall the efficiency of getting a chip out faster.
#
But I couldn't feel the overall broader picture.
#
And it felt that I was just one small cog in the overall wheel.
#
And that is why that is when I started thinking about the broader picture.
#
And in the general sense, I'm very passionate about semiconductors and technology in general.
#
So once I started working there, once you get deep into it, you kind of look at all the aspects of it.
#
And whatever we take for granted here, your mobile phone, your credit card and everything,
#
you kind of see the amount of effort that actually goes into it.
#
So once you see that, you will become very fascinated.
#
And that sends me down into a rabbit hole.
#
I wanted to figure out the industry in general, how it grew.
#
And that pushed me towards the broader picture.
#
I want to know, why is it the way it is?
#
How did this particular kind of architecture in a chip come into place?
#
So kind of tracing back the history, once you go back through history,
#
I think you will end up covering a broad range of topics.
#
And what started out as a very general theme becomes more and more specialized.
#
Like you were also telling, like we also mentioned in the book.
#
So when you go back to the beginning, it wasn't very broad, basic, very simple.
#
And from there, you kind of see how the story evolved.
#
I mean, kind of understanding how the broader picture worked out and how it all became silos.
#
So, yeah, I had an interest in understanding the broader picture,
#
but I started from the narrow down and went up there.
#
Yeah. And one of the beautiful things about your book is that you tackle not just a what,
#
So every stage of the way, including when you describe the science of the chips and so on,
#
you get a sense of the why also.
#
Everything makes sense. Everything is there for a reason.
#
And I would expect that in a geopolitical and economic sense,
#
but even in the scientific sense.
#
And I found that sort of as a layperson, really illuminating.
#
Pranay, what is your sort of take on this?
#
You have worked in semiconductors according to your extremely modest bio for,
#
you know, you've been a chip design engineer into Fortune 500 companies.
#
And towards the end of the book, you talk about how India's strategic strength is now going to be in chip design.
#
So you have a backup career, I guess you can go back to that.
#
But tell me about how you got into semiconductors and if you have any takes on what I just asked Abhiram as well.
#
Yeah, no, yeah, I don't have a career there because it's 10 years ago that I was working there.
#
The world has moved on.
#
But very interesting things.
#
For me, electronics sort of always fascinated like my first fascination was with the TV.
#
I think for many of us it will be like, how wonderful of an appliance that is, right?
#
I mean, and I was always fascinated, like, how is the TV, how does it work?
#
How is it that those signals are reaching us and we are able to see those pictures
#
which are converted into some electrical electronic signals?
#
They travel so far and then they get converted into human beings that you watch on TV.
#
Like, how the hell does this happen?
#
Right. So that was sort of always a fascination.
#
And right after 11, 12, I decided that I will try to understand this and go into electronics.
#
So I took that by choice.
#
And I always wanted to learn and never got very good at it, but I was always fascinated by it.
#
So I always wanted to learn.
#
And you know that the Clark's adage that any sufficiently advanced technology is indistinguishable for magic.
#
For me, it always meant the IC, the chip, the semiconductors.
#
That was the pinnacle of that statement to me.
#
So I was always interested in that angle.
#
But when I joined, after my engineering, I joined a semiconductor company, started doing work there.
#
The good thing was that the firm that I was working at was an MNC, Texas Instruments, and it started work in India.
#
It was the first MNC in Bangalore.
#
So you can literally say even Bangalore is Silicon Valley because the first company was about semiconductors, not about software as such.
#
And it started in 1985.
#
So by the time I joined in 2007, there was 22 years of ecosystem building in that.
#
So the work that was happening there was quite good.
#
It was not just that there was some really old back end work which was being passed over to India, but really good chip level architecture work also happening.
#
So that gave me a chance to meet very interesting people, people who are thinking broader.
#
They were not just thinking about the small little work that they were doing.
#
So that was an interesting experience, meeting people who are writing papers in this or interacting with others.
#
And one thing that struck me particularly was how what we are taught in college is very different from what is actually done in the companies.
#
You know, so I was very interested in that.
#
And that got me, gave me the teaching key to and I said, OK, I want to actually go back to my college.
#
And tell students how the work atmosphere and the learnings there differ from what we are taught in college and the concepts that you will require when you go there.
#
So I was teaching, went back to my college, taught this a couple of times.
#
So I was interested in the broader impact of this area.
#
And then, of course, I changed tracks after seven years.
#
I thought after seven years, I have had enough of this field.
#
I have learned a lot. But now let me try something different.
#
And that's when public policy and Takshashila happened and so on and so forth.
#
So that was sort of my interest in this.
#
But semiconductors, after joining public policy, sort of forgot, right?
#
Like, I mean, that was at the back of my mind.
#
But I got into many, many more interesting things.
#
And but 2019 was when it started changing.
#
So 2019 was so when Trump came in first, there was a restriction against a Chinese company called ZTE in 2017.
#
So basically, they said that American companies can't supply certain kinds of chips to ZTE in 2017.
#
And then the same thing happened in 2019 against Huawei.
#
And when it happened twice, then I thought, OK, there is something interesting growing here
#
that semiconductors, which is like no one used to talk about it.
#
Even when we used to talk about the word technology, no one mentioned the word chips or semiconductors.
#
It was always about software.
#
So this was thought to be a commodity which no one cared about.
#
But suddenly, this was at the forefront.
#
And it was very fascinating for me to understand how something which was considered a commodity is now being talked about in the geopolitical realm.
#
And I was working on foreign policy.
#
So, I mean, this was made for me like semiconductors and geopolitics happening in me.
#
So I said, OK, let's deep dive and understand this better.
#
So since 2019, I wrote first about this in 2019, then tried to convert it into a sort of a course that we taught at Takshashila.
#
That's when Abhiram joined that cohort.
#
And that's when we met.
#
And while in the last slide of the course was that I hope that we can actually write a book on this.
#
Is there anyone who is interested?
#
So Abhiram said, oh, I'm interested.
#
And that's how we started collaborating.
#
By the way, we've met only once in person.
#
So I have met RSJ more number of times than I have met Abhiram.
#
But over the last two and a half years, we've interacted every Monday, 9 p.m.
#
online, discussing the book.
#
And through that collaboration, over the two and a half years, we've been able to write the book.
#
So that's how it's been.
#
Yeah, I find it fascinating.
#
So you've met Abhiram less times than you've met your other co-author, RSJ.
#
And I have met RSJ more than you have met him.
#
So it's like, what's going on here, man?
#
You should kind of go out a little bit more, party shati, meet people.
#
I also like the mention of Texas Instruments, two of the heroes of your book,
#
Morris Chang and Richard Chang, respectively, in Taiwan.
#
And China did interesting things in the semiconductor world.
#
So maybe here's another Texas Instruments alumni who might do things for India.
#
Who is who can who can say?
#
And it's, you know, this is all saying, is it a saying or am I imagining it
#
that you can take a man out of semiconductors, but you can't take semiconductors out of a man?
#
Yeah. And like you said, you must have thought it saying by Amit Verma.
#
Yeah. No, I mean, you must have thought that, fuck, it was semiconductor.
#
I'm out of it. I'm in public policy.
#
And before you know it, hey, it's all semiconductors again.
#
So before we actually start talking about semiconductors,
#
one final question, which is not even so personal, but more about process,
#
the process of writing the book and so on.
#
Like, I find it really fascinating to talk to co-authors and, you know,
#
discuss how they write the books that they do.
#
For example, the first time I read In Service of the Republic by Vijay Kelkar
#
and Ajay Shah, I knew that everything is actually written, written by Ajay
#
because it's exactly his voice, though the ideas are both of them.
#
And it was a very collaborative process.
#
And Ajay described it to me in great detail.
#
Similarly, you and RLJ also kind of chatted a bit about what your process was.
#
How did you two kind of, you know, sort of manage that,
#
especially because one of you was a teacher, in a sense,
#
and the other one was kind of a student.
#
And how do you divide the writing?
#
How do you sort of divide the research and so on and so forth?
#
How does that all of that take shape?
#
Yeah, I'll begin and Abhiram, you can add.
#
But the R process was first, we sort of discussed
#
what will the chapters be like.
#
And Amit, what we initially wanted to do was actually put a lot of
#
this book as it is being developed into the public domain.
#
So we started with an obsidian vault
#
where we started putting all the things that we are discussing.
#
And, you know, this guy, Amit Verma, he has said,
#
perfection is the enemy of production.
#
So we took that very seriously.
#
I don't know if you know that guy.
#
I was quoting Venice Sitapati.
#
He said it to me on my podcast.
#
I really what we were doing is we also had a newsletter.
#
So I was trying to write about this every 10, 15 days.
#
So all this book was sort of one outcome of all the writings
#
that we've been doing over last since 2019.
#
So putting things out into the public domain was a really important part.
#
And then putting this into a shared obsidian vault also helped.
#
So we were collaborating on that.
#
Eventually, we didn't make it public, but we started
#
doing a lot of work and sharing ideas there.
#
Simultaneously, once we then broadly discuss
#
what the structure of the book is going to be,
#
then we started working on the chapters separately.
#
And a lot of history of how the semiconductor industry developed,
#
how the US came into the picture, how South Korea came into the picture,
#
that is all done by Abhiram.
#
He is really interested in the historical element of it.
#
So once he'd written sort of the first chapter,
#
I thought like your voice is much better than mine in this.
#
And I am not good at looking at history and trying to draw
#
a line through various elements.
#
I am not. That's not my strength.
#
So I focused on the framework bit, the frameworks in the last chapter.
#
And trying to understand why India didn't get to where other countries
#
could or what were the reasons why China's first attempts at semiconductors failed.
#
But now they are a significant power in this.
#
So the comparative analysis between USSR, US, China, India,
#
those kinds of insights were what I was interested in.
#
And Abhiram was interested in the historical angle.
#
So we play to our strengths, just like in semiconductors,
#
where there is comparative advantage based specialization.
#
And I'm reading every week a sort of help.
#
So every week we exchange notes that, OK, this is the new thing
#
I have learned in this chapter.
#
And then we would discuss, OK, how this could fit in into our book.
#
And that's how the process sort of progressed.
#
And would the weekly meetings put a pressure on you guys to kind of perform?
#
Because then would you feel that every week that I have to have something,
#
I have to have sat down and done something like Abhiram, especially for you.
#
I guess you haven't written in the past as much as Pranay would have, you know,
#
in the sense not just Pranay's book, but he's a regular newsletter,
#
writer, articles, papers, et cetera, et cetera.
#
In your case, how was that process of firstly finding the discipline
#
to be able to write regularly for a big project like this?
#
And also in terms of finding your writing voice, honing the craft.
#
Were there any models that you looked up to, ki aise likna hai?
#
And by the way, I find the whole book delightful,
#
but especially the historical parts, because I think what makes a book
#
really relatable is the stories in the book.
#
So it really works for me in that sense. So well done.
#
But, you know, how did you arrive?
#
Tell me a bit about your process, Abhiram, of kind of arriving at the voice,
#
at the discipline, at a picture of the book in your head as it formed.
#
OK, first of all, thank you for that.
#
So before the book, I think I had very, very limited writing experience.
#
I think the first time I sat down to write something was while doing the
#
while doing the GCTP course only, where there was one assignment
#
where it was like you have to write something and share it with the public
#
so that, you know, it's not just about learning.
#
It is about disseminating information you have also.
#
That is when I started writing.
#
And I actually I found that I enjoyed the process
#
of doing the research for the writing.
#
Let's say it's a thousand words.
#
It will take me two or three hours to write.
#
But the research behind that, it would lead me down into so many rabbit holes.
#
You know, like you say, you started double clicking on everything
#
and kind of understood the whole process.
#
But I would generally start out with one small theme
#
whatever I was writing about.
#
For example, we take one chapter.
#
I break down the chapter into different parts, basically where it started.
#
Let's say that is the theme of it.
#
And then I'll start researching overall.
#
And I'll collect huge amounts of information on that.
#
And like in obsidian wards or any other place, I'll just store all that information.
#
And one thing I would do is I just take a step back and read or read everything
#
that I have researched and see if there is a narrative forming
#
which connects to the overall arc that we are trying to see.
#
Or if it doesn't connect, how do we have to change the arc to actually,
#
you know, match what it what is there?
#
That was my overall process.
#
Just start with something small and have evidence,
#
research-backed evidence over it and start building it.
#
And, you know, eventually reach a story which matches what we want.
#
Other than that, the weekly meeting actually helped me a lot
#
because other than the course that I have done,
#
I don't have much experience in geopolitics.
#
And my only knowledge was how a chip works internally.
#
So I knew the industry as such.
#
I knew what was coming from where you worked there for five years.
#
I can understand where your chips are going, where your work is going.
#
But I had to build the rhythm to look at it from a geopolitical lens.
#
And Pranay helped me out a lot on that aspect.
#
Every week, we used to discuss.
#
I mean, depending on the chapter, we used to discuss what the country has done
#
and why it is doing exactly that, what would be their motivations.
#
This was something that didn't come naturally to me.
#
And all these discussions helped.
#
And obviously, because we had it every week, it put me in a place
#
where I forced myself to write a lot and research a lot.
#
There would be a lot of rewrites, obviously.
#
And, you know, I had to make sure I had enough content to show
#
what I had done for the week.
#
And, you know, there must be some significant progress over the week.
#
So it gave me a kind of schedule and trying to think about it
#
from a broader perspective, which was what Pranay was really good at.
#
He would help me, you know, break apart a problem into different parts
#
and looking at it from different lenses.
#
That was something that it was not coming naturally to me, basically.
#
My final question before I get to semiconductors is actually about knowledge
#
in the sense that, of course, you guys use Obsidian, which is an awesome software.
#
I use Roam Research, which is kind of similar with this nested entries
#
and it has bidirectional linking as well.
#
So these are great tools for knowledge management.
#
I often say Roam changed the way I work completely.
#
But I'm also thinking in a sense of not just assimilating
#
and organizing the knowledge that you get, but also growing that knowledge
#
by constantly iterating and doing things with it.
#
For example, you know, I found it very interesting what you said
#
that you created an Obsidian vault first as a way of thinking aloud,
#
even if you never made it public.
#
But I've heard from other writers that when they write a book,
#
they'll actually first workshop it in newsletters, you know.
#
And for some, it might be the other way around.
#
That I had an episode with Luke Burgess, who wrote the book
#
Wanting on Rene Girard's Philosophy of Mimetic Desire.
#
And what Luke told me was that he started his newsletter after the book
#
because he wanted to continue down these ideas.
#
And what I'm thinking is that often we get trapped by this tyranny of form
#
where we imagine that a book is the ultimate form of knowledge
#
and you've done the book and that's your last word on the subject and you move on.
#
But actually, knowledge is constantly evolving both in your heads and in the real world.
#
And all these are different parts of a puzzle, whereas a book may be a static view
#
where you sum up your view of what the world is like
#
and where things have come at a particular point in time.
#
A newsletter can be a continuous, continued debate with yourself
#
week after week after week as you keep updating your knowledge.
#
If one was to do podcasts, you could say that those conversations would be
#
conversations with the world and always teasing and fleshing out the subject as well.
#
So is that something that you've thought about?
#
Because Prane, I think, you know, you've been quite innovative
#
in terms of how you assimilate knowledge and how you kind of do all of this.
#
So what are your and even within Takshashila, you know, a bunch of you
#
do a bunch of different interesting things.
#
So what are sort of your thoughts on this, that when we talk about like,
#
you know, a key skill in the new world which is coming is, you know, being able to
#
like, I think the fundamental key skill in the world that is coming
#
is the ability to learn new things because the world is changing so fast.
#
And part of that has to do with assimilating knowledge and always constantly
#
needing it around and, you know, working with it.
#
So what are what are your thoughts on this question to both of you, really?
#
Yeah, I think, like you said, I think book writing for me is too much of a high
#
So you don't need you don't want to put a lot of weight in that.
#
So and like writing is a way to learn.
#
And I have started saying that we should not have reading list, but we
#
should have writing lists.
#
And because you're continuously writing every week in a newsletter or it can be
#
in any way, or you're doing a podcast on it or you're writing op-eds on it, that
#
process helps you tackle different things happening within the same domain.
#
So even in semiconductors, the amount of things that have been happening since
#
2019 is just, you know, breathless things happening every week.
#
There are three, four announcements and people jumping at it from many
#
different directions, sometimes people with who have not had any exposure
#
So you might not have those knobs to see whether a news is actually
#
worth focusing on or not.
#
So everyone is jumping at it as something which is really important.
#
And every single event is earth shattering, which it really isn't.
#
But because of that, I thought writing continuously was really important.
#
So just putting small posts out, we had a newsletter called techno politics.
#
So that was the one I started in parallel to anticipating the unintended.
#
So I started writing the semiconductor bits there and just putting things out.
#
The good thing is in Substack, you can also create sections.
#
So gradually we started using that newsletter to talk about many things in
#
high technology geopolitics, but I could create a section on semiconductors
#
and only focus on that.
#
So that was one really helpful thing.
#
And the other thing was teaching.
#
So the good thing with Takshashila is I could do teaching and writing both.
#
So, and teaching helped.
#
So we converted this into a sort of a course again, that was a way for me to
#
learn and where else can I get people like Abhiram who are working in that
#
industry to say that, you know, this is bullshit or I'm not explaining it well.
#
So that helped and talking to a large number of self-selected, interested
#
people helped me sort of improve iteratively my teaching and the
#
elements that went into it.
