Best Place To BuildSharan Srinivas | CTO, Mindgrove Technologies | “You have to be crazy to want to make silicon”| Ep.9
EVERY SPOKEN WORD
65 min read · 12,758 words- 0:00 – 2:45
Introduction & Funding Announcement
- SSSharan Srinivas
The ecosystem is set up for you to fail safely.
- SPSpeaker
Mm.
- SSSharan Srinivas
Are you willing to accept that failure? Shashwat and I spent about a year just figuring out what to build. You also need support from your PhD guide. Your PhD guide can make and break- make or break your career. That creates a question: Why are all silicon making companies that large? [upbeat music]
- SPSpeaker
Hi, my name is Amrit. We've heard that IIT Madras is the best place to build. [upbeat music] So we've come down to the Sudha and Shankar Innovation Hub. We want to meet some people. These are builders. We want to talk to them about their work, and also ask them, what makes IIT Madras the best place to build? [upbeat music] Hi, welcome to the Best Place to Build podcast. I'm sitting with Sharan. Sharan is the CTO of one of the hottest new startups from IIT Madras, called Mindgrove. They're in the semiconductor space, and we'll learn from him what he does, and about the semiconductor space, and about the build culture at IIT Madras. He's at the center of it all. Hi, Sharan, welcome.
- SSSharan Srinivas
Thank you for having me here. It's a pleasure to come back to this space.
- SPSpeaker
Yeah. Um, firstly, congratulations on your funding announcement. It's a big deal, and earlier this year you got the DLI nod also. Can you talk to us about your funding announcement?
- SSSharan Srinivas
Sure. We've, uh... We are actually quite grateful that, um, we've had almost every investor who invested in us in the previous round, doubling down on us. We take that as a sign that what we are- we are moving in the right direction, and we are trying to do the right thing, um, and then as a, as a vote of confidence. In addition, we have two new investors, uh, Rocketship VC and Mela Ventures, who will be, uh, joining the cap table of Mindgrove Technologies after our Series A round. And, um, we have, uh, raised a total of about 8 million US dollars, and that's, that's the capital we have to take things forward. We, we took our seed round with the pitch that we wanted to make a chip. Now, the, the, the, the pitch is, we want to keep making more chips, and we want to sell the ones that, that we have actually made. So that eight, $8 million, uh, in order to get around to do that, and we additionally have, uh, been approved for the DLI scheme, or the Design Linked Incentive scheme, that is offered by the government of India. This is approximately $2 million, INR, uh, 15
- 2:45 – 4:00
Understanding the Semiconductor Industry Basics
- SSSharan Srinivas
crores worth of, uh, uh, incentives if you meet certain milestones, and we have been, quite thankfully, uh, approved for that, so quite chuffed. But looking forward to the, the, the challenges that are almost certainly going to come, come at us, uh, in the process of doing this.
- SPSpeaker
And you also have some really good quality VCs. Uh, Rocketship, of course, IITM alumnus themselves. Um, there's Peak XV, there's Speciale. Speciale being, um, an investor who has repeatedly invested in IIT Madras-based, uh, tech startups. That's amazing. Yeah, and these are brave investors, because some of these will require 10 years, these, these company... Maybe not tech, but maybe, uh, to, uh, sort of pay back.
- SSSharan Srinivas
Oh, yeah.
- SPSpeaker
So especially in semiconductors space, globally, there have been very few startups, like traditionally, none in India or very... Whatever, whatever have come out now, maybe in the last 10 years.
- SSSharan Srinivas
Well, I would, I would not, uh, say that there have been none in India. There have been some, um, startups in India in the previous generation. It is just that due to the nature of capital and expenditure, uh, that is involved-
- SPSpeaker
Right
- SSSharan Srinivas
... most of these startups, that the previous generation of startups, either had to, uh, exit by selling
- 4:00 – 8:00
Components of the Semiconductor Ecosystem
- SSSharan Srinivas
themselves to a larger company, becoming a subsidiary, or getting absorbed by a larger company, or they've had to incorporate in places like the United States, where funding was much more prevalent. We're talking 2008 to 2012, 2013 kind of era. So it is not that there have not been startups, there have been more there, but life was, mm, relatively speaking, life is easier now in, uh, in than it was, uh, 10 years ago from a funding aspect. Because venture capital itself, in inside India, has, has evolved that much more compared to 10 year, 10, 15 years ago.
- SPSpeaker
Understood. Um, can you tell us what your products are about? But before that, can you give us an overview of the semiconductor space? Right, I think the government is doing a major semiconductor push, but... And we've all collectively learned a huge number of new words, you know, SoC, SoM, um, and, and, uh, nm, nanometer, I guess, uh, and all these terms. So just a fab, foundry, OTMP-
- SSSharan Srinivas
ATMP.
- SPSpeaker
... ATMP, OSAT. Uh, can you give us a overview of what these terms are? Where are you placed in this ecosystem?
- SSSharan Srinivas
That will turn into a lecture. [laughs]
- SPSpeaker
Let's keep it very, very simple.
- SSSharan Srinivas
Sure. So the semiconductor industry is one of the most fundamental industries in existence today. Why? Because most of the things we do has become digital, and if you need the word digital, you need the semiconductor industry. It is what makes digital happen. You want a cloud, you need, uh, uh, semiconductors to get server-grade chips. You need semiconductors to get server-grade, uh, storage systems. Your Google Drive or your, your OneDrive of the Microsoft cloud, same thing. And you, you, you write software that is hosted on AWS, you need server-grade chips. You are... You know, that, uh, dabba, that you take the QR code to make UPI payment? That dabba that says you have paid so much-
- SPSpeaker
Not...
- SSSharan Srinivas
... so much rupees, the sound box. I, uh-... dabba sounds better, you know?
- SPSpeaker
Okay, sure. [chuckles]
- SSSharan Srinivas
It sounds more fun. Soundbox is so-
- SPSpeaker
Yeah
- SSSharan Srinivas
... very formal.
- SPSpeaker
Yeah.
- SSSharan Srinivas
So that thing has a semiconductor chip in it.
- SPSpeaker
Right.
- SSSharan Srinivas
The process of you taking a QR code requires a phone, which is, which is one of the flagship items of the semiconductor world today. So it's everywhere. We- our switches are no longer, uh, mechanical. We have buttons that we press, or we have touch screens that... where we click at a certain point of the screen to increase or decrease fan rates. So all these things require chips. All these things require what are called passive components, like resistors and capacitors and all of that, and the underpinning of all of that is semiconductors. So what is the semiconductor industry? Everything is a semiconductor industry, because even for cooking these days, we use induction stoves-
- SPSpeaker
Right
- SSSharan Srinivas
... and induction stoves have electronic components in them, and electrical component, power, uh, components in them, which are also semiconductors. If you plug it or you run it by a battery, it has semiconductors in them.
- SPSpeaker
Right.
- SSSharan Srinivas
So, so it's a, it's a, it's a very wide thing.
- SPSpeaker
What part of this business are you running?
- SSSharan Srinivas
Mindgrove Technologies is a, what is called a fabless design, uh, uh, company. So we are a pure-play design house, in that we do not do manufacturing by ourselves. We design a lot of stuff. What we do there is that we design a chip, have it manufactured elsewhere, have it tested elsewhere, but... And then we take it, and we sell the chip. So it is only the design and the selling part that, uh, we work on. Manufacturing is typically outsourced to, to, to other people.
- SPSpeaker
Okay. So, uh, Qualcomm and Nvidia are also fabless, right? They also design, and-
- SSSharan Srinivas
Yes
- SPSpeaker
... somebody else manufactures.
- SSSharan Srinivas
Qualcomm and Nvidia are very classic examples of fabless semiconductor entities. A clear example of somebody who is, who, who do- does design and also
- 8:00 – 14:00
Deep Dive: Silicon Wafers and Chip Manufacturing
- SSSharan Srinivas
does manufacturing in-house would be Intel, uh, who make their I3, I5, I7 chips, who design it themselves and manufacture it in their own fabrication plants, and to an extent, Samsung.
- SPSpeaker
I guess Samsung is also like... When you look at Samsung as a company, uh, it does, uh, phones, washing machines, fridges, like, it, it, there's so many things, right?
- SSSharan Srinivas
Yeah.
- SPSpeaker
Like, I'm guessing that some of it they do themselves, and-
- SSSharan Srinivas
Absolutely. Absolutely. So Samsung, um, uh, to the best of my knowledge today, Samsung is the largest manufacturer of memory chips in the world. Um, that's quite a heavy... There's quite, there's quite a lot of competition going on, Samsung, Micron, SK Hynix, and so on. But as, uh, as far as my, uh, recent most knowledge is concerned, and feel free to correct me if I'm wrong, uh, Samsung is the market leader in the number of, uh, memory chips, uh, that, that are being sold.