#
So that writing and teaching combo doing that over and over again, and putting
#
things out into the public domain creates a very nice feedback loop, which
#
helped me develop this, right.
#
So book was sort of not in my mind till 2020 when I started teaching.
#
And after a couple of rounds of teaching, in fact, one webinar I had
#
done for Ajay's research group.
#
So when, and another webinar for Ahmedabad University.
#
And that's when someone else suggested to me that, hey, this is worth writing
#
a book on, said, okay, who will be interested?
#
Like, okay, I'm talking about this.
#
But who will be interested?
#
But they said, okay, this, I felt that there are people who are interested in
#
it, given that it was appearing in the news a lot more.
#
And that's when we thought, okay, maybe there's a book in here.
#
It's, you can convert this knowledge into a book.
#
So that was our process.
#
It was iterative, get small outputs as soon as possible out into the public.
#
And then maybe a book just happened along the way.
#
You know, I looked at some of the other books and semiconductors while I was
#
doing this, some of which you mentioned and looked at them in the sense, didn't
#
read them, just kind of glanced at them.
#
And my impression was that actually your book is, if not the only book, at least
#
the best book on the subject tackling it overall, because the rest might be, you
#
know, too Western centric or they're focused on one particular piece of the
#
puzzle and all of that.
#
And this seems to me to be like the definitive book.
#
It's, it's, it's a great book.
#
It's a world-class book.
#
Is it published only in India or are you going to get international publishers?
#
Do you feel that there is a sort of, there is some kind of blog there?
#
What's, what are your thoughts?
#
Yeah, I think right now it's published in India only.
#
And of course the Kindle version is available everywhere, but it depends on the
#
sales as our publishers say, but the idea here is, yeah, there have been a couple of
#
other books on this topic.
#
I think while we were just finished the first draft, the book Chip War came out.
#
That was again a book written by an economic historian, Chris Miller.
#
So that's also quite good.
#
It focuses a lot on individuals and their stories who went into semiconductors,
#
whereas we focus more on the institutional angle and why nation states become good at
#
innovation and some don't.
#
Those kinds of questions motivated us, but yeah, so I think we will see a lot more on
#
this field in the next coming years, given that with AI chips, etc., this area is
#
going to heat up even more.
#
But yeah, I think the book is relevant to anyone across the world.
#
It's not specifically related to India only, but we came at it from an Indian
#
interest, national interest perspective and an Indian angle largely, you know.
#
So the questions for us in India are always, you know, we have tried this when we
#
can be a space and nuclear power.
#
Why can't be a semiconductor superpower?
#
What were the problems?
#
And also I was like quite pissed off when we see and attended many of these seminars
#
where once the government announced this policy on semiconductors that we want to do
#
something that and a lot of those conferences ever like a lot of people from outside and
#
Indians asking white men about why can't should India build a fab or not or a
#
semiconductor fabrication facility?
#
Now, I am always open for ideas from outside.
#
Of course, we should be open to.
#
But this is a question that has to be decided by the government, society, markets
#
You can't ask, should ABC be done by India or not?
#
You can ask, what can we learn from your experience or other countries experience?
#
So I was like, you know, many of these books which are written have very little from an
#
So we need to make that point.
#
And India is a crucial note in this semiconductor supply chain.
#
I mean, both of us are examples, not that we have some great patrons or anything, but
#
we have been a part of this industry to understand that India does play a role in this.
#
So we thought, you know, there should be an Indian voice and an Indian perspective on
#
this. Chitwar, for example, mentioned the word India, a grand total of two times the
#
word and nothing about what India does.
#
So it was also that, you know, let's put this out from an Indian angle.
#
And there's a lot of things happening in the Indian industry, in the Indian government
#
on this sector. So we should put a view out.
#
So that is how we came to this.
#
Yeah, I mean, just for the sake of the listeners, there are like eight chapters in the
#
book and one of them focuses on India.
#
And it's sort of a great chapter.
#
Before we get to the contents of the book, just a final sort of topical question almost,
#
you know, which is about these rumors that came out after Sam Altman was no longer
#
CEO for about three days or whatever that period of time was, there were these rumors
#
that he had gone to the Middle East to raise money from funds for a chip fabrication
#
company or a chip design company where AI would try to design chips and take on
#
Nvidia. I mean, that's how that was kind of being positioned the way it was written
#
about. And your book, of course, has a lot about the dominance of Nvidia and so on.
#
Not a lot, but, you know, you explain it quite concisely.
#
So, you know, just from a layman perspective, you know, what is the take on that?
#
Because it would seem to me that, you know, one of the parts of your book which we'll
#
get to right after this is about the great innovation that was happening at Bell Labs
#
for a long, long time. And it seems to me that there is sort of a glorious summer of
#
that same kind of innovation happening in certain parts of Silicon Valley and certainly
#
in OpenAI. So, how dramatic is that?
#
Is it a big quantum thing or is it just not a big deal at all?
#
Right. So, just briefly on that Sam Altman, the rumor that we've heard of, but I think
#
there are two different angles here.
#
One is about AI chips and then making chips with AI.
#
These are two different things.
#
I think the second angle is what they were hinting at, that he's going to make chips
#
with AI and because he's making them with AI, they'll be so good that they'll compete
#
with NVIDIA. That was a broad gist of what I could get.
#
So, yeah, basically saying we'll make chips with AI, which will be the AI chips that that
#
can beat what NVIDIA does.
#
So, they're both two different things.
#
So, I think NVIDIA, a lot of what OpenAI does for the pre-training phase and the training
#
phase, they need specialized chips, which are called AI chips in common parlance.
#
These are basically chips which are optimized to take a lot of data and do parallel
#
computations. So, you can't do that with a general purpose Intel processor and all that.
#
So, NVIDIA actually made those for graphics.
#
So, there were the graphic processor units which were used for playing games, but it
#
eventually got used in AI kind of workloads and it was quite beneficial.
#
So, NVIDIA's stock really rose.
#
So, that is one angle and I'm sure OpenAI, given that this is what costs a lot, one training
#
run can cost you anywhere around 300k dollars, etc.
#
So, given that it is a high cost, I'm sure OpenAI would be looking at how do you minimize
#
that cost and currently, I guess, Microsoft is probably taking a lot of that expense,
#
but they would want to reduce this and probably in-house chip development could be one way out
#
because now they also know what are the exact kind of workloads that happen?
#
How would a transformer algorithm actually work on this kind of chip?
#
So, maybe they have an insight which NVIDIA doesn't because they are working at a layer
#
above NVIDIA and hence they might be wanting to do this.
#
The second thing using AI to make chips itself, whether it is an AI chip or any other chip,
#
there's a lot of work happening on this, but it has not yet reached a stage where it can replace,
#
get the human out of the loop.
#
But I know that DARPA, for example, has a plan to reduce, get the human out of the loop,
#
like basically from architecture to getting a chip, they want to do that in a 24-hour workflow.
#
So, a lot of things are happening, but these are not at that level where they can do really,
#
So, it will start probably in a couple of years for really small basic chips,
#
but we are not yet there where it can replace humans.
#
So, that's where my understanding is.
#
Because we were talking about designing chips using AI, that is kind of my point as well.
#
So, designing in general, it's not like just one company does it.
#
You design different blocks for a chip and there are places where you get third-party
#
companies to actually design that block for you and there are IP licensing companies like ARM
#
from where you can buy these particular blocks and you just place it in those chips.
#
All these are, at least the cutting-edge ones, these are all proprietary.
#
I don't think there is enough cutting-edge information available
#
open source for AI to actually train on and replicate it on its own.
#
So, like Pranay was saying, it might come up in a few years,
#
but the open source information that is available on chip design,
#
it is actually at least 10 years old and it is only for the older legacy chips,
#
which are not as complicated.
#
So, it will be difficult for AI to actually design chips.
#
Basically, they don't have the Wikipedia equivalent here easily available.
#
So, it is going to be tough.
#
I mean, Wikipedia equivalent would be needed by generative AI,
#
but if you are talking machine learning, I am guessing then you don't need that.
#
It is like AlphaGo teaching itself to play chess.
#
So, I then wonder hypothetically, could you teach yourself to design chips without having a
#
need for previous human iterations being available to you?
#
No, but even AlphaGo teaching itself to play chess,
#
the previous stages of that would have been where AlphaGo actually learned by
#
looking at various kinds of games.
#
How it worked in the case of AlphaGo was that the rules of the game were wired in.
#
That is the only thing that was programmed in.
#
Then it played itself for 24 hours and it played thousands of games against itself,
#
And during this time, even those tens of thousands of training games
#
that it plays against itself are really fascinating,
#
because it almost mirrors the evolution of human understanding of chess,
#
where certain openings come into fashion, they go out of fashion.
#
It starts playing the French defense and at one point it stops.
#
Then at one point, you know, Vladimir Kramnik made the Berlin fashionable again
#
at the start of the century in his match against Kasparov.
#
It discovers a Berlin and then it goes beyond that human evolution
#
and it starts doing things and humans enough to figure out why.
#
So, there was no LLM involved.
#
It was just the rules of the game and it played itself.
#
But obviously, chip designing is different because you're not playing a game.
#
It's just a completely different thing.
#
You know, let's kind of get to the narrative of your book.
#
And before we start talking about semiconductors per se,
#
I love the way that you set a backdrop for it in your book.
#
And I want to talk about that backdrop a bit,
#
because one of the things that you underscored and that comes across starkly
#
is how contingent everything is, how everything is almost dependent
#
on certain circumstances working out one way or the other.
#
You know, you started your book with this excellent quote from,
#
you started one of the chapters rather, not your book,
#
with this excellent quote from Bill Gates about how if you put him in a time machine,
#
he'll want to go back to 1947 in Bell Labs to see the magic that was happening there.
#
And Bell Labs, you know, which was a research arm of AT&T,
#
was as you've pointed out and as all of us who've looked at techno was,
#
you know, so many innovations came out of there.
#
It was the crucible of so much modern technology that we take for granted.
#
But what particularly interested me was the fact that number one,
#
there was tons of innovation happening in America at the time.
#
And two, that Bell Labs was the center of a lot of this
#
innovation while other companies necessarily weren't.
#
So tell me a little bit about that.
#
You know, why was that the case?
#
Why was that specific place buzzing with ideas and innovation and great minds?
#
And, you know, because in a sense,
#
a semiconductor revolution really comes out of that ferment.
#
When you look at Bell Labs or when you look at US as a whole,
#
I think during that point of time,
#
it was a melting pot of all different cultures.
#
And it was the first time a lot of people were actually coming together
#
and getting educated in the same place,
#
trying to understand from each other what was going on.
#
So that we felt was an important aspect where people from different mindsets
#
were actually working together on something with a united goal.
#
And what was that goal was what was currently happening in the 60s and 70s.
#
It was the kind of a Cold War era.
#
And the prime goal for the US was of being one step better than Russia
#
in technological superiority.
#
So you have all these different brilliant minds working together
#
in, let's say, one building or one place,
#
hitting ideas of each other.
#
And the one thing that was important in Bell Labs was that nothing was restricted.
#
If you were working in Bell Labs,
#
you could just directly go to any of the faculty there
#
or any of the senior researchers there
#
and spend some time researching that.
#
If you look at 1947, the time period Bill Gates was talking about,
#
that is when the building block of the transistor was invented.
#
And the transistor, if you look at it,
#
it was three different people from three different backgrounds.
#
It was just not physicists that were there.
#
People had a chemical background and people had superconductivity,
#
semi-conductivity background, and solid-state physics was required.
#
These were three different people with one goal
#
of finding out a device which should make it easier
#
to transmit over long distances.
#
So communication was the focus.
#
And at that point of time, AT&T was making long-distance telephony,
#
but you could go from New York to probably Chicago.
#
That is by when the signal would get so bad
#
that it wasn't audible anymore.
#
They were trying to figure out a proper way to boost the signals.
#
And what was before transistors and what we have now are vacuum tubes and all.
#
And they were just not that great to go long-distance telephony.
#
So they had one common goal in mind.
#
And the additional value that Bell Labs was providing was that
#
it was consolidated in a way that
#
Bell Labs, AT&T, its parent company, would provide the telephones,
#
would provide the overall distribution network,
#
would provide the baseband antennas.
#
And the government was also of the mind that
#
if this whole network already exists,
#
it doesn't make sense for another company to start from scratch on that.
#
So AT&T had the monopoly on communication.
#
They did not have to worry about competition
#
and the funding that they had.
#
They used to put it into projects of different varieties
#
from not only just on telephony,
#
but trying to understand the overall landscape
#
of how they can push electronics further more.
#
That was the golden period where there was enough funding.
#
They didn't have to worry about competition and this was growing.
#
But over time, it kind of went down and the company became complacent
#
and innovation was going down.
#
That we see in a longer time frame.
#
But at the beginning, this was, I would say,
#
one of the important key factors,
#
the lack of having to worry about constantly innovating
#
at a short-term mindset.
#
They could go long-term,
#
they could actually fund five years, ten years down the line.
#
The transistor, I think the project started
#
probably ten years before it was released.
#
So this long-term mindset could only come with that kind of foresight.
#
Experiments which could be afforded to fail,
#
which is what most companies do now.
#
We see Google kind of adopting the same kind of mindset.
#
But this was one of a kind thing that was happening then.
#
So this is the overall landscape for the company that was going on.
#
But if you look at US as a whole,
#
initially, at least for electronics and semiconductors,
#
as I said, the backdrop was the Cold War era
#
and the military was the only client for all the semiconductor companies.
#
And there were a few rules that these companies had to adhere to
#
to make sure that they had contracts from the military.
#
And two particular things that were highlighted were that
#
the military had the right to distribute information
#
produced under the contract.
#
So even if Bell Labs was working with the military,
#
they would provide quarterly reports
#
which are available to every other company.
#
So Texas Instruments could just take a look at what Bell Labs was doing
#
and understand the cutting edge of what was going on,
#
understand the science behind it
#
and probably could come up with something different.
#
So in that way, there was some kind of competition that was going on.
#
But even though Bell Labs had to manipulate,
#
the competition started brewing at that stage.
#
And also, there was this particular case
#
where at least 10 years down the line,
#
there was this antitrust suit filed saying that
#
Bell Labs had a complete monopoly and something had to be done about it.
#
And one of the main things to come out of that was
#
the patents that Bell Labs had, they had to be provided for free of cost.
#
So within a fortnight, probably around 7,000 patents became free.
#
And it was seen that the follow-on innovation
#
that was seen by other companies using Bell Labs patents
#
to come up with something new was around 17 to 20 percent higher
#
than what Bell Labs was actually doing.
#
So overall, I would say that Bell Labs had the funding to start out.
#
They were kind of getting complacent.
#
And with the antitrust suit and with military forcing a few clauses,
#
competition was brought back up.
#
They had the chance to put a foot into the industry
#
and they started innovating.
#
And these were small startups.
#
Some of them were the forefathers of companies like Intel.
#
So basically, it was this kind of storyline
#
where eventually competition started picking up
#
and it was not enough for the military to sustain
#
and they had to go commercial and started selling to other electronics companies.
#
So the growth, I would say, would be initial funding at the beginning
#
and then growing to other companies and going commercial.
#
So, I mean, before we get into this,
#
I think we should also briefly discuss
#
what is a semiconductor and a transistor really briefly, I guess.
#
I was actually laying the grounds for the environment in which it came up
#
and I'll come to the science and what is a semiconductor now.
#
But let me add to that itself.
#
I think I totally agree what you said
#
that there were a lot of contingent factors in this.
#
For example, monopoly, as we know from economics,
#
actually would lead to stagnation.
#
There's no incentive to do anything new.
#
But there were a host of factors which happened at the same time,
#
which led to also companies investing in research
#
rather than just not doing a lot of long-term thinking.
#
So one itself, you had this period after the World War
#
where there were a lot of great minds coming to the US.
#
So that was one contingent factor.
#
And after that, it's not just Bell Labs.
#
You also had organizations like RAND, etc.,
#
which came up in the 50s, 60s, etc.
#
The second factor was, like he said, there was this Cold War which started.
#
And the Cold War was also a movement where
#
people who study the politics of innovation
#
call this as creative insecurity.
#
This is a term used by Mark Taylor who studies this a lot.
#
So creative insecurity is the feeling that
#
the sum total of external economic and military threats
#
are greater than the internal political and economic divisions.
#
The more the creative insecurity,
#
the more that nation-states tend to invest in R&D and innovation.
#
So this is again a counter-intuitive thing for me,
#
someone who doesn't hold governments in high regards on many of these things.
#
But you can think of it from a very value-neutral way that
#
finally R&D investment involves opportunity cost
#
and no nation-state would want to do that
#
unless and until there's some exogenous shock or requirement.
#
I mean, a democracy would rather spend money
#
on making the life standards of the individuals and the people
#
who can't afford a lot of things better.
#
That would be the default option.
#
But when nation-states started entering into this sort of the Cold War period,
#
there was this creative insecurity which increased a lot.
#
And there was this Sputnik movement.
#
There were people in the US seeing the Sputnik satellite going around
#
with their naked eye, you could see that.
#
That sent a lot of and of course other political communism versus capitalism.
#
That debate, that narrative was so strong that nation-states,
#
especially the US started thinking that we have to do something very different.
#
There was a whole restructuring of the research ecosystems as well.
#
And of course, you had also the Vannevar Bush statement, etc. in the 1945.
#
Then you had changes come in terms of because of the Cold War in the 50s.
#
You also had these ideas of restructuring, starting of the ARPA which became DARPA later on.
#
And all those, also NASA, etc., all those start from that ferment.