- SPSpeaker
Right.
- SSSharan Srinivas
So they don't make it by themselves. They make other components, like image sensors and stuff like that, by themselves.
- SPSpeaker
So-
- SSSharan Srinivas
Um, do they, do, do they exclusively make all the components that go into their phones? I don't really know.
- SPSpeaker
So-
- SSSharan Srinivas
Same for semi, uh, uh, consumer electronic components, but they do make a lot of their own stuff.
- SPSpeaker
Fine. So there are companies like Samsung and Intel, which design and manufacture. Then there are companies like, uh, Nvidia, Qualcomm, and Mindgrove, which design but don't manufacture.
- SSSharan Srinivas
Yeah.
- SPSpeaker
I'm guessing there are companies which manufacture but don't design.
- SSSharan Srinivas
TSMC is a classic example of that. "We don't care what you build, uh, we don't care what you design. We'll manufacture it and give it to you." What they are- they are called pure-play foundries-
- SPSpeaker
Right
- SSSharan Srinivas
... because the only thing that they do is manufacture.
- SPSpeaker
Right.
- SSSharan Srinivas
So their job is to take a piece of silicon that's like this, a circular piece like this.
- SPSpeaker
Is it this big?
- SSSharan Srinivas
No, it's not.
- SPSpeaker
Okay.
- SSSharan Srinivas
Uh, it's only about, uh, 12 inches in diameter.
- SPSpeaker
Okay.
- SSSharan Srinivas
So it'll be about as big as a typical plate.
- SPSpeaker
Yeah, so the size of this iPad?
- SSSharan Srinivas
Yeah, pretty much. 12-inch, uh, circle, uh, circle with tw- of a 12-inch diameter, and what they do is they use lasers to etch transistors and circuits onto that silicon. So they will literally... It's like taking a knife onto this, uh, wooden table and, you know, making scratches on it, and then, uh, that scratches will be filled with copper, and then you have wires. And the same way, they will scratch a shape onto it, and then those things become transistors. TSMC does that and only that. They don't say how you have to... What kind of transistors you have to use. Uh, well, they do, because the machine has to be capable of manufacturing it. Other than the structural aspects, they do not, uh, say how you have to lay out anything, how you have to connect these things together, other than what is necessary for manufacturing it using the machinery that-
- SPSpeaker
Fair enough. They may come back and say: "Look, the design you have sent is not manufacturable," or-
- SSSharan Srinivas
Yes. So there is something called a design rule check.
- SPSpeaker
Okay.
- 14:00 – 17:00
The Technology Behind 3nm vs 28nm Chips
- SPSpeaker
Yeah.
- SSSharan Srinivas
At 3 nanometers, you can do way more than 28 nanometers for the same reason.
- SPSpeaker
Yeah. This is when, when people say that, uh, when man went to the moon, the computers used were as big as a room, and now you can go to the moon with your cellphone, I mean, with the computing power of your cellphone.
- SSSharan Srinivas
Yes. The 3 nanometer system, the, the ma- the piece of machinery, it's manufactured by a company called ASML-
- SPSpeaker
Yeah
- SSSharan Srinivas
... based out of the Netherlands. Each of this, each of the, e- each individual unit, each individual machine costs about $250 to $300 million US dollars.
- SPSpeaker
Yeah.
- SSSharan Srinivas
So that's what, um, $300 US dollars is 2,400 crores, give or take, 2,400, 2,500 crores per machine. And it, it's got, it's got a laser, and producing lasers, laser light of a certain, uh, frequency and wavelength, consistently and continuously, and deterministically for an extended period of time is a giant, gigantic task by itself. When you take that and basically go have a whole bunch of arrangements of mirrors-
- SPSpeaker
Yeah
- SSSharan Srinivas
... and then you, you use optics to reduce the size of the, the wavelength of-
- SPSpeaker
Okay, okay, I've lost, I've lost you at some point.
- SSSharan Srinivas
Yeah, yeah. The point here is-
- SPSpeaker
Yeah
- SSSharan Srinivas
... I, I went that deep-
- SPSpeaker
Yeah
- SSSharan Srinivas
... just to show how complicated the machinery is.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Right? So making the machines-
- SPSpeaker
A real machine that can make 3nm chips that TSMC will buy and-
- SSSharan Srinivas
It takes apparently several 747s, entire 747s to carry this one machine from-
- SPSpeaker
I know
- SSSharan Srinivas
... uh, the manufacturing, uh, facility to wherever the fabrication plant is.
- SPSpeaker
It's very interesting. Uh, we spoke about fabs, um, we spoke about fabs, fabless, uh, we spoke about, um, integrated companies which are both fab and fabless.
- SSSharan Srinivas
Yep.
- SPSpeaker
Now you are talking about a company that makes the specialized machines that fabs use.
- SSSharan Srinivas
Yes.
- SPSpeaker
And we have just spoken about till the wafer is manufactured. I'm guessing there's much more to that before we get our phones in our hand.
- SSSharan Srinivas
Yes. So wafer is manufactured, and you have these little squares inscribed in it. Then what happens is the squares get cut-
- SPSpeaker
Mm
- SSSharan Srinivas
... and the squares get put into boxes, like you would put a piece of jewelry into, you know-
- 17:00 – 20:00
Mindgrove's Product Portfolio: SecureIoT
- SSSharan Srinivas
and the, the, the package is the enclosure, uh, holding it. Because it needs to sit in a very specific kind of environment, and the package is, um, uh, is the one that gives you that environment.
- SPSpeaker
Okay, and-
- SSSharan Srinivas
And the package connects it to the external world.
- SPSpeaker
But when you use the word OSAT and ATMP, what are these words?
- SSSharan Srinivas
Um, outsourced assembly and testing, OSAT.
- SPSpeaker
Oh, so the, the assembly of the chip and the testing of the chip? Okay.
- SSSharan Srinivas
And ATMP is assembly, testing, marking, and packaging.
- SPSpeaker
Right.
- SSSharan Srinivas
It's just the functions that those plants, uh, uh, do.
- SPSpeaker
Right. So now I have a full picture. So there are design houses, there are fabs, there are packaging plants, OSAT, ATMP, whatever they're called, uh, and, and there are all these million factories that are supporting these.
- SSSharan Srinivas
There are a million subsidiary or, or s- indirect players, the, the component suppliers of the ecosystem, if you will.
- SPSpeaker
Yeah. And, and while we were talking about this, I, I figured that you've used three company names: ASML, which is a Dutch company; Samsung, which is a Korean company; and TSMC, which is a Taiwanese company. And these three are probably one of the largest or if not the largest companies in their countries, right?
- SSSharan Srinivas
Yes.
- SPSpeaker
Can you explain the Mindgrove products, at what nanometer node they are made at?
- SSSharan Srinivas
Sure. Mindgrove's products are made at 28 nanometer.
- SPSpeaker
Okay.
- SSSharan Srinivas
Um, pre- predominantly because we, A, we are not building for the state-of-the-art, we are building for the embedded world, which do not need the, uh, the technological complexity that you would find in a smartphone.... which therefore do not necessitate us to do it in three nanometers. Frankly, it's cheaper and easier to-
- SPSpeaker
No, no, explain this to me like I'm really dumb.
- SSSharan Srinivas
Now-
- SPSpeaker
I, I understand that three nanometer is better than 28 nanometer.
- SSSharan Srinivas
In some certain ways, not in every way.
- SPSpeaker
The... Okay, let me say that the technology that goes into building three nanometer is tougher than the technology that goes into 28 nanometer.
- SSSharan Srinivas
It's been a-
- SPSpeaker
But why do I, why do I care whether it's three nanometer or 28 nanometer?
- SSSharan Srinivas
Why do you care? You open your smartphone, and you flick it.
- SPSpeaker
Right.
- SSSharan Srinivas
You see lag. Do you like it, or do you l- don't like it? Do not like it.
- SPSpeaker
I mean, if I flick it, and it flicks after 30 seconds, then yeah, that, that-
- SSSharan Srinivas
Right? You open a video, and you're looking at doomscrolling Instagram reels.
- SPSpeaker
Right.
- 20:00 – 22:00
Commercial Grade vs Technology Demonstrators
- SSSharan Srinivas
a lower node cannot, because it'll drain your battery significantly faster.
- SPSpeaker
Okay.
- SSSharan Srinivas
But-
- SPSpeaker
So higher workload, faster, and-
- SSSharan Srinivas
But
- SPSpeaker
... lesser battery.
- SSSharan Srinivas
Um, the, the three nanometer... Okay, three nanometer node will give you much more performance as we think of it, right? The three nanometer will give you what is called performance per watt of power consumed. For every one watt of power that you consume, how many things can you get done?
- SPSpeaker
Right.
- SSSharan Srinivas
And, uh, three nanometers is better at, at that, than, say, 28 nanometers. However, a three nanometer node is exponentially more expensive than a 28 nanometer.
- SPSpeaker
Okay, so for embedded systems where Mindgrove chips are used, that much is not needed.
- SSSharan Srinivas
Yes.
- SPSpeaker
Can you explain a little bit more about the Mindgrove products? We make a chip at 28 nanometer, it goes into embedded systems.
- SSSharan Srinivas
Yes. The first product that, that we've prototyped, um, earlier this year is called SecureIoT, and as the name suggests, it is meant for IoT kind of use cases with a security component on it-
- SPSpeaker
Okay
- SSSharan Srinivas
... some security component on it. Um, because security is an ecosystem thing, and we are effectively doing our part, uh, in the hardware. Uh, it's intended for stuff like biometric systems, like the fingerprint scanners or at, uh, uh, you know, access control systems that, that you would find all over the place, or your Aadhaar verification stuff. It is intended for consumer appliances. It is, it is intended for industrial control systems, as a controller. So all those kinds of use cases where certain functionalities are important, certain voltages are important, certain electrical characteristics are important, and you will not necessarily get those electrical characteristics in, in, in the higher nanometers.
- SPSpeaker
Right.
- SSSharan Srinivas
The nanometer is evolved predominantly due to smartphones, because our phones are very, very small, and we, we, we need space to keep a battery, a board, and a display. So you need to crush in as much as
- 22:00 – 25:00
Inside System-on-Chip (SoC) Architecture
- SSSharan Srinivas
possible, uh, of the, of the electronic components into, uh, as small a chip as possible, while consuming less power and giving more performance. Pretty complex that. That is the complexity that created all this nanometer thing over the past 10, 15 years.
- SPSpeaker
Right. But the sound box, the Dabba doesn't-
- SSSharan Srinivas
Don't need it!
- SPSpeaker
Yeah, the-
- SSSharan Srinivas
Don't need it.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Um, for, for those kinds of things, 90 nanometers or 65 nanometers, which, which came out in, in, in, in, like, the early 2000s, are, are way more suff- more than sufficient.
- SPSpeaker
Right, because they'll also be cheaper then.
- SSSharan Srinivas
Exactly. And, uh, you don't need more than that. You can get away with it, you get away with it. So, uh, that's what they do, and that's kind of why we chose the 28 nanometer, because we feel it's a Goldilocks node. It's got just the, uh, right amount of performance, and it's got just the right amount of, uh, optimized power consumption for that technology node. Where if we insert our design capabilities into it, where you very care- if you ca- very carefully design your circuitry, you can, you can very quickly get, uh, uh, a very compelling product out.
- SPSpeaker
Okay. Okay, so this is the chip that you-
- SSSharan Srinivas
Yes, that is a-
- SPSpeaker
So the... I saw the, uh, news around it, was India's first high-performance-
- SSSharan Srinivas
Microcontroller, yeah
- SPSpeaker
... commercial grade-
- SSSharan Srinivas
Microcontroller, yes.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Because the processor we use here is, is the Shakti C-Class processor. The Shakti C-Class processor was, uh, developed, uh, uh, in, in IIT Madras, in the RISE Lab, under the, uh, leadership of, uh, Pr- Professor Kamakoti, then Professor Kamakoti, and now Director, uh, Kamakoti. And they've done this before. It's not that it has not been done before; it has been done before. It has been done, taped out on 22 nanometers Intel before, so... but it was not commercial grade.
- SPSpeaker
Yes.
- SSSharan Srinivas
It was a technology demonstrator.
- SPSpeaker
What is the difference between a technology demonstrator and a commercial grade?
- SSSharan Srinivas
Prove that you can do something, and prove that you can do something that can be sold. It's the same difference as what you would find in an R&D project and a commercial grade project.
- SPSpeaker
Right.
- SSSharan Srinivas
Not everything you need to do, do commercial grade would be quantified as R&D.
- SPSpeaker
You mentioned the Shakti C-Class processor.
- SSSharan Srinivas
Yeah.
- SPSpeaker
Um, but I've seen that, uh, the SecureIoT is referred to as a SoC.
- SSSharan Srinivas
Yes.
- SPSpeaker
So what is the... What is an SoC?
- SSSharan Srinivas
System on a chip.
- SPSpeaker
Okay.
- 25:00 – 27:30
Vision SoC: Next Generation Product
- SSSharan Srinivas
right? The, the, the, uh, the entirety of, uh, the SecureIoT SoC would be like, um, uh, single percentage points of the Snapdragon.
- SPSpeaker
Mm.
- SSSharan Srinivas
So it's like for, fi-
- SPSpeaker
In terms of number of transistors?
- SSSharan Srinivas
I- in terms of the, uh, uh, physical area that the, the chip takes, the size of the silicon die.
- SPSpeaker
And how big is the actual silicon wafer from Mindgrove or for SecureIoT?
- SSSharan Srinivas
Uh, that is something that I'm not allowed to say.
- SPSpeaker
Okay, done. I won't ask you.
- SSSharan Srinivas
But I can tell you, but I can tell you that after packaging, it's only as big as a... After packaging, it's only as big as a fingernail.
- SPSpeaker
Nice. So that's your current product, and so what is it going to be moving forward?
- SSSharan Srinivas
The next product is, again, we have a tendency to very literally name our products. It's called Vision SoC, and as the name very, very... hopefully, very clearly and, uh, by default points out, it is an SoC built for vision, computer vision use cases. So CCTV cameras, infotainment systems, and anything and everything that, uh, which can, which can take images from a camera, do some sort of processing on it, and either send it outside for further processing or put it to a display system, right? So that is, uh, the, the next chip that, uh, that we are working on, and this is the one for which we have approval from the Design Linked Incentive Scheme of the government of India. Uh, this is a multi-core, uh, there are ins- SecureIoT is a single core. There is only one Shakti C-class processor. This will have at least four. So the system can have more than a set of processors. So like, for example, a Snapdragon will have four processors to, to, to run Android, but there will be one very, uh, very much dedicated to processing videos, right? Encoding and decoding videos. There would be... There could be one processor that is dedicated to processing audio. There could... There, there's, there is a GPU, which is a graphics processing unit, and the graphics processing unit is usually dedicated to, uh, uh, putting stuff on display screens, as the process called rendering. So there are several processors in the system. There's not just one. SecureIoT has one, uh, but that's a matter of design. They can have more.
- SPSpeaker
Right.
- SSSharan Srinivas
And Vision SoC will have more-
- SPSpeaker
Okay
- SSSharan Srinivas
... for different purposes, uh, tuned around. It's a multi-core, uh, quad-core, um, as we, as we are designing it right now. So the design is not fully frozen. We're still talking to customers,
- 27:30 – 31:00
India's Semiconductor Mission
- SSSharan Srinivas
uh, figuring it out, and it will end up being the first of a family of processors. So the way we envision it, SecureIoT and Vision SoC will be a class, not just a chip. We only have one chip of SecureIoT out now, and soon enough, we'll have, uh, one s- uh, chip from Vision SoC. But we will continue to make-
- SPSpeaker
They will have variants and-
- SSSharan Srinivas
Yes
- SPSpeaker
... uh, and further improvements.
- SSSharan Srinivas
Exactly.
- SPSpeaker
Nice. So one is, uh, one is looking at the IoT industry, one is looking at the vision- embedded vision systems industry. Um, amazing, and of course, your mission is aligned to India's larger semiconductor mission, the ISM mission, and, uh, it's interesting to me, uh, uh, looks like of late, a lot of countries are investing very heavily into this. Of course, US through the CHIPS Act, and, uh, Japan, Germany, uh, India. What is the largest... Why are so many countries so, um, uh, gung ho about being, uh, overinvested in semiconductors?
- SSSharan Srinivas
As I said earlier, the semiconductor industry as a whole underpins everything digital, and we are moving everything paperless, right? We get our government services online. We pay online. I, uh, I, I literally went to a vegetable vendor, bought carrots, and paid with, uh... And asked him, I gave him cash, and he's like: "I don't have change, uh, GPay." Right, so a QR code for, for, uh, 250 grams of carrot. That's where we stand today. That's a great thing, but in order to sustain that, in order to con-
- SPSpeaker
That's a great use case. It's using vision, it's using cameras, it's using a phone, it's using satellite data-
- SSSharan Srinivas
Yes
- SPSpeaker
... it's using telecom chips.
- SSSharan Srinivas
Yes.
- SPSpeaker
Uh, it's like, uh, uh, it's using cloud server chips. It's like one use case, which covers a huge set of things, uh, for whatever, like 30 rupees of carrots.