#
So I think creative insecurity has something to do with making sure that the monopoly that
#
some companies enjoyed didn't get translated into those companies just
#
becoming static and not improving.
#
And they in fact led to a lot of investment in trying to get better fund long-term projects,
#
which was an interesting thing.
#
I am still not been able to reconcile that given my prayers that this should have actually led to
#
stultifying rather than growing.
#
And in fact, you will see a lot in the semiconductor industry.
#
Government intervention has played a role,
#
especially in the capital heavy parts of the segment,
#
which again is tough for me to reconcile, but that's the way it is.
#
Yeah, I mean, one of the things that fascinates me as both of you would know is
#
unintended consequences of stuff.
#
I recently did an episode with Timur Kurran where he described,
#
where he was talking about why the Middle East had stagnated for a thousand years
#
since about 1000 AD till the mid 19th century when they started picking it up again.
#
And the reason was that in circa 1018, and I won't go into details here.
#
You can listen to my episode with Timur Kurran or there's an everything is everything episode
#
on YouTube where I've discussed it with Ajay Shah as well.
#
But essentially they had incredibly progressive social laws
#
and incredibly good institutional innovations, particularly the wakf.
#
And they were far ahead of Europe in that regard.
#
But unintended consequences meant that economically,
#
they were stultified in the state back for centuries,
#
which then kind of made them socially backward as well.
#
And it's not something that you could have sat down a priority and figured out,
#
you know, from first principles.
#
And similarly, the fact that, you know, AT&T was a monopoly for so long.
#
And you and I have been taught correctly that monopolies are a bad thing.
#
They are bad for consumers, especially because consumers have to pay more.
#
There is less innovation, et cetera, et cetera.
#
But what it did was that AT&T got outsized profits because of this,
#
and they managed to put that into Bell Labs.
#
And their incentives there weren't short term incentives
#
that bring me something I can take to market next year.
#
But they were like, you know, not just research for the long term,
#
but research in other fields, which we have nothing to kind of do with.
#
And the other thing where I think you and I are on the same page in terms of cognitive dissonance
#
You know, it is not the business of government to be in business, as we know.
#
But state spending has often propelled so much innovation.
#
You mentioned NASA, which came up with anti-corrosion coating
#
and, you know, all kinds of commonplace things like scratch resistant coating we need now.
#
DARPA came up with ARPANET, which turned into the internet,
#
created an early version of GPS, a graphical user interface,
#
a computer mouse traces its origins back.
#
And as you point out, state spending during wars
#
have led to the invention of radar, microwave espionage,
#
cell technology, encryption and decryption algorithms.
#
And I read that sentence out from your book itself, which also doesn't compute.
#
And obviously, there's a seen and the unseen factor here.
#
There is a counterfactual of if the state hadn't spent on this,
#
what would it have spent on?
#
And we don't know that.
#
But the fact is, all these innovations did happen with their massive positive externalities.
#
And all of these, in a sense, were unintended.
#
Many of these were unintended.
#
They just happened to come out of the research
#
that you're sort of, you know, doing in sort of that area.
#
You have a beautiful part in your book where you talk about the Sputnik crisis,
#
about how, you know, after, you know, 4th October 1957, NBC radio broadcast,
#
this beep and all of America went into panic because this was a Sputnik
#
and you could actually see it at times with the naked eye.
#
And there was a sort of technological panic that, oh, shit, the USSR is so far ahead of us.
#
And this is a peak of the Cold War.
#
And therefore, this spurs so much sort of spending.
#
And I have sort of an anecdote from the other side, which is the Russian side,
#
which is, you know, one of the reasons why unintended consequence
#
chess became such a hugely popular sport that sometime in the 1930s,
#
a gentleman named Nikolai Krilenko, who was, you know, head of the Soviet chess association.
#
And he was purged by Stalin in 1938.
#
But for a while, he was a pretty big guy.
#
And at one point, he said that the USSR must sort of focus on chess
#
because everything that they did was part of this battle
#
to prove the ideological superiority of the Soviet Union over the US.
#
And chess was one way in which they could get a leg up.
#
And Krilenko's words were, quote,
#
We must finish once and for all with the neutrality of chess.
#
We must condemn once and for all the formula chess for the sake of chess,
#
like the formula art for us, our art sake.
#
We must organize shock brigades of chess players
#
and begin immediate realization of a five-year plan for chess.
#
And this is also, I guess, you know, coming out of that sort of push towards innovation
#
because you're panicking about others kind of going ahead.
#
We'll come back to sort of geopolitics in this period later.
#
But let's go back to that fundamental innovation of, you know,
#
the transistor to begin with and then how that goes ahead and what else happens.
#
And, you know, you spoke disparagingly of vacuum tubes,
#
but as you have pointed out in your book,
#
you know, four industries that dominate the world today,
#
long distance telephony, radio, gramophones and movies,
#
you know, came because of vacuum tubes.
#
So, you know, do not mock them.
#
They did their job before they were replaced.
#
So tell me a little bit about, you know, what is a conductor,
#
what is a semiconductor?
#
Take me through a bit of the evolution of this beautiful thing
#
that, you know, allows us to talk to each other right now.
#
Yeah, before that, Amit, just one angle I wanted to add
#
that cognitive dissonance the way I have sort of resolved
#
is that actually nine out of ten times that strategy would fail,
#
you know, and we've seen that.
#
In fact, we have many examples of how USA tried to do that.
#
India tried to do that.
#
It failed because all of them thought we'll make a city where this will happen
#
and we'll have a government run PSU which will make semiconductors
#
and it didn't work out.
#
But there were some unique things which happened in the US ecosystem
#
where they were thinking of AT&T not just as a government company.
#
You are still thinking of getting people from different places,
#
thinking long term and investing in R&D.
#
I mean, in 1945, US was not as prosperous as it is today.
#
And yet, what were the reasons why they were thinking about that
#
and not other things is a fascinating topic in itself.
#
A lot of people try to understand that.
#
And there are many interesting examples.
#
I mean, Denmark versus Sweden.
#
Both are GDP per capita wise great,
#
but Denmark doesn't do so well in innovation,
#
whereas Sweden and Nokia and Sweden and Norway and all do.
#
Those are all fascinating questions.
#
I think a lot of work to be done.
#
But one way we can understand is most of the times
#
government run programs would fail.
#
And I'll double down on that and point out that
#
when we talk about government programs that work, it's survivorship bias.
#
Like you said, I think 95% to 99% of all programs failed.
#
You have plenty of stuff in your book about the USSR, India and China,
#
also failed in some areas where they tried to kind of push this.
#
So state intervention, as we have seen through almost 80 years of Indian history,
#
is generally almost extremely likely to fail.
#
But there are these kind of spectacular examples of unintended consequences.
#
State interventions also have multitudes.
#
So let's begin by just talking briefly about the semiconductors.
#
And so largely semiconductor as the word suggests,
#
it's a material where the conductivity is between an insulator and a conductor.
#
So the idea is a conductor electron flow in that is so high that you can't control it.
#
It's perfectly useless if you want to control it.
#
And in insulators, the flow of electrons will be so slow or not at all that,
#
again, you can't control the flow of it.
#
So ideas, if you can control the flow of electrons in one part of the circuit or the other,
#
you can convert it into things like a switch,
#
or you can convert it into an amplifier and so on and so forth.
#
That you can't do in an insulator or in a conductor.
#
So semiconductors, largely silicon or germanium, if you see the periodic table,
#
all those which fall under, have four electrons in the outermost orbit,
#
those are things which are called as semiconductors.
#
And the idea was a specific arrangement of this treated semiconductor
#
can be made into a basic unit called the transistor.
#
And on which then the word transistor comes for the radio.
#
But the idea that it actually meant this arrangement of treated semiconductors,
#
which allowed this conductivity of electrons to happen in the most optimum way.
#
We don't need to go into how it is made and all that.
#
But the idea was that the superpower of the transistor was that you can control the flow
#
of electrons in one part of the circuit by controlling the current in another part of the
#
circuit. Now, it sounds very basic, but this is really a superpower.
#
Because you can convert this arrangement into a switch, you can convert it into
#
amplifier because you can control the current flowing in another part by
#
controlling some current in another part of the circuit.
#
These things really were a massive innovation.
#
And you could do that at a fraction of the cost that you could do by using a vacuum tube.
#
As you were saying, vacuum tubes could do all these things, but to handle a basic addition
#
process, you would require the amount of space of rooms worth and you would require electricity,
#
which would take a significant proportion of the electricity of a city at that time.
#
Those were the kinds of things you would require to do basic things.
#
But here you have something which is as small as it can be held in your hand.
#
At that point of time, it was still quite big.
#
And it could do all the things that you wanted to do at vacuum tubes.
#
You are basically transitioning from electrical to electronics through this period.
#
You are then dealing with really small currents or dealing with really small
#
signals and yet you are able to achieve the functionality that you wanted.
#
This was really a magical thing, the transistor.
#
And still it didn't solve the things because even though you could have transistors,
#
the connections between those transistors itself after you have, let's say you have
#
100 transistors and you have to connect between them, the wires itself would become really,
#
And beyond a point, the complexity of that circuit would grow so large that you can
#
connect things incorrectly and the failure rates would be high.
#
So in the 1950s was when the next step jump happened, when something called the integrated
#
circuit came into being.
#
And those are the square black things you see if you open your phone or someone throws
#
your phone, those ICs are basically collections of transistors which are connected in many
#
different ways to achieve specific functionalities.
#
And that IC was the next step jump because the connections between the transistors etc.
#
are inbuilt into it and you can achieve various kinds of functionalities within a circuit.
#
And that's why it's called an integrated circuit within a package.
#
So now this was really a fascinating thing.
#
And with that integrated circuit, the complexity of connecting transistors etc.
#
became simplified and you could then the aim became that how many transistors can you pack
#
in within this IC to achieve more and more complex circuits.
#
And the fascinating part of this is that in 1965, Gordon Moore, after whom this Moore's
#
law is named, which was just a vague prediction at that time.
#
He just said that in Fairchild semiconductors where he was working that the number of transistors
#
they were seeing in that IC was doubling roughly about every two years.
#
And doubling means you can achieve more functionality and you can do more complex things using that
#
So this was what he said in 1965.
#
And he also said that ICs will lead to such wonders as home computers, automatic controls
#
for automobiles and personal portable communications equipment.
#
Guess the word mobile didn't strike him that way back then.
#
But he said this in 1965 and the entire technological and economic revolution of the
#
industry has been such that it has become almost a law in the sense that this has held true
#
One IC now per millimeter square like in your Apple iPhone will have 134 million transistors.
#
Whereas when he made the statement and one IC had 64 transistors.
#
So just imagine how wonderful and brilliant this has been and the progress has been.
#
This is old fable about the king is happy with some guy and he says,
#
tell me what you want, whatever you want, I'll give it to you.
#
And he says, give me a single grain on the first square of a chessboard and then double
#
it with every other square.
#
And then it goes to some kind of completely mad place.
#
And as you point out, the current iPhone uses, you know, has about 10,000 times, you know,
#
nodes which are 10,000 times smaller than the ones used in the 1970s.
#
And I have sort of I'll take a quick tangential leap as I tend to do on this podcast and ask
#
another question, which is that one of the sort of something that comes along with Moore's law
#
is, you know, not only the fact that not only the doubling as it were, as it happens,
#
and we'll talk about Rock's law later, but not only the doubling of ICs every two years,
#
but also, you know, the cost of things therefore coming down substantially because of that.
#
And I think about that sometimes in the context of GDP, like I have done an episode on GDP
#
ages ago with Rajeshwari Sengupta.
#
But one common sort of complaint I have against the GDP is that there are many things that
#
And one of those is this, that you can see that I bought a laptop in 2023, which might
#
And I could have bought a laptop for 60,000 in 1999.
#
But the productivity that I get out of my current laptop is orders of magnitude more
#
than the one I would have gotten in 1999, that there are certain kinds of value that
#
GDP cannot capture because, you know, just monetary transactions cannot capture.
#
And they also have a lot to do with the well-being of a society.
#
So, you know, do you have any tangential thoughts on this question being a policy person yourself?
#
But the GDP in this case, the laptop case will include the additional consumption and
#
the production that has happened because of the because of the fact that it's this
#
technological leap has improved the productivity of the person.
#
No, but some of it could be joy to me.
#
Some of it could just be joy to me.
#
Like if I was, for example, in the 1980s or early 90s, if I had to collect music, you
#
know, I would have to scramble so hard to somehow get a mixtape together.
#
And most of the artists I want, I would not be able to get them in India, etc, etc.
#
Today, we can listen to whatever we want, read whatever we want, watch whatever we want
#
at the click of a mouse.
#
And how do we put a value to that?
#
And GDP is not meant to do that.
#
So you're holding GDP to way too high standards.
#
And it should be, right?
#
I can imagine GDP coming in front of me and saying, boss, what the fuck, this is not my
#
How much will you ask for?
#
How much will you ask for?
#
You know, it's so weaponized and people take it so seriously.
#
And I think that so much of human progress is actually incalculable through, you know,
#
crude methodologies like this.
#
One of the sort of things that you write about in your book is you detail out not just a
#
technological revolution, which you gave me a summary of just now, but also the sort of
#
economic revolution that happened in semiconductors, which led to this incredible supply chain,
#
which is so complicated, so miraculous, so mind blowing.
#
Like, you know, you've got a lot about the iPhone processor chip and how that's an illustration
#
of the kind of complexity and the kind of interwoven sort of connections that we have
#
And I particularly love that because one of my favorite essays of all time is I pencil
#
by Leonard Reed, you know, where he talks about how all the different components of a pencil
#
come from so many places that it is impossible for any one person to actually make a pencil.
#
There are so many countries involved.
#
So many the supply chain is so incredible.
#
And it's all the voluntary actions of people coming together that leads to this miraculous
#
thing called the pencil.
#
And this is an audio podcast, but we can see each other right now.
#
So I will let my listeners know that I am holding up what you could think is a pencil,
#
but it's actually a pen.
#
But I'm getting the point across.
#
So tell me a little bit about this miraculous supply chain, because at some level, this
#
incredible supply chain is at the crux of our geopolitical problems today as well.
#
So take kind of take me through and it's up to you whether you want to use the iPhone
#
processor chip example or not.
#
I absolutely loved it because it is so detailed and so mind blowing.
#
All the stages of designing a chip, all the materials and everything.
#
But, you know, take me through how the supply chain evolved.
#
What are the different components where at one point you might begin with one single
#
party saying that, hey, you know, I'll do everything and I'll make a semiconductor.
#
And today that is such a ludicrous thought that there are so many disparate parts to
#
And it is so complicated.
#
And so many links in the chain are so vulnerable that it becomes a huge humanitarian risk.
#
And therefore a geopolitical issue.
#
So, you know, tell me a bit about that.
#
So to get a basic understanding of the overall supply chain, we can take a look at the iPhone.
#
But what I would say is I would call it like a relay race as such.
#
You know, one part from somewhere in the world gets done and that gets moved to another
#
country where the next part gets done that gets moved to other.
#
And so it's kind of a chain that goes throughout the world.
#
And I think around multiple different countries have to be together.
#
And it is the case that most of the times there is only one single country in the world
#
that can do that part of the chain, particularly.
#
So I would classify the overall semiconductor supply chain into three different parts.
#
First, starting off with what is called the design.
#
You create the blueprint of the chip that you define what all it has to do.
#
If you talk about an iPhone chip, it has a specific set of functions that it has to do.
#
It has to be able to browse the internet.
#
It has to be able to receive signals.
#
It has to be able to play sounds.
#
And there are so many things that it does.
#
So someone, some team at Apple will sit and define what the next or whatever,
#
And they will start deciding on what all functionality that this chip has to perform.
#
And then a particular set of computer science languages, which are used,
#
very log or VSDL where you will particularly define what the chip has to do.
#
This is just a blueprint that is given out.
#
We think of creating one chip as building a city.
#
Basically, there are multiple different blocks for a city.
#
There is probably something like a town square.
#
And there is a sanitation department.
#
There is a water department.
#
There are roads connecting each of them.
#
And there are multiple cities that are there.
#
And there are some ways to connect to other cities as well.
#
All these kinds of functionality has to be built into the chip.
#
And the chip designers are the architects of that, basically.
#
So once this chip design is actually done,
#
it goes on to chip manufacturing.
#
This is where companies come in and it might be the same company
#
or it might be a different third-party company,
#
which only takes the blueprint and starts manufacturing the chip.
#
If the same company does it,
#
it is called an integrated device manufacturer, IDM.
#
Companies like Intel, they design their own chips
#
and they manufacture their own chips.
#
Whereas companies that don't do it are called fabless semiconductors.
#
A fab is what we call a manufacturing unit.
#
So companies like Nvidia, Apple, they do not manufacture their own chips.
#
So these are called fabless companies.
#
And there are some companies which only do the manufacturing,
#
which do not care about what kind of design it is.
#
They'll just take the blueprint and they'll start making chips out of it.
#
And companies like TSMC from Taiwan, they are the biggest players there.
#
All of Apple's iPhone chips are made by TSMC.
#
All of them all over the world.
#
If you look at manufacturing,
#
it also involves getting raw materials from around the world.
#
You can get it from multiple different places.
#
I mean, the US is one of the good places.
#
Australia, you can get sand from.
#
And you need to get all these chip making machines.
#
Those come from probably Japan or the Netherlands.
#
And then you start making your own chip.
#
I mean, you start making the chip
#
using the blueprint provided by any of these companies.