- SSSharan Srinivas
An event as simple as buying 250 grams of, uh, carrot from a vegetable vendor in an open market requires that many chips. Would you, as a state, allow almost all of that to come from outside the country?
- SPSpeaker
No.
- SSSharan Srinivas
That is why countries are, uh, uh, gung ho, as you put it, to have at least a portion of that built in-house. It's the new oil, and for a country like India, the import bills are going to be, if not already, way more than importing oil.
- SPSpeaker
Sure. I read somewhere that already the import of semiconductor components is about 2% of all our imports, or 1.5%, 2% of-
- SSSharan Srinivas
And it's gonna go up, so quite heavily, because we are trying to do everything digitally. We are paying for everything with a QR code, and that's gonna go up. So that alone kind of justifies, uh, the need for a local ecosystem. Now, the India Semiconductor Mission is-... has other connotations also, right? We are, what, 140, 150 crores population, and I read somewhere online that about 60% of our current population is working age, and going to be about 65% in the next- over the next five years or so. Who's gonna give them jobs? If you're importing most of it, who's gonna give them jobs? And, and, and we are training a lot of engineers, and it's not
- 31:00 – 35:00
Job Creation and Economic Impact
- SSSharan Srinivas
as if you, [coughs] you can give, um, a lot of low-end jobs to, to, to people to get by. You have really qualified people. That means you have to give jobs, and you have to create and sustain jobs at a level that matches the educational capability of these people.
- SPSpeaker
Does the semiconductor industry create that many, uh, skilled jobs?
- SSSharan Srinivas
Directly? Fabrication plants, no. They'll... Maybe every fa- fabrication plant can probably... A reasonably sized plant can do about thousands of jobs.
- SPSpeaker
Okay.
- SSSharan Srinivas
Uh-
- SPSpeaker
Oh, but there are all the support industries and the-
- SSSharan Srinivas
Exactly
- SPSpeaker
... industries. Okay.
- SSSharan Srinivas
You, you shouldn't look at a fabrication plant probably as a single, uh... as, as a job creator all on its own. You should look at it as a seed, right? And then you have to bring in the ecosystem around it. If, if I have a fab, somebody has to supply, uh, all these components to the fab. Somebody has to maintain this.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Somebody has to be security in that fabrication plant. Somebody has to feed those people, catering. Something has to realize that, right? Somebody has to bring them- provide them drinking water, and somebody has to collect taxes and ensure that these paper- people are paying taxes on time, government side. Somebody has to audit it, right? So if you look at it from that perspective-
- SPSpeaker
Right. The overall point is that, uh, r- reduce import bill, create an industry which supports a lot of new jobs. Uh, and majority of those new jobs are actually skilled jobs, which are coordinating to the kind of education that-
- SSSharan Srinivas
Yes
- SPSpeaker
... Indians are already getting.
- SSSharan Srinivas
The world is changing, and we also have to, uh, look at how we change our systems, uh, to, to keep pace with the world.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Everybody, everybody does GPay, right? I don't even take cash out these days. I don't have cash. I have a credit card, and then I... And, and which I barely use, and then I, uh... I mostly pay everything on my phone. It's just QR code everywhere. So if you are going to- if that is the way you are going to consume services, the rest of the back end sy- so-called back end system must be able to, uh, sustain that way of life. How do you go about doing that? How do you generate those jobs? How do you get those people up and running? And IITs, institutions like IITs, play a huge role in, in, in doing those kinds of things. That's where building gets so important. There, there is no one company in the world... Uh, 60% of our population is eighty- 800 million people.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Of, of that, say, let's, let's assume that about, uh, what? 80, 80 is 10%, eight, 8 million people is 1%. Let's assume 1% of that population graduates college every year and is looking for first-time job offers. You have to create 8 million jobs every year.
- SPSpeaker
Mm.
- SSSharan Srinivas
Who's gonna-
- SPSpeaker
Viable jobs.
- SSSharan Srinivas
Yeah, 8 million jobs that is commensurate with the kind of skill that people have.
- SPSpeaker
And, uh, the... Not just the skill, also the expectation, and the dreams, and the-
- SSSharan Srinivas
Everybody wants to grow. So, so to that end, forget the semiconductor industry. Keep it aside for now. In general, there is no one entity that can create that many jobs. The, the, the era of government-backed, you know, sarkari naukri is, is done.
- SPSpeaker
Yeah.
- SSSharan Srinivas
So the entrepreneurship is the way to go for that, in my opinion, at least.
- SPSpeaker
Makes sense. Uh, and, and just, uh, because you mentioned it twice now... So first, you mentioned that, um, the IIT Madras, uh, RISE Labs, R-I, R-I?
- SSSharan Srinivas
R-I-S-E.
- SPSpeaker
RISE.
- 35:00 – 36:00
Technology Innovation & Software Integration
- SSSharan Srinivas
corny, RISE Lab is a lab.
- SPSpeaker
Okay.
- SSSharan Srinivas
It's literally a laboratory where people come, um, uh, research work is done, uh, on, on, on chip design, um, amongst other things. Because RISE Lab is, is more of an agglomeration, where several professors come together and work on different aspects of the same thing. A lot of students, uh, participate in that. They do projects, they do their master's, uh, projects, they do their thesis, uh, or they do, uh, coursework, course projects, BTech projects, that kind of a thing.
- SPSpeaker
Right.
- SSSharan Srinivas
A lot of stuff gets built there.
- SPSpeaker
Okay.
- SSSharan Srinivas
And because of the different... fact that different kinds of professors working on different aspects of a larger domain together form the, the lab, it's not... You're, you're not deep diving into one thing. By- by the very nature of the lab, you have breadth as well. So there's always a, an element of coordination, cooperation, which is absolutely necessary if you're doing semiconductors. You can have depth in one thing, but without breadth, without
- 36:00 – 37:00
Industry-Academia Bridge: From Lab to Market
- SSSharan Srinivas
sufficient knowledge of the other things, you're not going to be able to understand what the other expert is telling you. You won't be able to grasp the significance of what's going on. And the RISE Lab, as it is built, will give you that kind of exposure, and that's very, very valuable when you go into industries like this.
- SPSpeaker
And it's a, uh, chip design lab, right?
- SSSharan Srinivas
Yes, amongst other things. They do a lot more than that. So they, they... It's not just the design of a chip, right? You have to be able to run software on it. You have to be run, able to run all the standard open source stuff, like run the, run the GCC compiler, or C compiler, or Python com- uh, programs, or Rust programs, or whatever on this processor, on this design that you're doing, and then you measure this performance, like, the, with, with, with standardized benchmarks and say... So these people will, will make an innovation in design, will work on and make an innovation in design.... they will, uh, show, demonstrate that, uh, hey, you can take all the standardized stuff where- that everybody in the world uses, open source stuff, uh, run it on these processors,
- 37:00 – 40:00
Inside RISE Lab's Ecosystem
- SSSharan Srinivas
and look, it's now- it's, it's, it's as good as anything else in the world. That's what I meant earlier when I said they have done a technology demonstrator.
- SPSpeaker
Right.
- SSSharan Srinivas
Right? What we do in, in, in Mindgrove is take that, "Ah, yes, this is done. For all this application, I will make the experience of you utilizing it much better."
- SPSpeaker
Makes sense. I mean, the role of academia is to, uh, invent stuff, discover stuff, put things together, prove that it can work, and then, a startup or an industry has to take that and sort of... I love this thing where your office is in the research park, the IIT Madras Research Park, which supports a lot of startups, and I love this thing where the- it's next to the IIT Madras, uh, campus, and there's a little bridge, and on top of the bridge it says-
- SSSharan Srinivas
Industry-academia bridge.
- SPSpeaker
Industry-academia bridge, which is-
- SSSharan Srinivas
Yes
- SPSpeaker
... cute. But I mean, it's quite literal-
- SSSharan Srinivas
Yes
- SPSpeaker
... uh, that you're crossing over from an academic institution to a, a... And it's almost like a bridge from, uh, the Shakti C-Class, uh, uh, demonstration tape-out-
- SSSharan Srinivas
Mm-hmm
- SPSpeaker
... the SecureIoT chip.
- SSSharan Srinivas
The- literally, right? The industry-academia bridge starts at the road, on the ground-
- SPSpeaker
Yeah
- SSSharan Srinivas
... and, uh, connects to the second floor of research parts, uh, park's, uh, parking lot. Mm, that's in a sense what IIT Madras represents and does. That is what a technological de- uh, a technology demonstrator is. You don't have to start from scratch, ground up, right?
- SPSpeaker
Right.
- SSSharan Srinivas
They give you a baseline showing that, "Hey, this is possible," and take it all the way up to second floor. From second floor to whichever block your, your office is in-
- SPSpeaker
Yeah
- SSSharan Srinivas
... that's on you.
- SPSpeaker
Yeah.
- SSSharan Srinivas
That's, that's the part called commercialization.
- SPSpeaker
Yeah, correct. There also, IIT Madras s- uh, support is quite extensive. The incubation cell is there. Uh, there are multiple incubation centers now.
- SSSharan Srinivas
Yes. So the way I understand it, IITM IC is an umbrella, and there are several specific, uh, incubators. There's one for rural technology, there's one for health technology, there is Pravartak, which is for cyber physical systems, so on and so forth. But there are a lot-
- SPSpeaker
There's healthcare, there's a fintech incubator, there's-
- SSSharan Srinivas
Yes, so there's, there's a lot.
- SPSpeaker
You are part of the... which incubator?
- SSSharan Srinivas
We are co-incubated, so e- everybody who works in here is, is- comes under IITM, uh, Incubation Cell. And under that, uh, we are incuba- co-incubated by IITM Pravartak Technologies Foundation, which is the incubator for cyber physical systems. And, well, we're building embedded systems that go into... that can go into industrial control systems, and robotics, and sensors, can use sensors, actuators, you know, uh, con- and other such industrial control systems. So we smack bang definition of, uh, uh, the domain of that incubator.
- SPSpeaker
There's an acronym for it, or for what was-
- SSSharan Srinivas
SNACS.
- SPSpeaker
SNACS. Hmm.
- 40:00 – 41:00
Startup Journey & Academic Balance
- SPSpeaker
number of people to put together something this big. All of them must have a builder's attitude. There has to be an environment which-
- SSSharan Srinivas
RISE Lab has about 50, 60 people in it, of various, uh... There are project staff, there are contractors, there are project staff, there are project advisors, there are students, there are-
- SPSpeaker
Yeah
- SSSharan Srinivas
... BTech, masters, PhD students, a whole bunch of-
- SPSpeaker
You're yourself a PhD scholar?
- SSSharan Srinivas
Yes.
- SPSpeaker
Why? That's why, isn't it? [chuckles]
- SSSharan Srinivas
Well, I'm also a co-founder, chief architect, you know, wearing multiple hats as a startup, uh, founder.
- SPSpeaker
Fair enough.
- SSSharan Srinivas
So I also do a PhD, is what I say.
- SPSpeaker
I'm guessing that you'll probably take longer to finish. [chuckles]
- SSSharan Srinivas
Oh, yeah. Oh, that, that is a given. [chuckles] Uh, there was this, there was this time when professor called me up, and he's like: "Hey, are you planning to finish your PhD or not?"
- SPSpeaker
Who, who professor?
- SSSharan Srinivas
Oh, Professor Kamakoti.
- SPSpeaker
Professor Kamakoti is your PhD-
- SSSharan Srinivas
Oh, yeah, he's my PhD, uh, supervisor, which is fun. Thankful that he's, he's my PhD supervisor, because it is because of him that I'm able to, uh, run a startup and, uh, continue,
- 41:00 – 46:00
Government Support & Institutional Dynamics
- SSSharan Srinivas
uh, the, the sweet dream of doing a startup and a PhD at the same time. Um, so he, he, he basically gave me the independence to do whatever I want. I'm part-time, so I, I don't- I'm not funded. Uh, so he was like, "You can..." So there's no project restrictions, and you have to work- choose your problem within the confines of the prod, uh, project, and so on. "So you can do whatever you want, as long as it is within the ambit of RISE Lab, and you justify it to your doctoral committee." So I'm like, "Okay, that's cool." Uh, that, that allows me to do a lot of esoteric things, uh, which is something that I am trying to do. But yes, uh, it's not all that easy to do, so...
- SPSpeaker
Fair enough. Just thinking about this thing that you spoke about earlier, that... So there's a lot of government support, you're a DLI-funded startup, uh, you're working in an academic- co-academic setting. Um, h- how is it like working with institutions like government or semi-government industry? Because I'm guessing it's like, from what I've heard from other entrepreneurs, sometimes it's quite hard. People have mixed opinions, uh, because of the scale at which things get done, because of the speed at which things get done. How has your experience been?
- SSSharan Srinivas
Shashwat and I... Uh, Shashwat is the CEO and co-founder of Mindgrove. Shashwat and I have a tendency to bucket- have three buckets of speed. There is IITM speed, there is startup speed, and then there is government speed. Now, by the metric of government speed, this government especially, our experience working with government for the India Semiconductor Mission has been fantastic. IIT Madras is a huge organization. There's, like, what? Some 18,000 people in it, including all the staff, the faculty, residents of the campus, the students, the professors, supplementary, uh, non-teaching staff.... uh, and, and so on. So with, with that kind of manpower and that level of size of operations, things do slow down. There, there, there is a necessary set of checks and balances that have been play- uh, that-
- SPSpeaker
Right
- SSSharan Srinivas
... that need to be placed for safe and proper, uh, operation of an institution this big. That slows things down, right? Startups don't have that. Startups are famous, or rather, in certain cases, infamous for not having control. We prefer to be on the famous part, not the infamous part-
- SPSpeaker
Yeah
- SSSharan Srinivas
... just for sake of clarity. Uh, so startup is the fastest, and IITM is faster than the government. But the government, if you look at the government, they're hell-bent on getting the semiconductor thing done. I mean, it's not just words. The, uh, the, the policymakers have been very open about it. They've come to conferences, uh, uh, press interactions, where they have basically openly said that, "We'll do whatever it takes to ensure that the ecosystem takes hold and becomes self-sustaining," to the, to, to the point of using those exact words. So kudos to them for that, for thinking like that. And it's not just... Our experience has been that it's not just words. Our experience, especially interaction with DLI, is that they have been very engaging, they've been very open to feedback, and, and there was- there were times when we have very vociferously complained, if you will. Uh, and, uh, our feedback was actually taken very quickly, and they, they, they have done everything i- w- within their ambit of power to alleviate problems, remove obstacles, pa- obstacles to pathways, and in general, ensure that we are able to... Uh, they, they set up an environment for us and get out of the way so that we can go and execute. That's government. That's the slowest of the lot. So you can see just how fast a startup works, and at that speed, a government is not gonna be able to match up. If you think IIT Madras, with its 18, 20,000 people, is going to be... needs its checks and balances, this is government.
- SPSpeaker
Yeah, many more checks and balances.
- SSSharan Srinivas
Right. So it's going to go through layers of approvals before anything gets done. That is the nature of government anywhere in the world.
- SPSpeaker
Yeah.
- SSSharan Srinivas
So the question is, how do you handle it? You handle... We have been handling it by being systematic to the point of paranoia. The government will say this is sufficient, but we will look up the law, and it's not just tending to the spirit of the agreement. Like, they say, "This is absolutely necessary. Anything more than that would be advantageous, but not nearly..." And if you have, you can give.
- SPSpeaker
Right, not just to the letter, but to the spirit.
- SSSharan Srinivas
Yes, but we have been doing it to the letter. What that has done-
- SPSpeaker
Isn't the other way around? The, the letter is the-
- SSSharan Srinivas
The letter says-
- SPSpeaker
... concerned about the letter
- SSSharan Srinivas
... the letter says certain things are mandatory.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Statutory, you have to do this. Certain things are good to do, they're optional, right? Uh, you
- 46:00 – 49:00
Breaking Records: 8-Month Chip Design
- SSSharan Srinivas
want to be-
- SPSpeaker
So you go well beyond that.
- SSSharan Srinivas
We go well beyond that.
- SPSpeaker
Understood.
- SSSharan Srinivas
And we have a s- we have a team, we have a team specifically, uh, set up to deal with government. Uh, it takes a team to deal with government. I mean-
- SPSpeaker
I mean, deal with the requirements and-
- SSSharan Srinivas
Yes
- SPSpeaker
... complexity.
- SSSharan Srinivas
They have, they have... They, they are, they are answerable to a whole f- they are answerable to parliament, they are answerable to... If they are f- if they are handling finance, they are answerable to the appropriate, uh, financial bodies, they are answerable to the people.
- SPSpeaker
And it is because of the kind of government support and, uh, IITM support that you're getting, you're able to move really fast, right? Like, because I, I, I remember you saying somewhere that it took, what, six months to design the first chip, or was it one year to design the first chip?
- SSSharan Srinivas
It, it took... Okay, it took about, uh, eight months-ish to execute the design of the first chip. Execute the design of the first chip.
- SPSpeaker
Right. How much is that in a normal, like, say, in Qualcomm or some large company?
- SSSharan Srinivas
Uh, if you want to build a Snapdragon, something of the scale of a Snapdragon, five years.