#
And what you get is a small square.
#
Probably it's the size of your nail,
#
which contains the entire blueprint of the chip.
#
And it is this silicon at that stage.
#
It is not protected at all.
#
That is when the third part of the supply chain happens,
#
where this particular piece of silicon is taken to assembly and packaging.
#
That is when you open the phone,
#
you have this plastic black color casing and all that.
#
That is when the assembly and packaging is done.
#
The silicon part of it is packaged.
#
And it is made sure that all the connections are right to be.
#
The chip has to communicate with your charger.
#
That connection should be open.
#
All those kind of stuff is then defined there
#
and basically made sure that the chip will be sturdy
#
in different kinds of temperatures.
#
It doesn't get damaged very easily.
#
These are the three main parts of the supply chain, I would say.
#
Chip design, chip manufacturing, and assembly.
#
And there are other peripheral parts
#
where there is the research that happens.
#
We were talking about how doubles the number of transistors
#
are placed each and every time, every year or once in two years.
#
The amount of research that is done is in the billions.
#
And the semiconductor industry is only second to pharma
#
in terms of the amount of part of its revenue that it spends on research R&D.
#
Roughly 15% of revenues of semiconductor companies go back to R&D.
#
And other than that, there are specific research columns as well,
#
which solely focus on one single part of the ecosystem.
#
Let's say you want to make machines better.
#
There is a whole research institute that works
#
on just making the manufacturing machines better and all that.
#
This whole supply chain overall ties back together.
#
And it's one part at a time.
#
Chip design happens, then manufacturing, and then assembly.
#
So this is the supply chain as such.
#
Actually, I add a lot to this about why the economic angle and how does that come in.
#
That is what fascinated me also, Amit,
#
like coming at it from a policy lens and an economic reasoning lens.
#
A lot of people who write on this have focused on the technological advancement,
#
but you can teach a lot of economics through the semiconductor supply chain,
#
like how the invisible hand operates or how the comparative advantage
#
based specialization operates and all that.
#
And that is what you see in this industry as well.
#
So if we begin the story again from where there were every two years,
#
you were doubling the content in a chip and hence you were able to do more.
#
But this was not coming at zero cost.
#
You have to keep improving your manufacturing.
#
If you want to double the amount of transistors in the same area,
#
that means each transistor has to reduce in size.
#
Now, in order to reduce in size, you will need to have machines
#
which are able to make smaller and smaller transistors.
#
And that is the capital intensive process,
#
which Abhiram was calling as the fabrication step.
#
And it gets really complicated beyond the time.
#
For example, some transistors now which are made are just a few electrons thick.
#
So it's that small, that precise.
#
How do you do that in few atoms thick?
#
How do you do that really quickly and at high precision and fidelity?
#
So once this started happening where transistors kept getting smaller,
#
you can't just use the same old machines to start manufacturing,
#
let's say a smaller transistor.
#
You need to do huge amount of capital investment to see
#
how the transistor at a smaller size can work equally well
#
because there will be some leakage effects and all that.
#
So every stage when the transistor becomes smaller,
#
you need to do newer in R&D, understand how the transistor works, etc.
#
So it is not just an incremental change, it is a step change.
#
And that's why the R&D requirements and capital investments are very, very high.
#
Now, once for earlier, one company thought that I can do all this on my own.
#
So in the beginning, it was that way.
#
All companies were IDMs.
#
Everyone did the design, manufacturing and packaging on their own.
#
But as this Moore's law sort of a prediction started becoming a law
#
and everyone started chasing this, the flip side of this to this
#
was the Rock's law, which said that sort of the capital cost
#
for manufacturing also doubles every four years.
#
So this became very tough.
#
So every all the companies which were initially doing this
#
at some point of time, they said, hey, every two years or I can't
#
keep investing in making smaller and smaller transistors
#
and because the cost is going to double every four years.
#
So this led to a comparative advantage based specialization
#
where some companies thought came up and said, why are you doing this investment?
#
Let me be the contract manufacturer.
#
So you, whether you're Nvidia or or you are AMD doesn't matter.
#
I am just a contract manufacturer.
#
You just give me your blueprint and I will mass produce it for you.
#
So this was a massive productivity gain.
#
So now companies which said I want to start making chips
#
didn't have to bother at all about investing in this capital
#
repeated capital investments every four years, which is doubling.
#
They didn't have to do that.
#
So they started focusing only on getting the architecture right.
#
So that's why fabulous companies, you know, boomed a lot.
#
So AMD came up and video came up and all those fabulous revolution sort of happened.
#
So they got really good at making architectures.
#
Whereas there was a small, there were some companies
#
which only invested in contract manufacturing.
#
So that's where Taiwan picks up.
#
So TSMC is now a very well known name.
#
Three years ago, apart from people in the semiconductor industry, no one discussed it.
#
But now a lot of it is discussed because it was in the background.
#
It was just a contract manufacturer, which got really, really good at
#
making the recipe that it makes it possible to manufacture these small transistors,
#
which are what five, six nanometer with those at high fidelity.
#
And that's what their expertise was.
#
So this comparative advantage based specialization led to massive gains.
#
You had capital investments being done by some companies,
#
some the design is a human capital intensive process.
#
You just need large number of engineers, et cetera, and architecture design work.
#
So that kept happening.
#
There was also an environmental angle.
#
So for example, at that point of time, this chip making process was really polluting
#
and people in the US didn't want that.
#
So Taiwan, Taiwanese company said, we are okay to take this cost because we want to
#
develop this industry and we don't have the expertise that is required in architecture
#
and design, but we have the expertise in manufacturing because they were doing
#
electronics level assembly there.
#
So we will take this cost and develop this.
#
So that was one interesting angle again.
#
So some nation states and some companies there decided that we will take this risk
#
in another area and develop.
#
Also remember the costs in the US labor costs in the US were very high labor costs
#
So that was another advantage for packaging.
#
It's a labor intensive process.
#
So you again, you need large amount of labor, which was available at cheaper cost there.
#
So the price also worked as a beautiful signal.
#
It led to different nation states and different companies focusing on the areas that they
#
And eventually you had this supply chain, which is a fantastic innovation in most normal
#
So no one even bothered about where the chip comes until three years ago.
#
Because the fact that it worked so perfectly, you didn't need to bother about where your
#
Every chip people have, the US semiconductor industry association has done a lot of studies.
#
Every chip makes almost four trips around the globe during this process because of the
#
relay race kind of thing that Abhiram mentioned.
#
Some companies are doing some part of it, etc.
#
But it just worked absolutely fine.
#
So in that sense, it is huge humanitarian benefits.
#
But just the fact that now because it is being seen, its nature of the supply chain also
#
makes it useful to be used as a geopolitical instrument, whether it should be used or not
#
is a different question.
#
I would say actually it is a, it is a net benefit to every human on this earth.
#
So it should, actually the goal should be resilience and every nation state should work
#
But that's not how geopolitics works.
#
So because of the supply chain where some companies are doing one small part of it,
#
the way it can be deployed geopolitically is that if you have control over that company,
#
you can deny access to another country.
#
And that's what has happened.
#
We'll talk more about it later.
#
But that is what the nature of the supply chain is a fantastic human achievement.
#
But it also makes it prone to be used as a geopolitical instrument.
#
And that's what has been happening.
#
And what I kind of loved about your book is how clearly it sort of lays out the diversity
#
of inputs that just goes into a chip.
#
And it's such a triumph of globalization.
#
Like you point out how the silicon could come from China, the design and the software
#
tools could come from the US, the primary manufacturing could be done in Taiwan, equipment
#
from Netherlands, Japan, the US and Germany could go into it, Southeast Asian countries
#
could assemble it, India or China could actually put the chip into the iPhone and so on and
#
And at one point you write, quote, a mere glance at Apple iPhone supplier list shows
#
that it has more than 200 suppliers from over 43 countries across six continents, breaking
#
down each component into its basic form tells us that 60 to 70 elements from a pool of 80
#
usable elements in the periodic table are used in this entire process.
#
And obviously, I think everyone would agree that this is a huge net positive.
#
It's happening because of freedom of trade.
#
And that if right at the start of the process, countries were to dig in and talk about nationalism
#
and we will support national champions and we will, you know, reduce our exposure to
#
X country or Y country.
#
Such a miracle may not have happened, but as we have found in these geopolitical times
#
and especially during COVID when these supply chains were kind of hit hard and there was,
#
you know, so much pandemonium during the pandemic, can you call it pandemonium, whatever, that,
#
you know, that this became an issue.
#
And I want to ask you this broad question now, you mentioned resilience and you pointed
#
out that the reason we got into the mess we did after COVID was because we had ignored
#
or rather the semiconductor industry had ignored and I think quite logically and rationally
#
had ignored resilience over specialization and cost effectiveness where they are all
#
maximizing for their, you know, their individual utility, the invisible hand is at play, they
#
are making perfectly rational decisions, they are using comparative advantage, but an unintended
#
consequence of it because of the complexity of semiconductor chips is that unlike in potato
#
chips, you know, you don't have resilience in the sense ki ek jaga supply bandh ho gaya
#
to kahin aur shuru ho jayega that if any part of that extremely complicated chain breaks
#
down everything downstream is affected and suddenly the world comes to a halt like the
#
classic illustration of that is obviously, you know, automobile manufacturers particularly
#
from Europe getting completely screwed after COVID because they couldn't get the, you know,
#
the semiconductors they needed and often a car would have, I think you pointed out about
#
300 of them sometimes, you know, and you simply couldn't deliver on that because all the demand
#
was taken up by consumer electronics in the chips that they needed. So what do you feel
#
about this larger question? How should we think about it? Resilience on the one hand,
#
which is something that nation states would want versus, you know, specialization and,
#
you know, the kind of value that brings, which, you know, is what would naturally arise because
#
of the rational choices made by people. Yeah, it's like at one level, basically,
#
you're applying this thinking at multiple levels and at one level, because of globalization,
#
everyone's benefiting and it's a great positive, but it only works when every nation state agrees
#
that globalization is in everyone's benefit and this kind of semiconductor industry is sort of
#
a common good which needs to be protected or at least held as a net positive. But when that
#
assumption itself gets questioned, that is when you enter the current situation, right? So until
#
this time, it was okay, but now geopolitically, there have been challenges and US and China's
#
relationship has worsened. And if you look at China's technology stack, it is only in a lot
#
of areas, it has caught up with US to some extent, but in semiconductors, it is really behind and it
#
is largely dependent on many parts of the supply chain from other countries. So that's where sort
#
of the US has chosen to attack it. Now, if you I'm sure in the US strategy cell, they would have
#
been a thought that this will lead to loss and downstream consequences for their own consumers
#
and their own companies and yet they have chosen this as an option. So they have chosen geopolitics
#
over the net gains that everyone has. So this is how it has operated. I would also say that there
#
were sort of three things for due to which it became so important and so globally relevant.
#
One is the geopolitics, which is US-China relations, also because Taiwan is a central
#
node of this industry and you have this period where Taiwan-China relations are also worsening,
#
right? So there is a lot of fear in the US and in many countries that what happens if for some
#
reason China-Taiwan relations worsen or if there is an invasion of Taiwan, right? What happens
#
to the fact that there are many crucial factories in Taiwan? And that is one important element
#
because remember, for example, if for the advanced jets that US is flying, the chip is still probably
#
getting manufactured in Taiwan, finally, right? So what happens if China invades Taiwan for one
#
reason or the other? So that is the second geopolitical reason why a lot of countries
#
started thinking about it. The second one is the geoeconomic imperative, which is illustrated by
#
the example that you said during the pandemic, but there are also concentration risks in the
#
supply chain, right? So what happens if one company just goes down for other reasons, for let's say,
#
investors don't have confidence in that company or it is malfunctioned, right? So that leads to
#
downstream consequences again. Or there can be even geographical risk, right? Just imagine there
#
is drought in Taiwan, which has happened in the past, or if there is some earthquakes in Japan
#
and there are certain industries which are supplying very specialized materials,
#
again, there is a shock, right? So how do you sort of overcome that? So this is the second one.
#
And the third one is the technological imperative itself, that you need semiconductors for any
#
other advancement, like you are talking about AI chips, or you are talking about 6G communications,
#
you will depend on advancements in new architectures and new chips, right? So all three things have sort
#
of combined and there is this unique concoction at this moment, because of which it has become
#
globally salient and that's why I call it metacritical, it is not just critical.
#
And that is what has led to the current moment. So I would say that actually just those companies
#
diversifying into comparative advantage based specialization was really great. But we have
#
this unique moment where all these three streams have combined and that made this sort of go in
#
directions where no one had anticipated. And now nation states are saying that, okay, now we need
#
to think of resilience, we can't just think about comparative advantage based specialization,
#
we need other ways. Now that doesn't mean just making everything back home, it might just mean
#
that maybe you need to have a set of partner countries in which you do comparative advantage
#
based specialization instead of earlier where it was whether you are an adversary or not,
#
didn't matter as long as all the consumers across the world are benefiting from it.
#
But because this industry will also lead to significant changes in other technologies
#
and other technology is perceived to be a bigger part of national power going ahead by many
#
nation states, whether we agree it or not, that is driving this fear that you need to be careful
#
where this industry is, even though it might not be in my home turf, it should be in a friendly
#
nation state. So that's the change. So, you know, after the break, we'll
#
dig into what your book does so well, the descriptive nature of geopolitics. But before
#
we get into that, I want to ask you a normative question quickly before the break, like throughout
#
your book, you've, you know, you've spoken about points of view that are essentially countries
#
points of view. And, you know, and as you know, I think that there is a danger always in thinking
#
like that a country, of course, will think like that. But individuals and you know, when we think
#
of societies and economy, sometimes that can be a little bit of a trap, for example, you know,
#
and this is something I discussed in an episode of Everything is Everything recently with Ajay
#
Shah, we linked that from the show notes, that when people talk of something like a trade deficit,
#
you know, as if it's a bad thing now, a fiscal deficit is a bad thing, a trade deficit, there's
#
nothing bad in it because countries don't trade with countries, individuals and individual firms
#
trade with each other and every trade is to mutual benefit and people are always better off.
#
And it doesn't, you know, matter where the balance is. For example, I have a trade deficit with my
#
domestic help, I keep paying her and she doesn't ever pass any money on to me. And that's perfectly
#
fine. It's a it's a positive sum game. We are both mutually benefiting. And you know, and when I
#
think about like one theme across your book is how different countries at different times have
#
used tools like tariffs. Now, one can understand why a country goes in for a tariff. But a tariff
#
is always essentially in my view, a redistribution of wealth from the poor to the rich, consumers
#
suffer, they get a worse price or worse, they get a higher price or worse quality, and certain
#
interest groups of local manufacturers kind of benefit. Now, there are other unintended
#
consequences which might be good, which might be bad, whatever. And I understand that in your book,
#
you're not recommending that people think about it from the country point of view. But how does
#
one kind of reconcile this? Because at one level, if I put on the hat of Oh, I am an Indian, then I
#
will read certain things and say, Yes, India should do this. Of course, we should have tariffs. But on
#
the other hand, if I just think as a human being and don't think about these boundaries, my entire
#
thinking kind of changes completely. So how is and I'm sure this is something that you have grappled
#
with as well, because on the one hand, you are deeply interested in economics where you get that
#
individual point of view I'm talking about. At the same time, you're interested in geopolitics and
#
foreign policy, which is countries looking at the world from the country's point of view.
#
And a nation state could easily argue that look, it's prisoners dilemma, as long as other nation
#
states are thinking of their own interest, I have to do the same. If all of us were suddenly saying
#
good of humanity, let people trade, let people flourish, then it wouldn't matter. But that is
#
never the case. So this is again another area of cognitive dissonance with me, where, you know,
#
your book has so many examples of, for example, Ronald Reagan, who is supposed to be a president
#
who is so good for freedom and all that. But every time you mention him in the book, it's in the
#
context of him doing something that is actually against individual freedom. And for the national
#
interests, get tariff lagao, you know, saw percent tariff lagao on DRAM chips or whatever from Japan
#
because they're catching us, etc, etc. I don't remember the specific examples of book is full
#
of them. So how do you reconcile this particular cognitive dissonance?
#
Yeah, it's tough to reconcile. But the way I see it is see in the when I'm talking about domestic
#
relations, liberalism and liberal values are really the guiding light, right? And all the
#
things that you said. But when you're talking about the nature of the international system,
#
there a lot of realism applies, you know, so a lot of things are for fortunately or unfortunately,
#
nation states are contesting against other nation states. And there is a fundamental
#
dissonance there. But that's the way that system is, you know, so it doesn't matter whether what
#
I want or not. But the idea is that just imagine, right, so US and China, for example, now, you
#
based on our principles, US sort of is the more closer one to our values, interest,
#
and all that it so you would want us to confront China, right in one way or the other.
#
But that also means if you do anything to China, it will lead to some downstream consequences for
#
Indian consumers, because China is so globally integrated in every manufacturing element that
#
if there is even if there's a decline in the economic growth there, there will be consequences
#
and negative consequences to us. So which which one do you choose, you know, which is your which
#
is the option to choose here, it is it is a tough thing to reconcile. But for me, it's always that
#
when you're talking about international relations, you look at interest, and you there are there is
#
something called a national interest. And that national interest could very well say that
#
well being of all my citizens is the national interest, you know, national interest doesn't
#
need to be territorial integrity and all that. But as long as you're thinking in terms of my
#
national interest is well being of my citizens and then start from that point, you can merge
#
realism and liberalism in the you can sort of draw a bridge between them in the domestic versus
#
international realm. But yeah, there's dissonance always exists. And there are hard trade offs that
#
you have to make some at some points. Multitudes, multitudes, no man is an island. Let's take a
#
quick break on the other side of the break, we'll, you know, dig a lot more into the fascinating
#
geopolitics and how we are all going to hell. So after a quick break.
#
There was a time when Pune used to be called the Oxford of the East. It was a hub of social movements
#
of educational institutions of intellectual ferment. But since the seat of government is in
#
Delhi, all the policymaking happens there. Well, Pune is back. This episode is sponsored by the Pune
#
Public Policy Festival. The Pune Public Policy Festival is an important step towards re-establishing
#
the city and the center of intellectual discourse. On the 19th and 20th of January 2024, Pune's
#
brightest experts of business and policy will interact with some of the most eminent economists
#
and thinkers of the country. This particular event is about the trade offs that India will
#
face in the next decade. Trade offs between growth and equality, development and environment,
#
convenience in technology and privacy. Professionals and bureaucrats, academics and students,
#
insiders and outside experts will gather together to discuss the shaping of the world. For more
#
details and to register, head on over to www.pppf.in. This will also be linked from the show notes.
#
The Pune Public Policy Festival 2024, a historic city building towards a brave new world.
#
Welcome back to The Scene in the Unseen. I'm chatting with Pranay Kotasune and Abhiram Manchi
#
about their excellent book about the chips being down. But you know, while we were away for the
#
break, Pranay thought and thought and he, you know, he couldn't, he was turning and twisting
#
because there was something he had forgotten to say in an answer to a previous question. But
#
there is always a second chance and that chance has come. He wanted to make a point about the
#
trade deficits and I had made the point that it's not a problem from the individual point of view,
#
but you have something to add to that as well. Yeah. So the cognitive dissonance that we were
#
talking about, I don't face that in the case of trade deficit specifically. And the reason why
#
I was saying is actually trade deficit can be justified on national interest grounds as well,
#
not just from an individual grounds, because even if you think the unit as a nation state,
#
now if there is a, you're running a trade deficit with respect to China or say any other country,
#
what it means is that you are buying a lot more from China and which means there are companies
#
in China and individuals in China who are dependent on you and your market, right? So in that sense,
#
you actually have a leverage and that is what China has used to get better technologies,
#
etcetera as well. They call it the TMFT, trade markets for technology strategy, right? So
#
trade deficit often is actually a strength. So if you saw between China and Australia,
#
for example, Australia is to export lots of things to China and China was such an overwhelming buyer
#
of things that it could utilize it as geopolitical instrument to say to Australia that hey,
#
we are not going to buy your wine and lobsters and that actually made things difficult for
#
Australia and not for China. So trade deficit is actually, it's fine even from a national
#
interest point of view. And that's why I see views that ban Chinese companies setting up in India
#
is actually really a stupid move because if those companies are set up here, the risk is theirs,
#
not ours. And this is something that Swaminathan Iyer had also written in a very nice article
#
many months ago. But the risk is theirs, their products are being sold here. We have the leverage,
#
not them. But when you think from a very short term perspective, many people start thinking any
#
trade deficit with another country is actually a problem. It really is not. You can think of it as
#
leverage. And I particularly, you know, feel the pain in the case of banning TikTok. Like in the
#
same month where it was banned, I had launched this course. I only taught one cohort of it called
#
TikTok in Indian society. And people talk about WhatsApp University. The course was actually
#
conducted on WhatsApp because I had to send hundreds of vertical videos to illustrate different
#
points I was making. And the easiest way to do that was on WhatsApp. You can't really do that on
#
Zoom. But WhatsApp, you just get everything in the group and then you scroll at leisure. And I
#
thought it was a huge net benefit to Indians, especially Indians in small towns and villages,
#
that combination of ubiquitous geo broadband and TikTok was simply mind blowing. And somehow,
#
I think even today, you know, Instagram reels and YouTube shorts haven't quite taken that space
#
allowed for that same kind of expression. Yeah. And I think of it even again, from a national
#
interest point of view, just think TikTok is global. Lots of people in the world are using it.
#
So even if the Indian state didn't have to think about individuals, let's say they were thinking
#
only geopolitics wise, you have lost your side of the story from the entire platform globally.
#
So for example, a person in the US, when they are thinking about, let's say something happens
#
between India and Pakistan. And TikTok is the primary mode of consumption for a lot of people
#
across the world. Which side of the story are they going to see right now, right? Like they're going
#
to only see the Pakistan story now, right? Unless they have also banned it, I don't know. But we are
#
they are not going to see the Indian side of the story. So there is an individual loss, of course,
#
like that one individual sitting in some village in Maharashtra could have been reaching across the
#
world, not just in India. So that is definitely a loss. And there is a loss from a national
#
interest perspective as well. So it doesn't make sense to me. There are views made that, okay,
#
it had some algorithms which things were being used by China in other ways. But there are other
#
ways to get around that. You don't need to ban it. That's a total loss. Wise words, May. Wise
#
words like yours percolate everywhere. And I should invite you on the scene and the unseen
#
sometime. Oh, but wait, okay. So let's continue down our geopolitical route. And to understand
#
the current geopolitical crisis, I think we need to do what you've sort of done in your book so well,
#
that go into the history of each country and its own relationship with technology and semiconductors
#
and the role that it has played there and so on and so forth. And I guess the US would be a good
#
place to start because it all started there. And as you pointed out, the US actually have been
#
part of two sort of geopolitical battles in which tech played a major part in the past,
#
one during the Cold War and one again in the 1980s. So, you know, and this is really like the
#
third time. So, you know, give me a sort of a potted history of America's relationship
#
with semiconductors and the role that it has played in its politics and geopolitics.
#
Yeah, so we were, we kind of alluded to the Cold War part of it where we were talking about
#
the Sputnik crisis. So basically, for the race for technological superiority, when
#
Russia was the first, USSR was the first to launch a satellite, there was a collective gasp overall
#
in the US. And the immediate next step that government took at that time was, you know,
#
start looking at its own programs, its own space programs. And then there was a funding increase
#
of over 100 million in education and R&D. And also at the same time, they would also start funding
#
NASA and DARPA and all, which, as we've already discussed, had a lot of higher magnitude effects
#
overall for the society as a whole. Yeah, I'm not saying that they had the plan to make the
#
society better in general, and they started doing all this. They had their own motives to
#
start working on technological superiority, and it ended up eventually helping out
#
everyone as such. So this kind of, the creative insecurity that Pranay was talking about,
#
this plays, I think we feel this plays a major role in making sure that the overall society,
#
you know, benefits from whatever technological advancements are there that are happening right
#
now. And other than that, I think what the US has gone on to do, eventually, that we need to focus on
#
is once it was able to get a technological superiority in semiconductors, which is what
#
its focus was on. And the initial Apollo missions also ended up using semiconductors instead of
#
vacuum tubes as such. And they look minuscule, they're nothing compared to what we have now,
#
but they were a huge step up when you look at it at that time. The Apollo mission and the first
#
moon landing had proper chips being used. And that is one of the major reasons for
#
USA's one-up over Russia. And eventually Russia tried to replicate it. Russia tried creating a
#
whole city, Zelenograd, it was called a semiconductor city. But they had their
#
incentives slightly misaligned where the focus was not on innovation, but the focus was on trying to
#
imitate what US had done. So the scientists at USSR would be given chips from instruments or
#
Fairchild or something, and they would be asked to replicate what this was doing.
#
So the culture of innovation as such was not propagated. And that would be one of the failing
#
points in the USSR. And the funding that was also given was also uneven. Different governments would
#
place different priorities on how much money should be given to semiconductors. And given that
#
it was only funded by the government and there was no marketplace as such where companies could
#
invest on its own, that was also a big blow for the growth of the industry in Russia.
#
Which is why we feel the healthy competition and moving to commercial in the US actually helped
#
out a lot in that part of time. The second one, the second technological fight that we were
#
discussing would be the US and Japan. We call it the calculator wars. I mean, for the cold war,
#
we all know it was an ideological conflict. They had to make sure either capitalism or communism,
#
which was on top and that was it. But for this, basically US and Japan were allied to a large
#
extent and US had actually helped Japan set up its own economy and helped fund the initial
#
growth of electronics in Japan. They provided free access to knowledge transfer, cutting edge
#
technology to make sure it could be replicated in Japan easily. They provided initial seed funding
#
and there were a lot of joint ventures that were happening at that time. And Japan was able to
#
quickly pick up and catch up to where US was at that point of time in memory chips.
#
So if you look at chips in general, there are two main differentiations. One is called logic chips
#
and one is called memory chips. Logic chips do all your operations, your addition, subtraction and
#
whatever your laptop is able to do. And memory chips store the memory. So memory chips were only
#
raised back in the 70s and 80s because they were easy to manufacture. They were not technically
#
too difficult to design also. So Japan got into making memory chips as such. And because
#
Japan was pretty good at manufacturing, they were able to optimize hard and the overall manufacturing
#
capability of Japan and the yield of chips that Japan was able to produce was over the next four
#
or five years was significantly much higher than what US was able to do. And in that sense,
#
Japan was able to sell chips for people in the US than the US firms could. And we were all talking
#
about the trade deficit that was there. And there was a trade deficit between US and Japan. And that
#
is where US kind of rigged out and they were thinking of what ways we can manage this trade
#
deficit. And one of the first things that the Ronald Reagan government did was
#
put price floors and set up tariffs as such. It was more focused on appeasing the semiconductor
#
giants that were lobbying hard to Ronald Reagan because you had companies like Intel and Micron
#
and Texas Instruments which were fighting hard saying, we are doing the best we can but Japan
#
is just dumping their chips onto our land for prices that we cannot even think of dropping our
#
chips to. And the other fact was that it was very difficult for American companies to actually
#
get a foothold in the Japanese market. I mean, logically it made sense because Japan chips were
#
much higher quality and cheaper. The penetration that Japan had consequently the penetration that
#
America had was much lesser than that. So this was another factor to say that Japan was doing some
#
kind of shady practices probably to get that. And the first thing was tariffs were set up.
#
And the second thing was that the US mandated 20 percent market share for American companies in
#
Japan. So the effects that these had on both the countries were that the prices of the Japanese
#
chips also went up and American chips were also no priced at an equal rate but the consumer lost out
#
in the US. And in Japan as well, the US chips were getting sold but those were not of the same
#
quality and those were much more expensively priced. So the consumer there was also losing out.
#
I would say the follow-on effect of that was because Japan was losing out on such a huge
#
customer because they were pricing at the same time, South Korea would come in and start pricing
#
chips lower and it was able to get a dominant market share in the memory chip segment.
#
Eventually both the US and Japan lost out in memory chips because they were not able to
#
compete with the prices that South Korea was doing. And the overall the effect that we see was that
#
there were some technical issues that were happening on both sides. And one of the
#
advantages that we got from that was South Korea coming in and the economy of South Korea developing
#
piggybacking on semiconductors. Samsung was just doing normal electronics. By then it moved to
#
semiconductors and the other effects that we could also see were US starting out with consortium of
#
sorts. All the major US players, all major semiconductor players in the US at that time,
#
you have your intellectual instruments, national instruments and fair child everyone, they walked
#
into a consortium and they were able to get funding from the government. There was an
#
industrial policy where the government would provide 100 million dollars every year for five
#
years to combat Japan's significant rise and significant market share in the memory model.
#
The result of this was kind of mixed I would say. There were some innovations that took place but
#
overall it seemed like it was money given to these companies to make sure they could price
#
their chips cheaper for a longer term. So that was one of the unintended consequences that you
#
could see at that time. The policy did not work as it was intended to. And at the same time
#
the other side of that would be because Japanese companies were doing so well,
#
they were actually acquiring Intel and this is when IBM got scared about what would happen if
#
its major player was in Japanese hands and they took a minority stake in Intel to help them
#
supply logic chips. The logic chips are much more complicated in nature and take significant
#
amount of R&D and it's not as simple as manufacturing memory chips. So Intel's focus
#
on logic chips helped them make the first chip for a general computer and that grew and grew
#
and you had your mobile telephony and you had your mobiles coming from that. So straight technological
#
fight was not an ideological fight but it was more of gaining market share and putting your
#
nation's importance first and the unintended consequences were the fall of memory chips in
#
both Japan and US but you saw that logic chips came about and you saw that South Korea came about
#
but there are points to make on both sides but we feel there was an overall technological
#
improvement that we could see in the US because of this conflict that arose. If I could add Amit,
#
there's a common like a macro picture if we take here. Many things start in the US and then there
#
is some technology transfer to other countries and that is also held by the post-World War
#
world order where trade is sort of seen as a positive and also US has all these allies
#
in East Asia. So these countries like Japan, Taiwan, South Korea, they get technology transfer
#
very easily. US companies are easily investing, there are no export controls as such for these
#
countries. So these countries companies in these countries have some collaboration with the US
#
company learn things and then they steadily improve their capabilities and eventually catch
#
up in some area and that sort of threatens US back home. So this all this industrial policy
#
thing that you're seeing now is not the first time. In Japan's case as Abhiram said we had the same
#
way where US government invested a lot for their local semiconductor industry.
#
But as again we see the advantage with the US was that all that industrial policy didn't succeed in
#
the way they wanted to but in fact it eventually led companies like Intel to focus on semiconductors
#
and focus on developing logic chips and that led to what it is today. So really not that what was
#
what the government policy intended which wanted to substitute Japan's dominance in memory chips
#
that didn't happen but Intel did get better in some other things. So it has been amazing to see
#
how US has constantly been able to re-innovate itself. Maybe their ecosystem is such that they
#
have been able to get things on track despite or in spite of government actions.
#
And you know at the start of the episode we were talking about how contingent everything
#
is and I love this little bit where you speak about post-war Japan. Like of course you lay
#
out the history of the 100 years before World War II and the Meiji restoration and now Japan
#
is a rising power and then suddenly the war destroys them and you know America come in with
#
all these sort of terms and at this point you write about their housing crisis. You know you
#
point out that they had a huge housing deficit in 1945 they had a shortfall of 4.2 million units
#
they lacked 3 million in 1955 so the Japanese government tried to make mass public housing
#
available. These affordable public houses were built outside the city and the middle class
#
loved them except that you know nuclear families shifted to these locations away from city life
#
and you know Japanese women when the men went to work Japanese women didn't have anything to do
#
there's no social life there's no you know all the things that a city has and they need televisions
#
and this leads to a boom in television manufacturing especially because 1964 Tokyo
#
Olympics is coming and everyone wants to watch that so there's a television boom and then at
#
one point they realized that hey this television boom is going to end what do we do then you know
#
and then they'd you know invest in personal calculator research and all of this is you know
#
one strand playing out another strand you point out is that there's an ideological war in a sense
#
where the US realizes that communism must be fought and therefore to fight communism in Asia
#
which is really North Korea and China they come up with what you've described as a containment
#
policy which is you know they decide that okay Japan, South Korea and Taiwan must be our allies
#
and we will support them in whatever way including making their future semiconductor
#
industries decades from now you know much more stronger than they would otherwise have been and
#
then that becomes a factor and this transfers of technology and know-how and all of that happening
#
and then you talk about this government project in Japan between 76 and 80 called the VLSI project
#
you know which also moves towards an innovative economy from what you earlier called an imitating
#
economy you know and then they are churning out patents they are dominating lithography manufacturing
#
and so on and so forth and you know and eventually then East Asia takes over DRAM memory chips and
#
I think they had some 85% by 1995 of the world market in that and it's kind of pretty insane
#
and you've just pointed out how the US reacts to that now give me a deeper sense of what's
#
actually happening in East Asia like I just shared these broad contours because I'm just so
#
enthusiastic about this story you know there's so much happening here and at another level it is
#
a simpler tale of you start down the value chain maybe you're doing assembly to start with and then
#
you start moving up and then you get into manufacturing and design and so on and so forth
#
so is it simply a case of moving up the value chain as you grow richer or these contingent
#
factors all really sort of play a part so give me a sense of East Asia now and of course Singapore
#
Malaysia are also part of that but you know Japan South Korea and Taiwan are the you know
#
the really fascinating stories yeah before Abhiram talks about the technical bits again there is a
#
larger economic story here I don't think there is like something spectacularly different about
#
semiconductor progress in East Asia and they just got better at many things like automobile
#
industry and all also so the story is similar that first a lot of people talk about for example
#
that Taiwan South South Korea has these chaebols and they are all national champions etc but a
#
lot of people forget that the story starts with lowering of import tariffs there was a lot of
#
trade which happened in the 50s this is before the military government came in that led to a
#
situation where they saw the benefits of integrating with the global market there was also
#
devaluation of the currency which happened in South Korea it happened in India also in 1966 but
#
India stepped back whereas South Korea kept on that track so that led to South Korea actually
#
becoming very integral part of global value chains when this world was just coming up right
#
where this comparative advantage based specialization was just beginning to happen
#
they were at the right moment there so a lot of American companies setting up some contract
#
manufacturing shops there all those happened and then they get become better over time just like
#
with any other country right there's no magic bullet in that they keep developing capabilities
#
and get better at it so with so was the case with Japan as well specific pathways are different for
#
example Japan was a lot more focused on having national companies whereas South Korea was more
#
open to MNCs establishing some bases here so those minor differences were there but overall
#
story was similar start at just electronics assembly you know just do original equipment
#
manufacturer white label product for some American company you learn that then you go to the next
#
stage where you think okay now I can probably do chip assembly you know not even manufacturing
#
then chip assembly after that you get better at okay now can I there's a sunk ladder theory
#
sinking ladder theory in software also it's quite similar so then you start capturing what is the
#
next bigger value that I can occupy in this revenue chain so then you start thinking of doing
#
fabrication and so on and so forth so that's what happened in South Korea Taiwan and Japan
#
all three cases there were some elements the difference was in China's case so for example
#
because as I said the tech transfers were easily easily available in these countries because
#
there were political convergence between us and these countries but with China that was not the
#
case right so China didn't get the technology transfer that these countries got so China tried
#
to do its own version of Atmanirbar actually Atmanirbar is a strategy used by Mao Zedong as
#
well it was one of the headings in my own research is China regeneration through Atmanirbar
#
so that struck me as well you haven't used it in your book of course but it struck me yeah so
#
they started doing that and these so their strategy was okay we'll do things through the
#
government route so government industry will start that will make my chips etc they started
#
doing this way back in 1956 so you know they also had this government plan that semiconductors are
#
important we'll do something there and as it happens in many government run programs there
#
was some initial start they were able to cobble together something but after a few years they
#
were not able to keep pace with these advancements the high capital costs required and they sort of
#
fell behind Taiwan Japan etc which took the lead and later on it was only through
#
FDI of American and other companies post liberalization you know in 90s that China
#
got technology transfers not because of good political relations with the US but largely
#
because it opened FDI and there were a lot of US companies which set up shop there and that's how
#
Chinese companies learned and then again got better so that was sort of a difference in
#
the pathways that Taiwan Japan and South Korea had versus China Abhiram you have anything to add
#
yeah so one more important thing which took place at the same time was the environment overall in
#
all these countries was getting substantially better the infrastructure was getting better and
#
people were actually focusing on making the overall livable environment much
#
comparable to how it was in the US so the return of people who were from Taiwan from
#
South Korea from China who had gone to the US that played a very important role so you know
#
you were talking about Morris Chang he also worked in Texas Instruments in the US but he came back
#
to set up TSMC so coming back of people from US to these countries where right now the
#
salaries were also comparable and the environment was also conducive that they could do their own
#
research and you know there was no hindrances to that that played a major role to make sure
#
you know the quality of the quality of the product that was produced was actually on par with
#
being exported that that was very important as well and another sort of a broader question you
#
know earlier you spoke about Zelenograd and we'll come to Europe but you spoke about Zelenograd set
#
up by the soviets to be like their Silicon Valley and that completely failed you've laid out all the
#
reasons for why it failed also and there was by the way they had a separate division called
#
Directorate T I completely fucking love these names Directorate T and their mission was to acquire
#
western equipment and technology but as you point out you know they weren't really doing original
#
work the emphasis was on copying not on R&D in the rest of the world there were substantial markets
#
for chips but for soviets it was only the military the supply was uncertain because a lot of the
#
equipment and raw materials were smuggled it was all a mess and it's easy to kind of see why it
#
you know failed and it's a sign of what the state was like like you know you and I both champion
#
nuclear power and you know how incredibly safe it is and people will always say but Chernobyl
#
Chernobyl and actually Chernobyl was a failure of the state of a really ossified state also at a time
#
when the technology was much older nuclear power is totally the safest form of power known to man
#
by an order of magnitude and we've discussed this in a past episode I think as well I remember
#
talking about this with you and Nitin but leaving that aside and you know and that's just a soviet
#
state that's why they failed and you see the failure happening with China as well like you
#
know at one point you cite the book paper tigers hidden dragons and you talk about how you know
#
Douglas Fuller the author of that book classifies firms into four categories number one neglected
#
domestic firms with little funding from the state two favored domestic firms considered national
#
champions three foreign mncs and four hybrid firms with chinese dna but access to foreign funding
#
or fdi as it would be and the state's intention was that the second category the national champions
#
would flourish but actually the state's intention didn't lead to outcome as it really does it was
#
actually you know because they opened up fdi and created that sort of enabling mechanism
#
those were the companies the fourth category which really flourished and sort of did well so
#
my again meta question to you is that we see different kinds of state intervention through
#
this story right we see the americans getting a lot of stuff right and we see the soviets getting
#
it completely wrong and china broadly getting it wrong and getting some things right by accident
#
without really knowing what happened or without intending to get it right in that particular way
#
what is there a lesson in this about state interventions as well that there are ways in
#
which interventions will always fail and there are ways in which you are a little more likely
#
to succeed for example i would imagine if you focus on creating an enabling environment rather
#
than directing companies what to do you're more likely to succeed you know as china did by setting
#
up the special zone at shenzhen for example or taiwan of course famously set up a special
#
zone like that much before that and south korea copied them and those enabling environments to
#
me would seem to be one reason but as you studied this way more than me what can states learn from
#
this yeah i think the one macro lesson for me was government should never run semiconductor
#
companies wherever they have tried to do that it has failed in the u.s case it was never government
#
running these you know especially not in design maybe the funding for some fabrication etc was
#
subsidized but not beyond that so that is one and to the wonderful list that you added ussr china
#
we should add india there as well where which was no different the same strategy we followed
#
there were two you indian companies government run companies which again got some head start
#
were doing something but what happens right in these industries one the only way you can survive
#
is to first serve the consumer market not just the military and the government market right because
#
the amount of money you need to claw back into the next stage of development can only come when
#
you are doing mass manufacturing for a large market which cannot be military you know militaries
#
will require some few thousand chips a year you can't sustain on that demand alone so you need
#
to get really good at one manufacturing for consumer scale products and second when you
#
are doing for consumer scale you have to be good internationally not just for your old products that
#
your citizens might want so wherever countries have gotten better at that and companies have
#
gotten better at that they have succeeded whereas when you are a government run entity
#
none of these incentives apply right so for me there were like three lessons actually one that
#
government run companies have no incentive to compete in a hyper competitive space which requires
#
constant you know capital infusion and technology upgradation so they start first and they can't
#
keep themselves in the race for long so even the eventually when they start getting things right by
#
that time the products are so costly that government companies itself will not buy from them they will
#
buy import from elsewhere so that is one thing you will see in all these stories second actually
#
because they are government companies the incentives run very differently and they are
#
shielded from internal competition you know so this I can illustrate with the Indian case where
#
you had two companies both were manufacturing chips again from some technology transfer from
#
the US etc so they got started but in the when you are when the money is being given by government
#
competition is actually a waste of that resources right so because why the government thing was why
#
should I put money in two companies what the hell like why can't I get them to focus only on one
#
thing and that's what happened so in fact Bharat Electronics Limited and Semiconductor Complex
#
Limited were both trying to do this but the government said SCL you only do chip manufacturing
#
BL you only do some overall component manufacturing not chips and eventually there was sort of no
#
competition for them to get any better that was not what happened in Taiwan for example so Taiwan
#
first the government led ERSO was able to spin off multiple private companies you know and there
#
was competition between them so you had TSMC, UMC, PSMC there were multiple players who were
#
competing with each other whereas in government that you know because government was giving the
#
money every money was seen as having huge opportunity cost so they eventually had no
#
incentive to make these companies compete so that was one this is fundamentally at odds with
#
innovation right how will you improve and third there was also Russia I mean USSR, China, India
#
at that point of time were all looking from trade from a very negative perspective so they were
#
all having inward looking trade and business policies that again proved costly you know so
#
the dominant narrative there was we have to save foreign exchange and how do you save foreign
#
exchange stop imports so you had all these companies which wanted to do things but they
#
themselves couldn't import equipment etc to really start churning cutting-edge products so their
#
requests were pending in some government ministry of trade for long periods of time
#
and eventually they just fell behind so this is one lesson you can draw from all government
#
run efforts in this industry. Let's talk about Europe now you know you've taken us through
#
sort of America's journey through this including the two sort of wars with you know first
#
I mean tech wars in a sense where it becomes part of the foreign policy you know first with
#
the Soviet Union during the Cold War and then with Japan in the 1980s when Japan starts getting ahead
#
in DRAMs and so on you know we've spoken about East Asia a bit take me a bit through Europe
#
because you would imagine that Europe is extremely well placed you know to do really well I mean we
#
spoken about what went wrong in Russia but otherwise you would imagine that as far as
#
Western Europe is concerned you know they should do pretty well you know you US allies like the UK
#
Netherlands France well Germany you know before the reunification were a part of the global
#
semiconductor ecosystem but yet at one level firstly started kind of falling back and then
#
there were these different initiatives which worked out in very interesting different ways
#
so you know to get us up to the current time just take me through a little bit of the history of
#
how this played out there both in a technological sense a geopolitical sense and
#
you know just how those industries evolved so when you look at Europe one difference was that
#
right when America was developing the semiconductor overall ecosystem Europe was right there with them
#
I mean it all started with vacuum tubes and it then moved into semiconductors and everything
#
right so Europe was also a major manufacturer of vacuum tubes they needed internally for themselves
#
as well and that's how it all started but I think the key difference would be that because
#
of their good ties with you know American companies it was easier for these American
#
companies to set up shop in Europe and it is a capital intensive business to start a semiconductor
#
firm start manufacturing and set it up and Americans were already doing it and they were
#
better at it and you know had to play catch-up at the beginning so at that point of time all
#
governments so I mean most of I mean actually the UK government and the France government
#
they started with the fact that you know since there are already good companies out there
#
they they were not as incentivized to start their own semiconductor manufacturing because
#
they were already getting what they wanted and all this retreat deficit was not something they
#
were worried about as much because I mean they were allied with the US and they just could
#
start using whatever US had so that is how it started but eventually even European companies
#
started understanding the importance of the role that chips played and you know how important
#
they were for the military and overall how they were one of the best of technology and right now
#
I would say there are a good set of European companies that are doing pretty well you know you
#
have pills that is there you have SGS I mean French company SGS which now has you know merged
#
multiple different times has become ht microelectronics you have German Infineon semiconductors
#
which makes I mean which is where I worked at and you know they make they are the number one in
#
automotive chips and all and you know even the UK had you know ARM that we've talked about IP
#
licensing ARM used to do very well and the Netherlands has ASML which is I mean which
#
is I mean if you look at the semiconductor space if we were talking about TSMC not many people
#
knowing it ASML stands equal to TSMC it provides very important components which no other company
#
in no other country can do so over time US has seen that you know competing in the mass
#
manufacture market of you know the memory chips or logic chips was not where it was cut out for
#
its labor costs were high and you know it's setting up costs were hey and they were all
#
it was mainly the European Union that was working as a whole because just multiple different
#
countries coming up with their own incentivized programs and they didn't simply have enough market
#
to you know focus on semiconductors and build on that so eventually they started figuring out
#
different places where each country would have its own competitive advantage and you know some
#
countries focused a lot on research there is this particular firm in Belgium IMEC Belgium which is
#
a research institute which is where every single semiconductor company goes to for cutting-edge
#
research you have TSMC working with IMEC Belgium you have ASML, Intel it works with IMEC Belgium
#
figure out what was the next right now we are probably you know the nanometer size of a
#
transistor where do you go after this 10 years down the line so this kind of research is where
#
a lot of European countries are actually working on closely and you also have ASML that I was
#
talking about where they are the only companies they are the only company in the world which are
#
able to make lithography machines that are able to make the transistors as small as three or four
#
nanometers so the machines are called extreme ultraviolet lithography machines and basically
#
what they do is they take the blueprint and they print it onto a chip they use specific masks use
#
very expensive very complicated lasers such that they're able to get onto a chip the size of your
#
fingernail so it's a very very complicated process and even for ASML it was not able to be done alone
#
it needed 15 to 20 years of funding from from governments from companies like Intel owning
#
stake and you know funding and it needed like 150 different products from multiple different
#
companies across the world and each machine also costs like 180 to 200 million almost
#
so this level of complexity competitive advantage that the Netherlands now has
#
and what they chose to focus on instead of going the mass market play like most of the
#
Asian companies did no I would so again Amit it worked perfectly like comparative advantage
#
based specialization theory would say so in in a world where this geopolitics and semiconductor
#
industry is not politicized Europe is placed quite well so because they have certain points
#
in that industry where if other nation states were to do something to them they also have
#
critical nodes which they control and they can again use that as a way to bargain so they do have
#
that option available to them but they are I mean labor costs are high running industry in the Europe
#
is not that easy so all those are the weaknesses so to say if you think of a very extremely
#
politicized semiconductor supply chain so that's where their weakness lies but yeah they do have
#
like Abhiram said right the beauty of the photolithography machine is also something
#
to be studied it's so damn complicated requires two planes to carry that one machine
#
and cost 180 million dollars a pop like but just the technical brilliance of making that work
#
is also really really amazing I mean I'm not even going into that
#
details but it it is something of some human marvel and again it got created in Europe so
#
Europe does have one instrument which no other nation state has yeah and you know in your book
#
of course you've got a narrative about how whenever the state tried to do something like
#
when it tried to put together things like esprit and the eurochip program they kind of failed for
#
different reasons which you've sort of detailed out but that you've also got these great success
#
stories like arm like imac like asml and by the way I discussed asml in an episode of everything
#
is everything with Ajay Shah called the three globalizations we shall link that from the show
#
notes and in that I asked him a question which I will now ask you Prane let us see if as an
#
expert in this field you know the answer to this and the question is this that if you recorded the
#
sound of one of asml's machines and played it in such a way that it gave you deep sensuous pleasure
#
and made the hairs on the back of your neck stand up what would you call the experience asmr asmr
#
by asml well done well done crack it I'm sure you've seen the episode no I haven't really I
#
haven't yeah yeah no no that's that's quite wonderful and like you said you know the perfect
#
illustration of comparative advantage and so on and I thought you were going to go with the bargaining
#
chip reference or something like that oh my god don't get me started Prane don't get me started
#
you know you don't know how this will end but yeah that's a good one great mind sync like and
#
fools seldom differ so let's let's come to matters of national interest I want to talk
#
about India what are we doing and you know we've had various phases I think you describe a total
#
of three phases in which India has sort of you know been a participant in this whole game three
#
versions version one version two version three so you know give me a potted history of you know
#
India's role with semiconductors and how we've sort of thought about it yeah so again the first
#
version we call it by the government and for the government so again it was in that period where
#
they thought government only a government company can do this so they tried a lot and we they
#
eventually reached only a stage where you had companies which can do some small amount of
#
manufacturing for space and for defense sectors that too a small bit nothing beyond that we have
#
no manufacturing facility in India which is at a commercial scale so that was sort of the first
#
version and I have described the failures of that comparing it to USSR and China's the second version
#
was when by the time government realized that this is not a matter for governments to solve and 1991
#
liberalization had also happened so thereafter the idea was can the private sector sort of do this
#
and there that stage that version I call the unease of doing business right that was the
#
period where there were a lot of attempts by companies to start things in India also by MNCs
#
to start some assembly not even fabrication but even assembly but all of them see one thing we
#
have to remember this is a capital intensive and high gestation period industry right so if you
#
start making a fab today the first chip will come out of it only three four five years from now
#
so you'll start making money after five years so you need to have an amazing amount of trust in
#
the trade business policy environments which Indian governments failed to provide for a long
#
time so at least in that golden period 2004 to 2008 there was an attempt to start an assembly
#
plan but then the global financial crisis happened so many companies backed off then in
#
2013-14 there were another couple of attempts but by that time you also had the Vodafone case
#
so there that set that made it very difficult for any company to do long-term investment of
#
billions of dollars scaring scared about what the government might do next right
#
retrospective taxation and things of that sort so my assessment is that it is because of poor
#
tax policies and also there was no GST I know GST has other problems but GST at least is a
#
uniform tax structure otherwise imagine you had to have so you needed to have your warehouses in
#
each of the states and for a globalized industry like this it would have been even more difficult
#
so those tax problems policy uncertainty and also the fact that we started increasing import
#
duties from 2015-16 that again makes it difficult because you need to import to export more in this
#
industry as well and that is again something that you couldn't do so this was the second version
#
where because of businesses running a large scale business being difficult none of these countries
#
companies came up and I will again draw parallel to it's not just in semiconductors again nothing
#
special in semiconductor assembly that's been India's weakness in other large-scale manufacturing
#
as well I mean we've we've started doing poorly even in textiles right so it's not just semiconductors
#
so that was the second version now version three is what the government is attempting now and it all
#
started again because of COVID-19 and the pandemic that led the government to again start
#
thinking that can we again start something new here the difference is that this time
#
one the Indian market itself has grown so now there is a demand which is local as well which
#
was not available 10 years ago there are also companies like Apple etc which are doing some
#
electronics assembly now so probably they would want to do some get some chips assembled here as
#
well right every it's in their interest to have a local ecosystem not so much in the now even if
#
the government wishes or not they would want to do that so that is the new force which led the
#
new schemes for doing something in semiconductors and the government is trying to do things across
#
the value chain in design in manufacturing in assembly which was not the case earlier earlier
#
it were piecemeal arrangements the third thing government did is because policy and trade
#
environments have still not improved significantly the government is putting capital investment
#
upfront not as a reimbursement earlier it was that you set up things and once you have set things up
#
reimburse 50 percent of the costs or whatever do you but which company will wait for five years and
#
on this promise right so now in order to get over that the government has said that for every rupee
#
you put in we will back you up with another rupee so that's a new strategy which probably will
#
encourage some sort of assembly but the fact that for the first company which has agreed to do this
#
is an assembly company from the us which is called micron even they have there are roughly
#
70 percent capital subsidies which have been given to an assembly firm which just illustrates
#
how difficult it is to do business in india and to do large-scale manufacturing so that still remains
#
and the way i see it is that government probably has three priorities in mind one is to
#
double down on your comparative strengths which is currently design because human capital intensive
#
process you require large number of engineers which we have so almost every top 10 semiconductor
#
company by revenue has a design center in india a fabulous company by revenue so we already have
#
that ecosystem for doing well in design also doesn't require you huge capital investment etc
#
so that could be one aim double down on comparative strength the second goal could be to actually
#
reduce your big major vulnerabilities you know so for example none of our defense equipment will
#
have chips made in india so maybe at some point of time you need to have things which are developed
#
in india so that would be the second objective the third objective would probably be to reduce
#
trade dependence on china now this is an objective i don't agree with but in stated policies of the
#
government this is an objective and they want to reduce the amount of chips etc which are coming
#
in from china so these are the three goals based on these three goals they have tried to do things
#
across the supply chain my assessment is that they have gotten the priorities wrong first of all we
#
should not worry about trade dependence on china but we should focus on the other two goals and
#
maybe doubling down on comparative strength should be the priority instead of trying to build a fab
#
and reduce vulnerable vulnerabilities today maybe you should start maybe have try to get a fab if
#
it happens it's good if it's if it doesn't happen it's fine but you will develop expertise in this
#
only 30 40 years down the line so if you start a fab today at the trailing edge maybe we'll get
#
to a reasonable stage 30 40 years later just like taiwan took such a long time but comparative
#
strength is designed so i think the goal should be can you develop more capable engineers can we
#
have indian engineers probably running semiconductor industries across the globe
#
not just as engineers but also as plant managers who can manage this complex supply chain etc so
#
if we develop that expertise you india will have some leverage in this supply chain from a
#
geopolitical perspective marvelous but i must protest pranay i must pranay protest that you
#
have let our nation down by saying we don't have fabs we have fabs i can totally imagine an american
#
from silicon valley coming here and going to fab india and show me the chips why do you have
#
these long long shirts show me the chips what are you doing but yeah what i'm going to try now
#
before i you know ask you my next question and ask you to elaborate on the present and the future
#
is sort of try and sum up in my own head this whole story in really simple language as i
#
understood it so you can tell me if i understood it correctly and fill in the gaps that there is
#
this period of great innovation in the u.s and you know the transistor is invented there in 1947
#
in bell labs and from there you get this whole semiconductor revolution happening and this is
#
given a philip by you know the ideological imperatives of the cold war there is what you
#
call this creative insecurity which is a therefore a drive towards innovation for geopolitical reasons
#
the state pumps in a lot of money and you have this whole industry taking place at the same time
#
you know there are different things happening in east asia and the u.s is part of its containment
#
policy of containing communism in china and north korea decides to support japan and south korea and
#
taiwan and initially they start at the bottom of the value chain and they're doing things like
#
assembly and gradually as the years go by they kind of move up they do they do this in different
#
ways they have different policies they have you know different kinds of success with it but they
#
all move up and taiwan becomes really critical because they become the fabricating powerhouse
#
of the world to the extent that if you know taiwan is to suddenly be shut off from the rest of the
#
world the world would basically be screwed it would be a huge humanitarian disaster because a
#
lot of things would just stop right there and so there is a danger that what if china you know
#
wants to invade taiwan and taiwan of course used to be in japanese hands at one time china gave it
#
to japan in 1895 and you know the first seno japanese war took place and you know they lost
#
taiwan there second world war they kind of get it back but then taiwan goes off on its own that's
#
where the nationalist government is and there's always sort of this collision so for political
#
reasons as much as anything else china would love to invade taiwan and get away with it and that
#
is the nature of the world's worry now because suddenly the other countries you know look at
#
what is going on and they realize that sure we have this incredibly intricate supply chain which
#
is miraculous and which has delivered so much value to the world but what is a feature can also
#
become a bug because any part of this can break down such as you know china invading taiwan or
#
you know natural disaster strikes a place there are earthquakes in japan or whatever you can't
#
afford any part of it to break down you know so what do we do and a lot of nationalistic
#
thinking kind of comes into play and india has been peripheral all this while it's made clumsy
#
attempts to be part of the game but it isn't and now your argument is we can you know be part of
#
the game if we focus on comparative advantage like europe did which is in terms of design but
#
otherwise whatever national policy has to be set i mean i would have a question in terms of you
#
know you have to think long term for something like this to work out and who's doing the long
#
term thinking and who is to do it is the state to kind of do it or are independent players to
#
follow incentives that are before them and to make investments accordingly like you point out in china
#
the idea of creating national champions simply did not work out what worked out was allowing fdi
#
and allowing investors to come in and invest and for that the question is that you need a
#
conducive business environment and after all that has happened in the last couple of decades
#
you know is foreign capital really going to trust india to that extent again what can we do
#
to build up that trust so this is sort of the overall picture that i see so now i want you to
#
talk about the present and the future because it seems to me right at one level this is now
#
a fundamental shift in geopolitics that even though we have an outright war going on and
#
you know russia ukraine are fighting and all of that the age of outright wars is kind of over and
#
you could argue with so much globalization even trade wars is you know completely different in
#
nature from what it used to be and now today technology has kind of come to the forefront
#
so tell me a little bit about this you know both uh you know when you guys started noticing this
#
pranay you said in 2019 you first became aware that fuck i have to start thinking about semiconductors
#
again so take me through your sort of like did i state this more or less correctly uh you know
#
what would you add to this and you know give me a sense of where we stand now because what has
#
happened in the last three or four years where do we stand now what are all these different
#
stakeholders involved how are they thinking about this what are the things that they are doing
#
right so i think you had the summary spot on i will just add one element to this on the taiwan
#
angle right so the interesting angle there is again tsmc the specialty about it is that it
#
has mastered this recipe to produce those chips at minimum loss right so if you have uh there's a
#
something called a wafer so again another chips and wafers so there's a wafer which is sort of
#
you can think of it as a pancake and there there are chips on top of it right now the now what
#
happens is taiwan you can on a wafer you will have some chips which will work perfectly but some
#
because the production process so complex they won't work as you desire it to be so you need
#
companies to develop the perfect recipe so that the yield is high right so most chips on that
#
wafer will work and hence your profitability will be more that's what tsmc has been able to do better
#
than any other company now if you think of it this recipe doesn't come from some machine or
#
anything it's because of the processes they have put in place the human ingenuity and the learning
#
that they have done over 30-40 years so if that resides it's a knowledge problem and not a machine
#
capital problem it means it can move to other countries as well right so that's why what you're
#
seeing is that taiwans and tsmc's option is that you try to establish bases in other countries now
#
so us is attracting tsmc to set up base there japan has already uh in starting a tsmc plant there
#
so the hope is that even if china were to invade taiwan there will be other alternatives which will
#
come up by taiwanese engineers and tsmc engineers just going to other places as well so that is sort
#
of the hope and the flip side to this is even if china were to invade taiwan and take control of
#
all these machines in at tsmc it's very unlikely that they will actually control the supply chain
#
like taiwan did because tsmc itself is dependent on so many equipments from asml and materials
#
from japan which will stop to china if china were to invade taiwan so in that sense i think taiwan
#
is important but its importance to the semiconductor industry is overstated in the politics of today
#
you know so a lot of nation states are precisely doing industrial policy because of this reason
#
but i think their other alternatives will come up there will be a shock but it is not as if
#
china will occupy taiwan and there will be a tsmc operated by china and then all countries are have
#
no option but to just buy from this china owned tsmc i don't think we'll get to that world so
#
but nevertheless this is the narrative and that is driving a lot of and also this narrative
#
is beneficial to the local semiconductor players right so intel would want to say this to the
#
u.s government give me money otherwise this will happen in taiwan and that is also driving a lot
#
of industrial policies in every country so the present is coming to what you are talking about
#
the present so that is where we are at this moment where a lot of nation states are trying to pump
#
in money japan south korea taiwan china india europe u.s all of them want their own local
#
semiconductor industry to a certain extent at least more than what it is now so and a lot of
#
them are focusing on fabrication because taiwan is the cited reason right so that is one angle
#
that most countries are trying coming to your india point i would say that i agree you know the
#
better way would be one to integrate with the global supply chain right so we are not part of
#
rsep we are not part of any trade formation that means that a lot of the imports and trade with
#
the world is difficult you know and that makes it difficult for any mnc to set up shop here
#
if the products that will be made here are not competitive at a world level right
#
that is one thing that indian government has to do and tax policies trade policies business policies
#
are still not settled that is one thing which governments should focus on but always governments
#
like nice shiny things buildings like a fab are always nice jewels that companies a country
#
governments would want so industrial policy comes very naturally to many governments and that's
#
where we are we are on that path in all countries including in india yeah that's that's where the
#
present is i'll come to the future later so i have a follow-up question you know to buttress your
#
point you know i'll quote from your book where at one point in fact you quote tsmc chairman mark
#
leo where you say he went on record saying that quote if china were to take over taiwan it would
#
find tsmc's facilities unusable because a secret ingredient is human capital and real-time international
#
collaboration with companies for material software hardware and know how stop code and this reminds
#
me of perhaps the most tweeted sentence of the last week which was again during the open ai
#
controversy where at one point to show solidarity with sam altman all of these open ai people who
#
were threatening to leave tweeted out a quote open ai is nothing without its people right so you know
#
that again speaks to the same point and my question to you then is that surely the secret for any
#
country which wants a leg up in this game is attract the best people figure out ways to attract the
#
best people create network effects the same way that you know silicon valley would at one point
#
in time and everything will kind of work out and you create the and i don't think we are attracting
#
the best people in fact i think in some ways we are driving good people away but to attract the
#
best people you want to you know have good tax policies you want to figure out what kind of
#
incentives can make you know nris indians in silicon valley for example return here the same
#
way as abhiram said that you know in taiwan and in south korea a lot of taiwanese and south korean
#
diaspora actually went back to be part of their country because living standards were better and
#
they felt hey let's come here and build our own country and it's a good life anyway and to quote
#
your colleague nitin pai you know the best foreign policy is a rising gdp you know so what what do
#
you sort of think about that thought of it that perhaps we pay a lot of attention to all these
#
intricate details keep you know where is the fabrication where is the assembly where is this
#
where is that but one key part of the element is just figure out ways to get the best people to
#
where you are create those conditions and the rest will happen and comparative advantages will
#
present themselves no i agree to that and in fact many countries are now thinking of having
#
something called silicon visas you know and it's not visas made of silicon but visas for the
#
semiconductor industry to attract a lot of you know talent in this area to their own countries
#
for example us the bottleneck that us faces today is not money they are pumping in huge amounts of
#
money but the bottleneck is really talent they don't have enough people to work in the new fabs
#
and new companies which will come up that will eventually and chinese people can't go there
#
because this industry is now deemed to be geopolitically sensitive so it will become
#
difficult for them to so which is the country from where many people will go it will be india right
#
now it is it benefits the u.s probably benefits india as well if these people come back but this
#
is how it will play out in many countries human talent i think will be a very important instrument
#
of statecraft and also a domain of geopolitical contestation to attract the best human talent
#
in some cases it might also be to make human talent movement difficult for example because of
#
china japan australia relations now japan is thinking that there are certain strategic sectors
#
in which chinese students coming in to japan should be kept away you know so that is the
#
flip side of it but so human talent is a really important ingredient it will in the future also
#
become a lot more contested and yeah you need to attract the best people here in in any country
#
that will be important for many people in this industry let's talk about the future now in your
#
book you have this fascinating section on four likely trends in you know geopolitics regarding
#
the tech wars so you know so tell me a little bit about those likely trends you know that are going
#
to come up and how foreign policy is going to be impacted or going to change because of that
#
yeah a few trends that sort of i see going ahead in the future is one as we see trade wars are
#
largely going to be technology wars and that you also mentioned because nuclear weapons make
#
geopolitical conflict i mean big large-scale wars unlikely similarly there's so much benefits
#
to globalization that complete decoupling is also unlikely for example us china the trade is still
#
going on there is not much of a dip in that as of now but the thing that is different is in technology
#
domains nation states are still not clear what are the actual losses of going down this path so they
#
are trying and i think it will lead to losses for everyone but every nation state is trying and it
#
is also attractive because the narrative is that technology is important for national power so
#
why shouldn't you do more things domestically why shouldn't you decouple right so that's why you
#
will see that there will be some decoupling in high tech areas it might be at the level of materials
#
machines humans or ideas but it the precise pathway will depend on the technology whether
#
it is ai or whether it is semiconductors that might differ but this decoupling is something
#
we'll see for the next few years the second trend that i see is that because of this aggressive
#
national competition there might be some non-linear breakthroughs right so if the flip side as we saw
#
discussed again is there is more creative insecurity so there are nation states which
#
are trying to invest more and more into science and technology because china has restrictions
#
put on it they might try alternative ways to achieve the same goal right so there might be
#
some non-linear breakthroughs which wouldn't have happened in the past but might happen
#
in this decade the third one is there is probably going to be higher alignment between
#
private high technology players and the national governments and that again is happening because
#
of the politicization of technology areas right so for example intel just a few years ago was
#
happy to invest in china to start fabs in china but once this idea happened since 2019 it has
#
stopped its investments further investments and it now because the government is ready to pay a lot
#
more money back home they are doing a lot more things either in europe or in us itself so this
#
kind of alignment between private high tech players and their national governments is more likely
#
earlier you didn't have right we had 30 40 years of a period where you just said
#
private companies followed profit followed whatever was whatever nation state and
#
environment gave them the best value and that was fine the national government's positions were not
#
that important but in these high tech areas including semiconductors it is likely that these
#
private players will now need to seek some export control waivers or some sort of alignment will be
#
necessary and the fourth one is that actually we might see selective international cooperation
#
in semiconductors and other areas but something which is subject to geopolitical considerations
#
so for example us china high tech collaboration will probably decline but us india collaboration
#
or us japan collaboration might actually increase because even us realizes that you can't build
#
everything at home you know that's the story of globalization and the way i put it is that
#
plurilateral cooperation is a necessity and not a choice in all these areas so you will need to
#
have some pcb manufacturing happen in india and some materials happening in japan so you might see
#
more international cooperation but it will be sort of bracketed by geopolitical favorability as well
#
and that's why you're actually seeing a lot of action on the technology front in the foreign
#
policy realm so you know the national security advisors of usa and india meet the if you see
#
the joint statement generally joint statements of nsa will have will do something against terrorism
#
etc but if you see this readout the latest there were words like hackathon and joint
#
starter ecosystems and things like that so that it signifies change when the indian pm went to
#
us and their readout also the first line of that is all about semiconductors and what india you
#
know it's very unusual to see that kind of thing in a us india geopolitical foreign policy talk
#
and then quad again something with india us japan and australia doing there again technology
#
and semiconductors are an important item of discussion there is a semiconductor supply
#
chain initiative by the quad as well so it shows that there will be more international cooperation
#
as well provided some geopolitical conditions are met so these are sort of four trends that
#
i foresee going so i have a question for abhiram but first i want to kind of seek a clarification
#
from you pranay what is the difference between multilateral and plurilateral the first time i've
#
heard that term yeah i think this is a term that gov government of india others use multilateral
#
when we mean we mean organizations even like the un right very big one tent organization where all
#
countries come up plurilateral are smaller groupings so you might just have four five etc
#
fair enough so here's my question for you abhiram that we've you know spoken about
#
semiconductor geopolitics and semiconductor geoeconomics but the third arm of all this
#
no pun intended even if i said arm but the third arm of all this is semiconductor technology and
#
there i find that there are also unknown unknowns there and maybe some of them are not quite so
#
like you write about the open source revolution there like how so far you know chip designs been
#
dominated by x86 and arm and now you have the open source alternative called ricv which has
#
you know been developed by the university of california no licensing fees china loves it
#
because nobody can stop them from using it it's public it's open source you point out that in
#
terms of eda also you know right now it's an there's an oligopoly but even here you know
#
there was an open road project launched in 2018 with support from darpa and so on and that again
#
you know is exciting plus you speak about how there are possibilities of looking beyond silicon
#
and maybe taking a big step up there as well so tell me a little bit about this also that what
#
are the kind of developments which are happening what are the trends happening and what are the
#
possibilities i'm not asking you to predict what will happen but just thinking in terms of what
#
could happen in different directions that could change the game a little bit
#
yeah you were talking about open source and you know using other materials and all right
#
so what we see here is there could be new players coming into the picture and you know existing
#
players who are at a disadvantage you know eliminating that disadvantage basically like
#
there are two types of architecture that are mainly used the arm based architecture and x86
#
so internships are mostly x86 and arm is mostly in mobile phones and three four years ago apple
#
famously moved from an x86 to an arm based architecture and their performance like went
#
up multi-fold both of these are i mean x86 is owned by intel and used by amd while arm is owned
#
by uk based from um uk based it is uk based basically so both of them are not allied
#
particularly to china and they have no incentive to be providing licensing to china as well so if
#
you look at it that way an open source alternative to both of these fits china's needs perfectly well
#
it doesn't help to rely on either of these countries and that is one way that is one way
#
to actually move forward without having to depend on the us or uk the same goes with
#
eda tools eda tools are basically software which are required to create the blueprint
#
i mean there are billions of resistors there so i mean you can you know draw it out with a drafter
#
so you will definitely need some kind of software which will do multiple computations to give you
#
the perfect scenario the perfect placement of the chips and the two main players are
#
canons and synopsis both of them are us based and no other company has been able to come
#
close to that level of domination and any chip design form which work at the cutting edge level
#
it will definitely have to use either of these softwares and this is when again
#
china will have to depend on countries like the us at least there are open source alternatives
#
coming up or the open mode was by darpa again but we do not know the level to which they
#
able to perform it is just in the starting phases and very basic chips will be able to
#
be made there but it's still a long way to go but it is still because china is looking at and
#
it is also investing significantly to you know make open source eda tools available for everyone
#
as well and this is also china strategic advantage where it was previously lacking behind but you
#
know i'm not going overly positive with both of this they might be able to help out at the
#
beginning but there's still a lot of work to be done over the next couple of years for them to
#
actually reach a stage where you know they become as commonplace as like linux operating system
#
which is open source or something like that so that is regarding the open source part of it
#
and our other is we were talking about silicon so we don't know silicon is the best material
#
for semiconductors but that might not be the case in the future there there is a lot of research
#
for silicon carbide graphene and in fact the uk has a separate silicon carbide zone that is coming
#
up with all the companies only focusing on research on silicon carbide even you know india has
#
invested significantly tata has invested in silicon carbide as well you can see even
#
different parts of chips you know it might not be for general purpose chips but for power
#
electronics which derive a lot of power silicon carbide might replace silicon and you might
#
actually see india you know taking an advantage there these are the places where each country is
#
trying to focus and get a competitive advantage and this may change the order of things and
#
which company is favorably seen as well yeah i i think just to add to this open source is
#
quite an interesting space to look at from a political angle one
#
it also benefits india right anything made open source benefits entire world including india
#
and it benefits china as well so actually now the u.s government is reconsidering that whether
#
there should be restrictions on open source going to china which defeats the purpose of open source
#
like how do you even do that yeah yeah so but there are they're trying to think of putting
#
export controls etc so all these are new things which we are literally finding out as we go ahead
#
i mean closing open source what what does that but it's being thought of and that this open
#
source idea is happening throughout the supply chain for example eda tools also like we discussed
#
you you mentioned then also you discussed in specific blocks like finding alternative to arms
#
which is risk five but it is also happening at the fabrication level so for example right now what
#
happens is that fabrication process is quite complex and the machine the files which are
#
required to optimize that process are called the pdk files process design kit and those are only
#
shared with specific companies which work with these fabs over time to fine tune that process
#
now there are other companies which are saying that why don't we open source this these files
#
as well so that anyone should be able to actually produce the final chip not just the blueprint but
#
the final chip after paying a cost for the machines materials etc so google for example
#
with skywater fabs which again and they have started an open source program so they had a
#
pilot where you can submit your design and that will probably get fabbed and you can have a chip
#
of yours and they had a limited run but this is something again which will be an interesting
#
area to watch out for going ahead that's something to look out for the other thing amit i wanted to
#
mention from a political angle my hunch is that nation states will realize that this
#
self-sufficiency path that they are trying to achieve will be futile and eventually go back
#
to some understanding of some modus operandi with china where trade goes on as it is generally what
#
happens is the semiconductor industry goes through these phases there are exuberant phases and then
#
suddenly it drops down so right now it's in this exuberant phase because of what happened during
#
the pandemic demand raised everyone wanted more of it so that that's why people are okay to pump
#
in more money etc but when the industry goes down all these assumptions will be challenged and
#
many suppliers in us might say if chinese companies don't buy we have no other buyer
#
so what else do you want us to do we either survive by supplying to china or we don't so
#
those kinds of things might actually make us regress to the norm in some way regress to the
#
mean and we might have some some of these assumptions go away but right now we are
#
dominated by this thinking that every chip is strategically important and hence it needs to be
#
made at as close to home as possible and that's the dominant thinking in many governments as of
#
now so i've done an episode of everything is everything with my co-host there rajesh and
#
it's called the three globalizations and we were sort of talking about the different waves of
#
globalization and my sort of lament which he tried to assuage but it still remains a lament is that
#
you know we are becoming a less globalized world that countries are increasingly becoming more
#
populist and nationalist and there's a danger of becoming more insular and ajay's point was that no
#
globalization isn't over it has just changed in its nature and the term he coined for it was a
#
third globalization and he said what is happening is that you still have the vast majority of
#
countries being effectively globalized trading with each other playing the game but a few countries
#
which are treated as rogue or potentially rogue nations like russia like china there are conditions
#
on them and you use you know economic statecraft and all the tools you have at your disposal to
#
keep them out of the globalization game and i'm not entirely convinced by that and obviously
#
ajay was also being descriptive and not normative in saying this is the way things should be but
#
he was just being descriptive that this is where we are and it seems to me that a term that i think
#
you must have coined because i haven't heard it before it's in the book silicon craft and it seems
#
to me that silicon craft is in a sense about this sort of third globalization about this kind of
#
economic statecraft that states carry out particularly in the domain of technology and so on
#
so i first want you to confess to coining the term if you have indeed done so and then tell me a
#
little bit about how you would define it and what are the kind of strategic objectives that countries
#
can have what are the things they can do how do they do those things
#
right yeah we have coined that term and before we go to that i'll just talk about that globalization
#
point i sort of agree with you largely that in the sense nation states will still see that
#
today it has happened to russia and china so india will see tomorrow it might even happen to me so
#
it ends up creating a reality that you want to prevent right so even though u.s might be saying
#
that this is only against rogue nations and not against you nation states will respond to this
#
from a different perspective and big nation states are not going to be totally aligned with
#
the u.s as well so when you are in the middle you will always think that you know what if this
#
happens to me and that sparks more industrial policy in third countries apart from russia and
#
china so that is a worrying sign that actually globalizations positives are largely being
#
overlooked but yeah the hope again is that you will go down this path realize probably that
#
this is not the path early enough and then go back to enjoying the benefits of globalization
#
but for now it doesn't seem likely at least in the next two three four years
#
that's my thing so silicon craft coming to that point the way we thought of it is try to imagine
#
what are the kinds of instruments of geopolitics that might be used in this domain of semiconductors
#
right so these can be political instruments or economic instruments and what are nation
#
states likely to use again this is a descriptive framework not normative one so here there are sort
#
of if a nation state were to think of what would be the objective of doing anything in this domain
#
i could think of largely four objectives one you want to deny something to an adversary the second
#
goal could be to outpace the adversary third goal could be just to increase supply chain resilience
#
thinking that this is for the benefit of everyone let's make it more resilient that's all and the
#
fourth one would be maybe you want to use this to influence the minds and actions of the citizens
#
in your adversary states so if you think of these four objectives we try to map out what are the
#
instruments that nation states might use for achieving all these objectives so for example
#
if the objective is denial nation states might use instruments like sanctions and controls which
#
are already in place or restrict the movement of semiconductor human talent third they might
#
actually have induced restrictions as in semiconductor trade goes as it is but if some
#
chips are used in military applications then there will be restrictions and fourth there might be
#
investment screening as in if some us company wants to invest in china in semiconductors there
#
will be investment screening going ahead in fact us government is considering this as we speak so
#
each of these we try to map out what are the underrated repercussions or unintended consequences
#
of these actions right so for example sanctions and controls like us has said us has put secondary
#
sanctions that is us has said a thaiwanese company tsmc cannot supply chips to a chinese
#
company because the thaiwanese company finally uses some intermediate product from an american
#
company and us can do it because it has the power to do it and now this works fine for tsmc i mean
#
it puts a lot of strain on tsmc right now but they are okay to absorb it
#
because of the political realities but it becomes difficult to sustain over time right so what if
#
the industry goes down and tsmc is not able to survive they will find some way to do a
#
backroom deal with a chinese company to transfer and make money there so it is difficult to sustain
#
they have started it but i don't know how long you can sustain this kind of sanctions and controls
#
second obviously restricting movement of human talent if that were the goal you will slow down
#
a lot of technical progress in all of the industries right talent interaction multiplies
#
research and that will stop the third one if there are induced restrictions like let's say you don't
#
want things to be done on the military front again it is very difficult to police this ensure
#
that some chip doesn't get used in military and gets used elsewhere so very difficult to
#
maintain it and it in any case encourages the adversary to develop local capacity in
#
that precise area because you have denied access in that area they know that is it is
#
vitally important to do something there so that is one so similarly if this the second objective
#
let's come to that if the objective is just to outpace the adversary not to sort of deny
#
then you could have certain instruments like you might some nation states might want to do
#
industrial espionage to gather secrets or to do targeted poaching of talent from other companies
#
or you could on the positive side if you just want to outpace an adversary you don't need to
#
indigenize everything you just need to build partnerships with other players so that
#
collectively you are ahead of your adversary rather than you yourself trying to do everything
#
the third one is indigenization and industrial policy which every nation state is doing now
#
the fourth one could be economic coercion which china is trying to coerce many companies
#
and that is something that nation states might use the fourth one would be to encourage the
#
movement of semiconductor talent again with your partner so that you can outpace the adversary
#
and the final one would be you just actually increase dependence of your adversary on your
#
own ecosystem such that your adversary is always two steps behind your own capabilities so these
#
are some kinds of instruments that we might see if the goal were outpacing the adversary
#
if the goal is increasing supply chain resilience and not denying or outpacing the adversary
#
then nation states might deploy actually promoting open source hardware kinds of instrument
#
or just to map supply chain such that if there were a shock in taiwan due to geographical
#
reasons or whatever you have mapped the supply chains and you are not in the pandemic like
#
situation where everyone is suddenly finding the reality you if you have mapped things you can
#
put mechanisms in place to absorb that sudden shock and the last one which appeals to me a lot
#
is again if you want to increase supply chain resilience build partnerships to create some
#
resiliency in the supply chain instead of thinking a lot of doing that at home in every other country
#
so this is how i would sort of map it and we were trying to develop this framework
#
we don't have a prediction on which of these will be used but we can definitely see a trend
#
that we are seeing a lot of the more self-sufficiency kinds of instruments being
#
used a lot now but we've tried to put this framework to highlight that there are other
#
instruments available which nation states can deploy and that don't amount to you always trying
#
to do things everything at home fabulous extremely enlightening and i should clarify to my listeners
#
that you know as they know i don't like using jargon but i've used a little bit of it myself
#
so i'll just clarify what i mean by those terms where you know we use descriptive and normative
#
and by descriptive i mean you know the world as it is and by normative the world as it should be
#
right so often we will be describing something and it might seem to an outsider that we are
#
endorsing it but we are usually just giving an explanation and not a justification so that's
#
important to point out that the descriptive is super important you know to understand to know
#
what to do to know what is the right way of action you have to know the world and understand the
#
world so the descriptive is super important and then later you can get to the normative and say
#
we should do this we should do that and what i loved about your book so much is that descriptively
#
it is incredibly rich and you know you don't you stay in the opinion of facts and analysis and
#
etc etc there's no hectoring at all no so two sort of final questions for you guys and each
#
of these final questions is for both of you and the penultimate one is that if we now look into
#
the future and we focus only on how this game is going to play out and what its consequences are
#
which will geopolitics in the semiconductor space if i ask you to look 10 years ahead you know what
#
is a best case scenario you would love to see and what is a worst case scenario you're really scared
#
of abhiram let's let's start with you yeah so best case scenario would be going back to the
#
level of cooperation that each country had already had you know having tariffs down and international
#
trade agreements set up so that there is zero customs and export and import fees but the worst
#
case scenario is where you see two different silos basically china is trying to do everything
#
on its own already right now and it has a sizable set of population as well which it will exclude
#
from others and it will try to self be self-sufficient on its own while the other countries will also have
#
to do the same this would be a blow for both of them there is less transfer and you know each
#
has to do it separately and the price of each ships would also be increasing yeah so i would
#
say the siloing would be the problem yeah to add i agree with the best case scenario actually
#
pre-pandemic world with some corrections to absorb to make things resilient and that government role
#
will be there at least to map the supply chains encourage companies because many of these companies
#
will not know who their suppliers are it's such a complex supply chain so mapping that might be a
#
role for governments so that might be the positive most positive scenario where we realize that
#
the pre-pandemic world trade was the best way with some corrections but and the worst case is
#
where you will have actually very high consumer prices consume very high prices lead to lower
#
demand lower demands will lead to lower sales made lower sales made lower revenue and lower revenue
#
will lead to lower rnd so you will actually have stalling of technological progress because of
#
the over politicization of this area and the entire geopolitics around it so i think that to me
#
is the scary scenario so you you will lose out on the gains that you have had you might not
#
have the next generation chip which might make chat gpt 20 possible right or something like that
#
so that is something to worry about i will think i'll answer this question from the
#
indian lens as well right what would be the positive case from the indian lens and
#
so maybe 10 years around the line the positive case for me from the indian lens would be that
#
we probably have one fab which starts up and through private investment some government
#
incentive and india starts learning this process we won't get to an expert stage but maybe you
#
carve out a specialization in that so for example abhiram was saying you don't need to play the
#
game that tsmc is playing you there are some compound semiconductors there are other kinds
#
of lower cost chip fabrication which india can start off with and occupy a space there so that
#
is a best case scenario in fabs if the best case scenario in design would be that the semiconductor
#
talent across the world in most companies is indian and probably they some of them return
#
they start their own industries companies in india i tried to think of hardware be in india
#
being where software in india was 10 years ago actually where you didn't have products in
#
software 10 years ago in india but you had a lot of talent here so eventually we got to a stage
#
where products are also indian maybe 10 years down the line that might happen in hardware as
#
well so that's sort of the most positive scenario the negative worst case scenario is actually where
#
indian government tries to do all this without actually making running business easier so you
#
have a fab starting and then it flops and that will just mean that government will not want to
#
touch this area for next many many years and next decade and then you will be in a situation when
#
all companies are spooked and just like what happened with the vodafone case and no one wants
#
to invest in this area so that is the scary scenario for india see pranay this is why you're
#
such a frequent guest on the show because i ask you one question you answer that question and you
#
answer another one as a bonus and that was a damn good question about india you know no no it's
#
great because as soon as i asked that when abhiram was talking i said shit here india ke baare me bhi
#
puchna chahiye tha but it's as if you read my mind and at this point i will tell you a story
#
of what happened to me many many years ago when i was a poor student and used to travel by buses
#
so once i entered a bus aur main waha baitha tha and i noticed this very unusual thing that the
#
conductor half the time to he is doing what he should do giving tickets taking money the other
#
half the time o backseat pe batke he is recording a podcast with someone so i went to him and i said
#
ki boss what are you doing what is your job he said see half the time i'm a podcaster you could
#
say i'm a semi podcaster so i said the other half what are you semiconductor well done well done i'm
#
sorry that was i couldn't help that so my that's terrible isn't it but luckily it's come right at
#
the end so only the two believers have stayed around my final question for both of you and
#
it's slightly different form for each of you because pranay has been on before but abhiram
#
since it's your first time my last question to you will be to give my listeners and me recommendations
#
of books music films that you love and that you love so much that you want to share them with the
#
world not just related to semiconductors or this field though that is welcome but just generally
#
what makes you happy now okay i actually spent a lot of time listening to the scene and the
#
unseen and all things falls in everything so i would definitely recommend just taking up any
#
episode of this podcast that would be very informative and very helpful i especially
#
yeah so dividing books i think we were already discussing uh we had talked about two different
#
books here uh the chip war and paper tigers and hidden dragons so those are two books that
#
we had to go through and both of them are amazing amazing books and regarding movies i am a very
#
masi kind of person so i like proper no pulpy indian movies so i love dhawan actually so anyone
#
was not seen i came here and saw it and i i would want to watch it again as well so that would
#
something i would recommend thanks so much and thanks so much for the kind words about the
#
show as well pranay you've already in different episodes recommended different things so i'll
#
ask you a more narrow question narrow question in the sense that in recent months what are the
#
things whether they are books or films or conversations or whatever which have really
#
sparked your mind and made you think in new ways and you know sort of given life to your brain as
#
it were quite a bit of things but one thing i have started reading a lot more is fiction
#
i know last time when we discussed i was so terry pratchett is one i have started reading
#
a lot so that is something quite enjoyable especially all the series which
#
focus on government and policy there are a lot of those novel series as well the patrician
#
he came up with the immortal phrase semiconductors all the way down
#
okay never mind carry on yeah terry pratchett fans will get that yeah yeah so but yeah so that
#
is one thing that i have quite enjoyed the other thing that i'm trying to explore a lot more is
#
the intersection of evolutionary biology and human societies so trying to read a lot of work of
#
supolsky of joe henrik so those are some things which probably will answer some of the questions
#
about why societies function the way they do so that is another interest area that i want to
#
learn more about so guys you know once again congratulations on a wonderful book i'd recommend
#
to all my listeners that they read this is extremely readable and made me just a little
#
bit smarter but because my memory is so bad i will be a little bit less smart again but
#
hopefully learning is always two steps up and one step down so thanks so much for coming on
#
the show this was great and more power to both of you thanks so much for hosting this wonderful
#
conversation amit thank you thank you thank you so much amit if you enjoyed listening to this
#
episode head on over to your nearest bookstore online or offline and buy when the chips are down
#
by pranay kota sani and abhiram manchi superb book you can follow pranay on twitter at pranay kota
#
you can follow abhiram at manchi garu we'll link these from the show notes you can follow me on
#
twitter at amit varma a m i t v a r m a you can browse past episodes of the scene and the unseen
#
at scene unseen dot i n thank you for listening
#
did you enjoy this episode of the scene and the unseen if so would you like to support the
#
production of the show you can go over to scene unseen dot i n slash support
#
and contribute any amount you like to keep this podcast alive and kicking thank you