- SPSpeaker
Oh.
- SSSharan Srinivas
Five years from scratch, because that's what we did. We are not... Not disclaimer, we have not built anything resembling the levels of sophistication of a Snapdragon, so it's not-
- SPSpeaker
Understood. But what you're saying is that you start from zero, uh, whereas a company like, uh, Qualcomm, if it decides to build a new chip, they, they have some body of work which they're carrying forward.
- SSSharan Srinivas
Yeah, but we spent one year just... Shashwat and I spent about a year just figuring out what to build.
- SPSpeaker
Right.
- SSSharan Srinivas
We didn't have any employees. It was just the two of us, uh, and, and our COO. So late, he became COO later, Uma. We were just sitting and just throwing ideas across to each other, just figuring out what we- what to do. So by the time we got the tools, by, by the time we got seed funding, by the time we were... We, we, we did... We got the market research up and running, we had conversations with customers where we realized if we, we, if we do this, we have the highest possible probability of getting a product to market fit.
- SPSpeaker
Mm.
- SSSharan Srinivas
Right? We had already done that work beforehand, so execution was eight months.
- SPSpeaker
Right.
- SSSharan Srinivas
Eight months flat.
- SPSpeaker
Got it.
- SSSharan Srinivas
So other-
- SPSpeaker
Clear what you wanted to do, you had the exact support, you had the Shakti processor.
- SSSharan Srinivas
And then we had all testing equipment. This is where IIT Madras comes into play.
- SPSpeaker
Yeah.
- SSSharan Srinivas
There's an enormous amount of infrastructure available for students and other people to use.
- SPSpeaker
Yeah, through Pravartak also.
- 49:00 – 52:00
Behind the Scenes: Work-Life Balance
- SSSharan Srinivas
more expensive than university licenses.... especially depending on what kind of, uh, tools you want. The more sophisticated your chip, the more sophisticated tooling that is, that is necessary, and the cost growth over there is exponential in relation to sophistica- levels of sophistication.
- SPSpeaker
Right. It's, this sounds like quite a difficult life, like, you're, you're, you're the architect of the chip yourself. Uh, it's not like you're a hands-off engineer. You are the engineer on it. You are, uh, a PhD student-
- SSSharan Srinivas
On paper, at least
- SPSpeaker
... a part-time PhD student-
- SSSharan Srinivas
Yeah
- SPSpeaker
... uh, and a CTO building a company. So like, how do you balance all of this? You're also newly married, uh-
- SSSharan Srinivas
Well, yes.
- SPSpeaker
Relatively.
- SSSharan Srinivas
Yes.
- SPSpeaker
Um, so how do you manage all these, you know-
- SSSharan Srinivas
That's a very... That, that, that last part is a very dicey question, so I will [chuckles] not field that.
- SPSpeaker
Sure. Uh, but let's talk about work-life balance, right?
- SSSharan Srinivas
Yeah.
- SPSpeaker
Like, how do you, how do you figure how to give time to what to prioritize?
- SSSharan Srinivas
Sure. Well, anyway, I was just joking. I, I've been married since 2018. My wife is also a PhD student-
- SPSpeaker
Oh, wow
- SSSharan Srinivas
... at IIT Madras.
- SPSpeaker
Right.
- SSSharan Srinivas
Uh, and she's part of the answer to your question. My wife and my mother, um... my father is no more. My, my wife and my mother have given me unconditional support, and heavy emphasis on the word unconditional, right? Whatever you go, if I say I have to go for this, regardless of what I've been doing, regardless of the consequence of dropping something in the middle and running away, they'll fix it, they'll figure it out, they'll, they'll clean up after me. And I've heard Shashwat saying the same thing. He's got two young children.
- SPSpeaker
Yeah.
- SSSharan Srinivas
And-
- SPSpeaker
Oh, even one.
- SSSharan Srinivas
Four and ma- four and slightly over 12. Bit over tw- uh, uh, somewhere around 15 months.
- SPSpeaker
Ah, so, so a, a child and a toddler?
- SSSharan Srinivas
Yes. Fun.
- SPSpeaker
[chuckles]
- SSSharan Srinivas
Every, every, every parent of young children, uh, knows that, and, you know, I, I see all that's happening, and I'm like, "Okay, I'm looking..." Uh, I, I, I don't have any yet, but I'm definitely looking forward to that, uh, aspect of life. But family support is foundational. If your family is not supporting you, you, you cannot do this. It's just impossible to do, do... Forget the PhD and startup part, you can't do w- either one or the other. I've seen enough PhD students who started with me, especially unfortunately female students, who got married, uh, after joining a PhD, or who came after marriage to join a PhD, and who quit because they've had family problems.
- SPSpeaker
Right, because PhD itself is quite hard.
- SSSharan Srinivas
It's very hard. It, it is not a... It's, it's not something that... It's a process of learning how to get insight into a domain.
- SPSpeaker
Right.
- 52:00 – 54:00
The Birth of Shakti Chip: "It'll Come When You Make It"
- SSSharan Srinivas
click, and you'll do something, you'll write your thesis, you'll publish it, and you'll get some insight, and done. And that's why I said there are people who are working, like, three times as hard as I do-
- SPSpeaker
Yeah
- SSSharan Srinivas
... and don't have the, uh, platform to get that exposure, and that's, that's kind of why I'm kind of chipping in for them.
- SPSpeaker
Yeah. And, and the subca- subtext here is that this is... You're talking about PhD in a, in one of the most coveted, uh, computer science, uh, department and labs in the country, right? So it's, it's hard.
- SSSharan Srinivas
It's hard. It's very hard.
- SPSpeaker
Sure. So, um, we spoke about hard, uh, and you said that one of the foundational support is that you have a lot of support from your family. Um-
- SSSharan Srinivas
You also need support from your PhD guide. Your PhD guide can make and bre- make or break your career. So that is... That's why I'm very grateful to Professor Kamakoti, if not for... Well, he was the one who made me do the, the, the Shakti chip. If you can make it, it will come.
- SPSpeaker
What is the story? Sorry.
- SSSharan Srinivas
I will get to that.
- SPSpeaker
Okay.
- SSSharan Srinivas
But he, he's allowed me to go explore. And I don't see... Today, unfortunately, the, the process of getting a PhD paper or degree paper, amounts to you publishing something in, in a particular type of a journal or a conference-
- SPSpeaker
Okay
- SSSharan Srinivas
... and presenting it and getting, getting the approval of-
- SPSpeaker
Your doctoral thesis committee
- SSSharan Srinivas
... an appropriate- yes, an appropriate group of, uh, people who are empowered to do that.
- SPSpeaker
Sure.
- SSSharan Srinivas
But if you look at what the spirit of what a doctor of philosophy is, philosophy is all about gaining insight. So it's, you, you don't have to figure out the, the, why life is the way it is, right? You don't have to, you, you, you just enjoy it. Uh, but you're, you're trying to figure out the insight into an engineering problem. You're trying to figure out an in- a particular piece of insight into a domain, and a PhD is one way of doing
- 54:00 – 57:00
Finding the Right Market Position for Shakti
- SSSharan Srinivas
it. It's not necessarily the... Off late, I've come to the conclusion that it's not necessarily the only way of doing it, and that particular piece of insight itself comes from what professor did. So we didn't start off, uh, to build out chips, we started out building systems.
- SPSpeaker
Mindgrove started off.
- SSSharan Srinivas
Mindgrove started. We were like, "We will... You- I will do the hardware, I, I, Shashwat will do, do the hardware, system hardware, PCBs, and that kind of a thing. I'll do software." I am a computer scientist. I am not a hard- an electronics engineer by training. Excuse me. That's where we started, and, and we realized that the chips, that this is smack bang middle of pandemic, 2020, 2021, and we were trying to figure out what to build. We were looking at, uh, components that we could take to build. So we realized, after a whole bunch of, uh, searching and discussions and all of that, that Shakti was perfectly placed right at the center. It was not too performant, it was not too powerful. Things that have extreme performance are usually they consume a lot of power to do that performance.... right? Uh, the, uh, the ones that are super power efficient didn't have the necessary performance, and you needed something in the middle, and Shakti was sitting in that middle. So we went to professor, who he was not yet director at that point, and we were like, "Sir, when can we get Shakti processor as a chip?"
- SPSpeaker
For your, uh-
- SSSharan Srinivas
Yeah
- SPSpeaker
... product that you had imagined?
- SSSharan Srinivas
Exactly. And just signing a bunch of papers, and then he was like, "It will come when you make it."
- SPSpeaker
[laughing]
- SSSharan Srinivas
That single sentence, we were like, "Huh?" You, you don't make a chip just like that. Where's the tools? Where's this? Where's that? Where's this? Where's that? And it took us a while to get through... I, I mean, I'm not bombarding those questions, but it took us a while to figure out all those questions.
- SPSpeaker
Yeah.
- SSSharan Srinivas
And we just kept going to him, "Hey, you, you asked us to s- you are telling us to do the chip. Sir, how can we do this chip without this?" And then he kept answering each and every one of those things. Right?
- SPSpeaker
Interesting.
- SSSharan Srinivas
And his, his standing was, "Look, IIT Madras is built as a research organization. It's very, very well set up to prove that a technology exists, to create technology, to engineer, and to innovate. It is not set up to commercialize."
- SPSpeaker
Right.
- SSSharan Srinivas
So his standing was, "Look, we've proved this technology. We showed this is possible. We showed, uh, i- it is no- it is- there is no compromise. It is no less performance, uh... no less performant than any other equivalent in the world. Now you have to take it and do stuff with it."
- SPSpeaker
Fair enough. He said-
- SSSharan Srinivas
Set it free
- SPSpeaker
... we need a partner on the other side of that bridge, and if none exists, then you just have to be that partner. You have to be-
- SSSharan Srinivas
No. He's like, "You also be that partner."
- SPSpeaker
Ah.
- SSSharan Srinivas
"You can be one of the people who can take it and do something with it. You do something, whatever you want. I'm not telling you to do this or do that.
- 57:00 – 59:00
The Reality of Making Silicon: A Billion-Dollar Industry
- SSSharan Srinivas
You do what you, what you think you can do with it, but do it. Because if you don't do it now, we'll make another 300 pieces of this chip-"
- SPSpeaker
Yeah.
- SSSharan Srinivas
" - I will end up framing two or three and putting it in s- in, in, on the wall in RISE Lab, and the remaining, uh, 297 odd chips will sit in, uh, sit in, sit in, um, cupboard. And whenever some dignitary visits, we'll open, dust it, show it, and put it back in." [chuckles]
- SPSpeaker
I want to just clarify here, you are not the only startup that's come out of RISE Lab, right? There, there are multiple-
- SSSharan Srinivas
There are, there are multiple startups. Uh, uh, they are- they're working on different aspects of the thing.
- SPSpeaker
I'm sure that Professor Kamakoti has said that line, "No, when you make it," to multiple startups or multiple people.
- SSSharan Srinivas
See, making silicon is hard.
- SPSpeaker
Okay.
- SSSharan Srinivas
You have to be a special, uh, grade of mental-
- SPSpeaker
[chuckles]
- SSSharan Srinivas
... to, to be, to able, to be able to want to make silicon in today's world. Who else makes silicon? Think about it. Let's see. NVIDIA. What's their market cap again?
- SPSpeaker
[chuckles] I don't know. A few trillion dollars, um-
- SSSharan Srinivas
Largest company in the world.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Done. Who else makes silicon? Apple. Whole different league.
- SPSpeaker
Right.
- SSSharan Srinivas
Who else makes silicon? Intel. Let's forget their recent problems. Intel was-
- SPSpeaker
Intel, 60, 70-year-old company. Yeah.
- SSSharan Srinivas
Not just that, they make $60, $70 billion in revenue every year-
- SPSpeaker
Right
- SSSharan Srinivas
... and, and, and their profit after tax is something like 20, 25%. NXP Semiconductors, $14, $15 billion in revenue.
- SPSpeaker
Right. I-
- SSSharan Srinivas
Their R&D spend is a billion dollars.
- SPSpeaker
You're basically saying that this is a business where the incumbents are really, really large. They are all billion, trillion dollar companies, so it's-
- SSSharan Srinivas
And then that, that, that creates a question: Why are all silicon making companies that large?
- SPSpeaker
It's fair to say that maybe because it takes that much money to-
- SSSharan Srinivas
Like, I was telling Shashwat, "Hey..." He was like, "We, we are, uh..." We were dreaming, right? And we were like, "We, we
- 59:00 – 1:01:00
The Journey from Research to Commercialization
- SSSharan Srinivas
should, we should do this much in seed, this much in Series A, get that much, that many million dollars in revenue, do IPO, and become a unicorn." And I'm like, "I'll give you $100 million to spend. How long will you take it, uh, will, will it take you to spend?" And I was like, "Five minutes flat."
- SPSpeaker
Hmm.
- SSSharan Srinivas
That's all. It, it literally takes only five minutes to spend-
- SPSpeaker
It's that easy to spend in, uh, in that kind-
- SSSharan Srinivas
I just spend, I just splurge on license fees and do it in seven nanometer, 100 million, boof! Instant.
- SPSpeaker
Understood.
- SSSharan Srinivas
That's the problem with the industry. You have to be very careful when you're designing, and that's why it, it becomes very hard, because it's very tempting to do it that way. But if you dump in that much capital, you have to have a way of taking all that capital out and making a profit within a reason- and breaking even in a reasonable time span, which investors would agree.
- SPSpeaker
Right.
- SSSharan Srinivas
Do I have $100 million to put in of my own money? Nope. Shashwat?
- SPSpeaker
If, if-
- SSSharan Srinivas
Nope. Nobody else. So we, you, you... This is where, uh, venture capital, the venture capital industry of today comes into play. It is mature to a point where we can actually go make this kind of a pitch, and then they will not laugh at- laugh us out just on the sheer principle of-
- SPSpeaker
Quite opposite, right? They're saying, "We believe in you, and-
- SSSharan Srinivas
Yeah
- SPSpeaker
... let's do it."
- SSSharan Srinivas
Forget, forget selling and all of that. Making a chip is by itself a big deal. Uh, designing it, taking it to production, getting it back, and getting it to work is a big deal. What is the risk? The risk is utter and absolute failure, if it fails, if it completely screws up. And there are also, there are also the venture capitalists, from what I've seen, are also the kind of people who look at the positive aspect.
- SPSpeaker
Yeah.
- SSSharan Srinivas
"Oh, there's so much risk here, it could fail." Yes, they do look at that, and they do look at aspects for mitigating that kind of risk. And this is something I learned from Rajan, Rajan Anandan, when I was, uh, speaking to him, and he asked me this, uh, Shashwat and I, this question: "What will happen if this thing succeeds?"
- SPSpeaker
Hmm. It's very interesting.
- SSSharan Srinivas
And then we just told him the process, and then he kept digging and digging, and
- 1:01:00 – 1:02:00
The IIT Madras Ecosystem: Support Systems and Infrastructure
- SSSharan Srinivas
we kind of, uh, uh, arrived at TAM, we get kind of arrived at service available, available market, potential revenues, and so on. And then he was like, "Look at the positives, look at the negatives," right? "If this kicks off, it's gonna be huge. If it doesn't, you're gonna be one among the 90% of startups that fail."... And that is the kind of risk-taking that's necessary for this kind of an ecosystem, unfortunately. And fortunately. Fortunate- the fortunate part is that there are people who are willing to do that. There would be- there are people who are playing in a smaller scale, with semiconductor experience, uh, who, who, who are doing those kinds of things.
- SPSpeaker
I'm just thinking in my mind, for, for somebody in IIT Madras to dream that, "I'm gonna build this," uh, and we say best place to build for a reason, right? Like, it requires so much support. It requires something to have happened in an academic setting for you to build on top
- 1:02:00 – 1:04:00
Understanding and Handling Failure
- SPSpeaker
of. It requires very supportive professors who are building flexible systems, um, a research park or a incubation center which provides you s- sort of access to tools, people, gives you a l- that level of validation. Uh, maybe just pats you on the back and says, "You know what? We'll go through this journey together." And the investor community, like Speciale, uh, Peak, and all these other companies which are coming here and saying, "You know, we are looking into technology seriously, not as a college project, uh, but something that can move forward." Um, it's, it's... And, and you mentioned your family, which is giving you a lot of, uh, uh, which is backing you up for everything else in your life, right? It's amazing.
- SSSharan Srinivas
Well, you missed one thing.
- SPSpeaker
[chuckles]
- SSSharan Srinivas
Failure and the ability to handle failure. What do I mean by that? Let me split it into two parts. The ability to recognize failure... There are people who are extremely stubborn. Stubbornness is not a bad thing. Stubbornness will make you keep trying over, and over, and over, and over again, and the difference between success and failure is, is one more attempt.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Right? So that stubbornness in that perspective, in that context, is a huge asset, but your greatest asset is also your greatest weakness. You never know when to stop. Some things are not meant to be. You just have to accept that they don't work.
- SPSpeaker
I could call them gritty people also.
- SSSharan Srinivas
Yes, grinders.
- SPSpeaker
Yeah.
- SSSharan Srinivas
The gaming terminology is grinding.
- SPSpeaker
Yeah.
- SSSharan Srinivas
And grinding is good.
- SPSpeaker
Yeah. I mean, Rahul Dravid is a gritty guy.
- SSSharan Srinivas
Yeah, absolutely.
- SPSpeaker
Stubbornly at his crease. [chuckles]
- SSSharan Srinivas
Absolutely. I remember one cricket, one test match, where he took ninety-two balls to get off the mark-
- SPSpeaker
Yeah
- SSSharan Srinivas
... to score one run. He scored ninety-two, the entire stadium applauded, and he raised his bat as if he scored four hundred runs.
- SPSpeaker
The weather does stop.
- SSSharan Srinivas
The ecosystem is set up to f- for you to fail
- 1:04:00 – 1:08:00
Personal Story: The ETH Zurich Challenge
- SSSharan Srinivas
safely.
- SPSpeaker
Mm.
- SSSharan Srinivas
Are you willing to accept that failure?
- SPSpeaker
Right. Okay, understood.
- SSSharan Srinivas
I was one of those people. I couldn't... I, I have only had two failures, uh, in my student life, and that, too, uh, it'll sound really silly, but in my eighth standard, m- my maths annual exam paper, I think, I got fifty-nine out of hundred. I hadn't got anything under ninety-five in any course until that point in time.
- SPSpeaker
Right.
- SSSharan Srinivas
I thought I was the world's- uh, world-class failure.
- SPSpeaker
Class A, it'll feel quite-
- SSSharan Srinivas
Right? When I was doing my master's degree, I had this one course that was a bachelor's course.
- SPSpeaker
Where did you do your master's?
- SSSharan Srinivas
Uh, ETH Zurich in Switzerland, and their system is such that you are allowed to fail in a course only once. If you fail a course twice, you are debarred from taking that course. The logic there is, it is not for you. There's a-
- SPSpeaker
...
- SSSharan Srinivas
- whole bunch of other thing else, so you go take something else. But for me, unfortunately, it was a bachelor prerequisite course, and I had to pass it, right? If I don't pass that course, my master's program will be canceled, and I'll be sent out of Switzerland. I'll be de- sent back to India, and my father's health was declining at that point. Guess what happened? First time, fail. Again... And then I take- retake the exam.
- SPSpeaker
Mm.
- SSSharan Srinivas
And usually, what happens is, the professors will give you, like, eight to ten problems to solve. You can choose to solve any problem, and then they'll usually come and say, "Hey, if you solve three problems in its entirety, you will get the grade six. Perfectly, in entirety, you'll be able to get a grade six." And I was like, "Okay, six problems. I can do two, and I'll be able to pass." Professor comes and, like, "You need to-- This is a very easy paper." And I was like, "What?" Because it didn't look very easy to me. Computer science student sitting in the Department of Mathematics, writing a theoretical paper. Yeah, and I was a guy who thought, "No maths," right? I failed JEE because I, I, I, I failed, uh, ma- the math part of it. Not patient with it, don't... I, I prefer things that I can imagine, and so therefore, physics and chemistry is easy, maths is not. Professor comes and said, "Hmm, easy German, easy paper. You have to get four questions to pass."
- SPSpeaker
Mm.
- SSSharan Srinivas
And I was like, "Fish..." I somehow end up attempting four questions. Then I sit, and then I sit, and then I pray. Nine thirty pm, I get an email saying, "Your grades have been updated." You know what I got?
- SPSpeaker
Four. Yes. That was joyful. [chuckles]
- SSSharan Srinivas
Not more, not less. Four. It was twelve years ago. Absolutely irrelevant today. But I failed and nearly destroyed my career. My, my father was also, uh, uh, you know, one of those over-accomplishers, never failed, and his, his health was completely, uh, declining, to the point where we were wondering whether he will live for another year or two. That was the situation, and then I, and then I was failing, sitting alone in the middle of Zurich, not knowing what to do, not, not being able to share it with anybody, and then I passed.... Passed, four. 1:00 AM in the morning India time, I called my mother up and said, like, "I bloody passed."
- SPSpeaker
Nice.
- SSSharan Srinivas
Right? Ten years hence, you know what that has given me? You came through that.
- SPSpeaker
Yeah.
- SSSharan Srinivas
You came through that, so you can go through this also.
- SPSpeaker
Yeah.
- SSSharan Srinivas
A little bit of farm problems in, in corpo- in, in, in, uh, uh, uh, government, uh, bureaucracy, and statutory compliances, and all that is not going to shake up your world.
- SPSpeaker
Right.
- SSSharan Srinivas
It may cause delays. Yeah, let's roll with it. If it causes a delay, it causes a delay. Worst case scenario, you have to go and make an explanation
- 1:08:00 – 1:10:00
Finding Joy and Balance: Personal Interests
- SSSharan Srinivas
to your customer.
- SPSpeaker
Yeah, or you have to just try again.
- SSSharan Srinivas
You have to reschedule your, uh, timeline again. It's not... This too shall pass. That's the first thing that I learned. The second thing that, that I learned is, you can try as much as you want, but if you, if, if external factors require you to take a beating, then there, there's nothing you can do about it. You can try all you want, but if you have to take a beating, then you have to take a beating.
- SPSpeaker
Yeah.
- SSSharan Srinivas
That is, that is what I mean when I say you have to accept your failures.
- SPSpeaker
Okay, last question, I promise. What do you find joy in? How do you relax?
- SSSharan Srinivas
These days, I sleep.
- SPSpeaker
[laughing]
- SSSharan Srinivas
It's a big deal today.
- SPSpeaker
Yeah.
- SSSharan Srinivas
But otherwise, um, I am a bit of... I'm kind of, "I do whatever I want. I do all sorts of things," kind of a guy. I cook, uh, I stress clean. Uh, my mother is very happy when I do that. [chuckles] I listen to all sorts of mu- I listen to all sorts of music. Um, I watch movies. The, the more, the more nonsensical the movie, the better. Uh, I, I am not a big fan of movies that make sense, because I-
- SPSpeaker
Not Inception?
- SSSharan Srinivas
No, not, uh, not Inception. No Malayalam movies also.
- SPSpeaker
[laughing]
- SSSharan Srinivas
They're very good, right? So I generally... They're very serious, they have emotional-- You have to invest emotions into the... You become, you kind of become a part of the character, and-
- SPSpeaker
Yeah
- SSSharan Srinivas
... all that stuff. No. Tamil movies, Telugu movies are my thing. Go beat people up, make a lot of noise, blow up a whole [chuckles] bunch of things, have popcorn, and relax.
- SPSpeaker
Yeah.
- SSSharan Srinivas
Uh-
- SPSpeaker
It's interesting.
- SSSharan Srinivas
Yes.
- SPSpeaker
And I guess you watch a lot of cricket also, because you mentioned-
- SSSharan Srinivas
IPL, these days, yes. Uh, I used to. I used to be a very avid follower of cricket, uh, to the, to the tune of watching test matches five days, ball by ball, but test matches are not what they used to be. Uh, and IPL, I,
- 1:10:00 – 1:11:30
Closing Thoughts
- SSSharan Srinivas
I, I also watch... I, I watch reviews of IPL or recordings of IPL matches that went to the wire.
- SPSpeaker
Right.
- SSSharan Srinivas
Uh, sometimes it is so dominated by, by one side that somebody comes and scores 200 runs in an IP- in a 20-over match. It's-
- SPSpeaker
It-
- SSSharan Srinivas
... Come on, man, there's no contest there. Or some, some bowler comes and takes six wickets in the, in, in like two overs, and-
- SPSpeaker
Yeah
- SSSharan Srinivas
... there's no point in, it's boring. You need that balanced competition, where people employ techniques, strategy, tactics in a, in a balanced team. That's when I start watching matches these days.
- SPSpeaker
So cricket.
- SSSharan Srinivas
Music.
- SPSpeaker
Mindless movies.
- SSSharan Srinivas
Mindless movies. Yes. Exercise is not counted as a leisure activity. It's a part and parcel of life.
- SPSpeaker
Yeah. Nice. That's damn cool. Um, always exciting to talk to you, Sharan. Um-
- SSSharan Srinivas
Likewise.
- SPSpeaker
Very insightful.
- SSSharan Srinivas
Hopefully, you can make me look good on the camera.
- SPSpeaker
Yeah, 100%. Thank you for watching. We were with Sharan at the Center for Innovation. This is the Best Place To Build podcast. I hope you enjoyed this episode. Do like, share, and subscribe. Thank you, Sharan.
- SSSharan Srinivas
The camera team wanted me to say the same thing, but it's not my channel. [chuckles] I'm not gonna do that. But if you do like it, go ahead. These are good people.
- SPSpeaker
Thank you. [outtro jingle]
Episode duration: 1:11:31
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Transcript of episode L59QHD1rG50
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome