Indias 10 Billion Semiconductor Plan Neelkanth Mishra Explains Sparx

TITLE: India’s $10 Billion Semiconductor Plan | Neelkanth Mishra Explains | SparX CHANNEL: SparX by Mukesh Bansal DATE: 2026-01-31 ---TRANSCRIPT--- We are kind of where the Chinese were 25 years back. So one American company went bankrupt, one Japanese companies

because of rise of this Chinese companies. China is massively massively investing in. So our question was how are you going to compete? We’re trying to discourage people from starting a chip design company. No no no. So I recently met a team of people who think that what can globally be done with $250 million will be done at 100. This is the ecosystem disruption that India can take advantage of. So the mission is to create a semiconductor ecosystem in India. We’ve come a long way but at the same time we need to stay at it for 20 years. We need the government funding to keep flowing. A number of large Indian groups when they were approached by the government they had zero trust in the government. So we need to build our credibility. I remember one very embarrassing interaction. I will not mention the company’s name. So they had a partner. That guy gave us a lecture for 15 minutes effectively saying beggars can’t be choosers. The foreign company was taking 5% stake in this and that company was not putting its own money. So it was just saying that I will share technology and in return I’ll get 5% stake. My thing was you can walk away in time. Where is your skill in the game? Welcome. Welcome back to Sparks. Yeah, I think you continue to be very much in demand. I think we feel like we need to come back to you every 6 months perhaps sooner. Uh the pace of progress in the world you know seems to be accelerating. I’m just amazed like I’m really struggling to keep up. So talking to you is a good way to really know what’s happening in the world, what’s happening in India, how should we navigating and I think you are in the middle of a lot of things. Today I want to focus on I mean you have a specific role with the semiconductor mission which I think finally making a lot of strides in last one or two years from everything I read in the news. But it also seems like overall tech is becoming very very important for as far as policym governance is concerned. So we’ll talk about you know probably both aspects today. Starting with semiconductor. So let’s first of all maybe let’s start with your interest in semiconductor. I think I think one of the podcasts I remember talking to you in your early days of your career you spent time working as a semiconductor analyst. Yes. Uh so that was 2003 onwards and uh so I was hired to cover IT services but my boss uh was struggling with not struggling but he wanted help in uh coverage of semiconductor foundaries in Taiwan. So uh so within a few months I was transitioned. So I covered uh Taiwanese foundaries um memory I as a as a strategist a tech strategist I went to Taiwan and I covered chip design companies myself. uh I was the I was the part of the team that took SMIC public uh the biggest semiconductor company of China and uh so yeah so it was and it is such a fascinating field that even though I had not I mean actively covered it for 17 18 years after that uh I was still keeping track of it uh as to how the node transitions are happening what the big companies are doing it’s been a fascinating fascinating problem and look our world is today run on tech and pretty much entire tech is run on semiconductors. I’m pretty sure this mic that we are using most but uh yeah the way you’re laughing it seems like you know more than what you want to I don’t think but point is you know there is not a single electronics gadget or device and semiconductor mission is trying to change that. Yes, again it has had I think uh it has had little checkered past. I think we’ve been trying to do it for a while but looks like just what is the thought process behind semiconductor mission so we can feel optimistic that is that is very much possible. I I must say that uh you know when our uh Mohali facility in India was was set up um uh it was uh I would say even ahead of uh of course ahead of China at that time but uh in the early 80s it was even ahead of Taiwan m uh and then we let the ball drop we we didn’t India was ahead of Taiwan in ’ 80s 80s because we’ve been we’ve had that the laboratory in Mali for a even tuting mini computers like so much advantage in 70 HCL computers you know so on hardware PC manufacturing as well I think uh we had a fairly decent manufacturing setup two generations lost out but anyway so periodically I think people did wake up and say oh we need to two semis. So I think 2006 7 there was an attempt uh 2016 1516 also I think there was an attempt but both these times and and and that’s why the credibility also for India had gone down India is not able to execute on the plans it makes and so when this third attempt was made so December 21 India semiconductor mission was formed this time it was back right from the top see the prime minister himself is personally involved now so he has approved a $10 billion budget and to tell the country and the bureaucracy how deep his conviction is. He has inaugurated every single Semicon India conference. He also meets the heads of global semiconductor firms regularly around Semicon India. There is a group CEO meeting. One mentioned that in September 2025 the PM was making notes as he spoke and then promised that all the issues that were mentioned will be addressed well before the Met next. Semiconductors are now part of the diplomatic agenda on summit meetings say with the leaders of the US, Japan and Korea which have the technology that we need. You would have seen pictures of the PM and Intel CEO Lihutan recently after the Intel and Tata electronics signed an MOU. He has also put the right people on the job. So the minister Rashini Vashnau not only understands technology and he’s also from IP Kpur but understands business has been an administrator and also a policy maker as you know he turned around leway capex you can see he’s imprinted not only on the structure of the policies like the significant capital subsidies but also the decision-m process foreign firms as well as the Indian applicants now get the comfort because he meets them personally that there is someone who understand what their needs are. Agencies first help ISM and the ministry vet the detailed proposals. Then external technology and financial experts like me evaluate the applicants capabilities in detail. There are multiple rounds of deliberations and applicants are asked to show up with parts of their plans like you know bolstering the team say you need more equity capital get a better partner and things like that then this goes for cabinet approval. If an approved project is getting stuck because the local bureaucracy doesn’t understand the importance or are you know behaving generally how they do with projects he then goes in and understands how to address the bottlenecks. In fact in the early years there was another minister uh Rajiv Chanhiker who has himself worked with Intel and can talk about assembly language coding for example who also helped shape the policies and processes and more importantly build a conviction among global technology firms that India is serious and understands what it wants to do. The formation of the mission itself was a very important step forward. So the folks at the ministry of electronics I think they realize that uh you know in the bureaucrats are uh you know transferred very frequently. So they are excellent at what uh you and I will relate to in operating system context switching. So they they transition from one role to another very fast. But the the systemic memory which tends to be kept in files um on sophisticated things that is not enough. So you need a permanent body which will be the repository of information. So December 21 ISM was formed uh uh around I think March, April if I remember right. They constituted an advisory board. When you see the advisory board, you realize that there’s so many legends in the world who of Indian origin. Yeah. Who one are not only you know at the cutting edge of technology but also are very willing to give back. So so uh global leaders across firms. Um and so the that that body was formed. The government allocated at that time $10 billion. So 76,000 crores. Initially there was a cape subsidy planned that for advanced uh semiconductors it would be 40%, for slightly less advanced be 30%. Uh but in the first meeting itself I think we all of us agreed that look right now we have nothing and making it too complex on you know being this being that so let’s keep it very simple. Let’s increase it 50% cape subsidy. uh on top of this the state governments have topped up. So some states offer 20%, some states offer 30%. And therefore by putting in 20 30% of the cipex you get to own a whole fab. So given the large resistance a huge resistance to uh to setting up a facility remember that these are very capital intensive large complex give a sense of capital like to good fab is what few billion dollars more depends on which node you’re talking about. So just like you know in a steel plant uh there was a time when you could have you know 200 500 cubic meter furnaces but today if you have to be viable you have to have a 4,000 5,000 cubic m furnace. Now a 5,000 cubic m furnace if it runs through the year will produce 4 4 and a half million tons of steel. In the same way for a wafer fab a 12-in wafer fab so as you know silicon wafers are of 8 in or 12 in. So 12 in uh started about 20 25 years back for it to be viable you need to have at least a 40 50,000 wafers per month fab if you are trying to do uh 28 nanometers or 40 nanometers you know so that that range it can cost anywhere between 5 to 7 billion if you are trying to do a 2 nanome fab or a 3 nanome fab uh it could be $20 billion and just to put in perspective today state-of-the-art in the world is 2 or 3 nanome and right now India but we are attempting in the 28 nanometers. Yeah. So 20 well we are first attempting 40 nanometers. Okay. Which is 2028. Okay. Uh the plan is to get 28 nanometers by

Understood. Understood. And these 40 nanometer will will cost $45 billion and 70 80% of that is because of semiconductor mission subsidies by the state. So in some ways the cost of entry for people who produce capital has been reduced dramatically. So what tends to happen? See this is and again I must be honest that initially uh because I was still thinking like an analyst uh I was not wholly supportive of this whole exercise uh you know in in semiconductors the value chain is very very long it’s a very large collaborative ecosystem so you one you can’t do anything on your by your own on your own so the complete atmir is impossible and we can discuss that as we go along so um so there are chip design companies these are called the fabless companies they don’t have a fab semicond to fab. Uh then there are the wafer fabs. Then there are the packaging and testing companies. What you do is you get you wafer fab at nanometer level. But when you’re putting them on motherboards obviously those are very fragile require very high levels of clean air and all that. And of course then also they dissipate heat. So you need packaging uh which brings it to micron level and so that you can put it on like a packaging is putting components together in a functioning chip. Yeah. So you can do that. So there are there are packages which incorporate multiple chips but mostly it is encapsulating that piece of silicon okay into a ceramic or a plastic or a metal packaging and it’s almost like you know those those millipede type centipede type things that come out. So those are the wires and they are micron level mm level. So that’s packaging wafer fabrication and packaging are together about depends on chip to chip but could be 30 to 40% uh of the while while you know in the Nvidia blackwell chips Nvidia makes 80% margin so the packaging and testing is only about 20%. M but and and conversely if it’s a commoditized uh chip then the fabulous company makes very little money mostly it is manufacturing even in the manufacturing the value ad is split between u the equipment companies because the sophistication and the complexity is often in the equipment company itself. So for example this 2 nanometer very famously the Dutch company SML they are the only ones who can make the lithography I mean using that kind of lens. But if you just think through it, so if you’re if you’re generating ultraviolet light at um 2 nanometers, it tends to get absorbed everywhere and therefore the mirrors you use to focus it are also manufactured only by one or two companies anywhere. The lens that can focus that is grown as a single crystal over months and and it also can be done only by Zeiss or some one of these companies. A lot of the value ad is actually in the equipment or or even like testing there are just two companies terodine and advant which control 90% of the testing equipment market. So as you can imagine the manufacturer doesn’t have too much bargaining power and it is one of the characteristics of a system where you add a lot of capacity in bulk but your demand goes up steadily is that there are vicious cycles. So there’ll be bull markets, bare markets. So inventory up cycles down cycles and through a cycle very few companies cover their cost of capital. You know when we uh look from semiconductor mission is the objective is to have all these companies or mostly fabs the fabulous companies, fabs, packaging companies, testing companies. Awesome. So uh so yeah so the mission is to create a semiconductor ecosystem. Okay. So all of these type of companies. So the 76,000 cr project the the budget that I’m talking about is for wafer fabs and display fabs and packaging plants. there is a separate design linked incentive scheme to have uh chip design. We’ll come to that but uh to finish that argument because this is a viciously cyclical industry and returns for manufacturers are actually very low. So in the last several decades there are only three companies that have actually covered their cost of capital meaning that if it is say 10 12% globally globally so Samsung and TSMC they used to be Intel but now Intel is also struggling. So my as an analyst my first response when not that they were asking me but when I heard about this was that why are we wasting money right this is this is never financially viable and then I was told that look you know you’re thinking like a financial analyst think about it from the country’s perspective do we need a semiconductor ecosystem yes then how do we build one so we have to get people through the door that convinced me and I I think that’s the strategy from there uh to actually get large companies to start producing producing in India. So, so one of the things I realized was that uh there are so many people of Indian origin who work in some of the most advanced plants like you know so whether it is applied materials or nollas or even ASML or TSMC or you know you name it right everywhere Micron full of Indians uh or Indian origin and many of them wanting to come back. problem was that there was nothing to bring together the capital formation that was necessary for that one facility where people would start coming back. Now the challenge is that you needed this capital which was of course a necessary condition but you also needed someone who had the technology because uh it is kind of like an apprenticeship uh problem that an Indian company cannot just start wake up and start doing 40 nanometers looks oh it’s so 20 years behind the curve pos it’s 40 nanometers I mean you know so uh we’re talking about really really precision manufacturing um when I started covering semiconductors the transition was from 130 nometers to 19 nome and even that was cutting edge. There were so many problems with that. So you need a foreign partner who has the technology, who has the experience of how to manufacture chips. But at the same time, most foreign companies will struggle to work in an Indian environment. I mean for obvious reasons, right? So while the central government may be convinced uh and the state government is obviously convinced and they’re giving the capeex, but the local government may not even know what this is about. Like for example, what is happening now is that some of the plants that are coming up, if they’re importing equipment, the customs guys don’t know how to classify the equipment because they’ve never seen that equipment. So there is a learning curve across and so it is easier for a domestic firm to handle the local bureaucracy about you know managing the land, the permissions and this and that. So you have to do a lot of matchmaking. Now in the first couple of months I must tell you it was absolutely depressing because some of the projects that were coming in and remember this is like what what you all do in in say angel investing that a bunch of entrepreneurs come with a presentation and you say oh you know what okay fine you know take this money and build a business. This was like that because there’s nothing on the ground beyond a presentation and some of these people didn’t even have a team. I remember one very embarrassing interaction. I will not mention the company’s name the foreign company’s name. So they had a partner I mean a foreign partner and I was uncomfortable foreign partner and I was uncomfortable because that company was taking taking five the foreign company was taking 5% stake in this the source of capital for this company wasn’t very clear to me and that company was not putting its own money so it was just saying that I will share technology and in return I’ll get 5% stake so my thing was you can walk away anytime where is your where is your skin in the game and then that guy gave us a lecture for 15 minutes that you know effectively saying beggars can’t be choosers So it was really embarrassing that you’re throwing so much money on the table yet you get uh you’re not getting good good people through the door. So things really started turning on two fronts. Uh the first was that Micron came in. So I remember I was taking my kid to the dentist uh in the evening and and when we heard it was Micron I said look you know we we can’t wait for the right time before they change their mind. Let’s get them through the door. Micron is a American company. Micron is an American company. It’s a it’s a memory company. So memory is again so you will see that in all of semiconductors uh there are 50 to 70 blocks of expertise. So for example the software in which you design chips is called EDA electronic design automation and there are only three companies effectively that run the business. When you are talking about say lithography tools there used to be three now effectively there are two. uh similarly in in memory there are only three companies effectively I mean there are more but large amounts of market share uh are with just uh Samsung skhics and uh and micron so Micron uh decided to was exploring how to do a packaging plant okay so it is it is less capital intensive these are smaller plants so about two to three billion because they’re setting up packaging plant for memory chips in case of micro memory chips so another thing which is very interesting is that memory chips As if you remember, you just have to store. So it’s like, you know, creating a grid and you know, so it it’s it’s a reasonably simple design and therefore they’re always ahead in terms of node transition. Okay. So memory 2 nanometer would have happened way before logic 2 n because logic chips are obviously complicated. So then fabbing it is is is is complicated. Designing it is complicated. Uh so memory is always a bit ahead. uh and of course there are new types of memory requirements like now this in AI servers now you need high bandwidth memory so there different types of packages so today the state-of-the-art for memory is also two two nanometers uh I think they’ve gone one nanome yeah so they’ve gone below that as well but that’s that’s on the wafer fab what what micron is doing in India is packaging packaging sure so when that packaging plant came um we uh was so excited that this was uh because see the the the problem for approving some of these plants is that it is very exciting and very important for the country to do it. The the the challenge is that you are approving grants of several hundred million in some cases a billion dollars based on nothing but a PowerPoint presentation. Yeah. You have to make sure that and it’s it’s going to be a 3 to 5 year setup journey. The team has to have the ability to deliver on it. You also have to be sure there’s no gold plating and remember that never no one has ever done any semiconductors in India. Yeah. So you don’t even have any benchmarks. So like clean room, I’ll tell you a very very interesting, it’s a fascinating story of how all the risks we know, all the opportunities we’re not aware of. So for example, clean rooms. Yeah. So when you’re manufacturing semiconductors, uh obviously you’re talking about nanometers. So even a dust particle falling will spoil the chip. Yeah. Just so that I mean just I I every time I think about this, I get so excited. So like you know the the ingots the silicon ingots that are used for semiconductors have 10 9 purity. So 99.99999 whatever 10. So 10 9 purity uh because even one atom because we’re talking about you know stitching together atoms and so the clean room is like you know class 10 like you know in one cubic meter you need 10 particles or class one or whatever. I don’t know if class one exists or not but so obviously and but if you’re doing packaging you can run with class thousand or class 10,000 now it is very expensive because you have to create large rooms where in some of the most advanced clean rooms in TSMC’s fabs the air circulates 90 times a minute so like ridiculous levels of purity right now of course it’s very expensive so when you are when you are seeing a project plan that this is how much the clean room will cost both sides are guessing because the the company has guessed and you are also guessing whether that is correct or not and the fact that no one in India has done clean room assembly and so they will have to get foreign suppliers and so people say okay it’ll cost more sounds like a reasonable but how much more we don’t know so if there is evidence of gold plating and it gets discovered 3 years later someone’s neck will be caught so how do you do benchmarking so we as we started we started developing these frameworks that look uh we want uh you know capeex benchmarks how is your equipment and if your if your capeex is too high you will not be competitive it’s a capital intensive industry then we also wanted uh uh manufacturing cost benchmarks because remember that today China is massively massively investing in uh semiconductors already in a certain type of semiconductors which are called compound semiconductors so just like you know so semiconductors like you know you have a silicon wafer and you dope it with nitrogen and phosphorus or whatever and then Or if you want high voltage and certain types of usage so you can have a gallium arsenide or you know silicon carbide so these are in three years Chinese fabs as they’ve ramped up have brought down wafer fab cost by 50%. So one American company went bankrupt one Japanese companies uh because of rise of these Chinese companies. Yes. So our question was how are you going to compete given that it’s in both our interest that you survive. We want to ask the tough questions up front. So but with Micron look there was low risk that the company would run out of money solid American listed company. Uh second they would have a market. Uh so we checked with them that look you may have dozens of plants globally but uh this is going to be our only plant at least for that time. Yeah. So how will it be utilized? So they said look 70% we do in house 30% we outsource. If the DM demand or memory demand starts to fall then we cut the outsource guys first. So this fab will be fully utilized. So we are we assured that it works and they won’t be goldplating with the US listed company anti-corruption measures and all that. So it was it was and and plus the credibility boost that comes from that micron is is doing it in India. So so it was a huge jump for us and then so this project was kind of fasttracked. Yeah. But not fast enough uh for our patient. So, so we I think approved it around October 22 and the so I’m part of so ISM uh the advisory board uh they created uh two groups called the techno financial advisory groups there was a three member group for wafer and display fabs a two member group for packaging so I was part of both because as a financial expert u and there were technology experts on both so in TF for wafer fab one for packaging so we approved and then of course there was a process of cabinet approval and and so on so forth. So that was happening. Now in between because see remember Micron is going to bring its own technology. So they needed clarity on how much tax they’ll have to pay. It’s a classic advanced pricing agreement kind of structure. But there was no framework to uh to sign an APA for a semiconductor company. So it took 9 months. Now all of us were getting restless because see memory is a very cyclical industry even more cyclical than than say foundary business and semiconductor and it was worrying for us that it was taking so long but apparently it was the fastest uh an APA was ever done. But so where are we with micron today? No micron was cleared uh in May 2013. Plant is being manu is set up plant has been set up. I think their uh their test batches are out. I think mass production should start. So this is the very first production outcome from No, the first first no first production is actually uh CG power. So that has already happened. Okay. So the the the first uh chip coming out uh was CG power and Renaissance. Okay. Which which kind of chip is that? It’s a it’s a logic chip. Same 40 nanometer. No no that is packaging. Packaging. Okay. So wafer fab we only have one plant. Okay. This is the one in Assam which starts that is the packaging plant. That is also okay. So there were there were six. So uh there were six large projects that were initially approved. Now there are some smaller ones that have also gone through and now the first first budget that 76,000 crores is exhausted. Right. So now we are waiting for the second. So this just to for me to understand this 76,000 crores is distri distributed across what about it 10 odd projects. Yeah 10 o 12 13 project and majority of them are packaging. Yeah. one wafer fab and the rest and and a couple of sim compound semiconductor projects and uh so I’ll tell you so the the big ones are uh Tata wafer fab which is bulk of that 76,000 crores uh there is Tata packaging um there is the CG power and SAS and Tata’s uh technology partner in wafer fabs is PSMC power chip semiconductor it’s the third most important fab in Taiwan understood um and then uh we have Kane’s technology mh M and and Micron right and then there was HCL tech fox con is there like semic semiconductor mission 2 yes so so that’s the expectation uh at the semicon India this year so so the the the semiconductor industry is so collaborative uh and it needs to be so collaborative because as I said you know all these blocks need to work together uh so semicon is the conference where all these companies come together so semicon India uh is a new fixture uh so earlier there was semicon Yeah. California, Semicon Taiwan, Semicon China, Semicon Korea, and now there’s a Semicon India. So at the Semicon India this year, which the honorable prime minister inaugurated, he effectively said that look, the second package is almost done. It’s going through the files. Okay. So it’s safe to assume is going to be hopefully dramatically more than the outlay. Dramatically, I don’t know. I mean I’m not privy to what the size of the of the thing is but uh yes I do think that the first one uh we did not not blindly but you know we we had no idea um of how uh good or bad things would be. Now we have a much better sense. So I can tell you what I would like it to be. So uh given that uh whatever projects are approved under this would be coming online 203132 at at the earliest uh the from the semiccon 2 semicon 2 because it’ll take a while for projects to come through there’s there are people waiting for it to come through so I think this time the flow will be much better uh but we would like to be in the 7 to 12 nanometer range because by 2032 I think that will be the mass production okay uh nodes but but we’ll see how it happens coming just closing that argument on that story on uh clean rooms. Whereas we initially assumed that clean rooms in India will be more expensive than those elsewhere. What was fascinating was that there’s this company called Yes Technologies which provides packaging equipment for Micron. So they have this plant in Kimaturur and they were so excited that look uh when we set up our clean room in Kimaturur of course it’s class 10,000 so it’s not that cutting edge but it’s still still pretty pretty complicated. He said it’s a fifth of the cost of our clean room in Arizona and as you know Arizona is a new chip manufacturing hub that is that is getting constructed and they said that look when we told Micron that look uh you know if our this guy and all local suppliers so as we start getting into this you will find lots of positives as well. So there are lots of unknown problems as I said you know the the customs guys don’t know how to classify the equipment and all of that or that uh you know the the soil in Dera is too sandy and therefore establishing that platform turned out to be more complicated. So there are things that you will discover but is semicondues to be like big industry play or the scope for new age players as well. The crop of companies that are large has been changing. I think 80s and 90s the Japanese were very important. I I think then they slowly started fading out. The Taiwanese became much more important. Then now in terms of wafers produced per month, China’s share is going up dramatically. So, so there are always new companies that come in. There are also new challenges that come at every node. So for example, Intel struggled because they couldn’t get extreme ultraviolet. You know the there will be new challenges. So as we uh so some one of the equipment companies was showing me uh and I got it validated I mean I read it somewhere else as well that till 15 years back uh you know men’s periodic table. So there were 15 of those elements that were being used in a in a cutting edge chip or 15 20 years back today there are more than 50 elements 50 out of roughly 100 known elements. 50 are used in a chip. Yeah. and and many of those if you remember are artificial like they just created it in for like few seconds in some some uh uh labs. So uh so it is remarkable how com so so what has happened now is that 1 nanometer is is like 10 anstrom the moment you go to five angstrom six angstrom three angstrom you’re talking about like moving atoms around you can’t make transistors smaller than that and one angstrom is roughly the diameter of an atom exactly so so you are uh so you can’t go smaller than that so now new types like you know 3D semiconductor 3D uh transistors uh different types of uh integration so different types of packaging uh how to get the same MS law type of productivity gains and reducing power consumption. So all of those things will create new technological Mhm. uh challenges whoever solves them first will of course be able to progress. Got it. So the story that I’m that’s uh at least I’m able to understand is you know because of uh you know these 9 10 projects which got off the ground because of semiconductor one uh mission one and these are all getting into production soon uh semicon 2 is uh going to come hopefully soon and uh we can be little more ambitious with that. So in some ways the foundation stone for the semiconductor ecosystem is being laid in this decade. So our our objective as I’ve uh heard um from the policy makers is to get 12 to 15 large-sized. So of the 123 we’ve done I think four five are very small. Uh I mean they’re just there because they’ve been doing semigs. See there are companies in India which have been operating since the 60s and 70s they’ve just not grown in size. So there are some small ones as well. So we want to get 12 to 15 large projects on the ground and then the government can pull its hand away and then let the ecosystem flourish. So for example, you know, if you’re setting up a wafer fab and something breaks down, remember that some of these fab some of these chips have 200 300 layers that of sort of stacking material. Uh they take 4 months to process, 5 months to process. Now imagine if something breaks down in between. The risk is that a lot of the lots will get have to be discarded if you have to wait for too long. Now how quickly can the engineer come? Now the engineer may not be close by. I mean they have to take a visa and then you know come down and it doesn’t make sense for an equipment company to have an engineer located if there are no fabs. So so these are the starting points and which is why I think the first fabs required a lot of capital subsidy uh because initially projects will get delayed and that means a cash burn. So in a way if you think about the the way discounted cash flows work because these are very capital intensive industries. So even if your return on capital employed is negative meaning that you’re burning cash in the first couple of years given the amount of cash that has been handed over to you and what we made sure is that in the first batch the large projects they are all backed by solid balance sheet so uh so I must have been hated by everyone for insisting there’s not going to be any debt that you know because companies fail when they take on debt and then they’re not able to repay the interest that’s when the bankruptcy and the banks start come calling uh If you put your own cash then you know you will you will destroy wealth but you will not go bankrupt uh for a long time. So so we made sure that these firms can survive for 8 10 years and in those 8 10 years these are all solid entrepreneurs who are investing for the long term they’ll figure they were out. So so we are kind of backing the teams to to do it on their own. So if you have 12 to 15 large plants operating in India then there will be technicians, engineers, innovation ecosystems and what will also start happening is that the likelihood of large chip design firms coming up. See India has uh uh 1/5if of the world’s chip design engineers. Uh so out of one and a half million chip design engineers in the world 300,000 are in India. But how many chip design firms in India? Yeah, I’m I’m sure there are a handful now but uh but no one of size. zero effectively. So if factor is this 20% of uh chip design chip designers in India they’re all basically working for multinationals and so so Intel for example apparently has 30,000 people working in India for them and today what is the barrier to entry for a you know let’s a really ambitious entrepreneur to start a chip chip design company very very excellent excellent question and this is where uh when I was talking about disruption or what do you use to enter and how the industry changes so think about chip designing as like how say I’m sure Arnova played with the circuit design things like you know you’re placing things and connecting them and all that some of these chips today have billions of transistors so imagine I mean of course there are libraries that help you do that it’s not like everyone writes C code right so there are libraries but you have to place those things and uh to do a reasonably sophisticated job uh you you know the the cost of a chip which used to be because it takes a long time to to to to design Mhm. And then uh tape out meaning that you know converting it into something that earlier uh you know like 1960s people used to draw circuits shrink them through photographic things and then print them like you print and all that but today of course you can’t do all of that. So it has to be chip design has to be translated into something the fab can understand the fab machines can understand. So these are all very expensive. The EDA tools are very expensive. The the license to those the amount of processing you need to just make sure that imagine like rendering a video you know you can say that what Mukri and I used to do the man picks the flower and gives it to the woman to write in English. Mukri and you used to do yeah that’s a bet project. Pick the flower. Okay. All right. So, we we were tech project. No, good for you if that you got away with that. No. No. So, I I did the English parsing and Mukre did the the graphics behind that. So, so rendering it used to take a long time because you know the computer needs the bits uh to be there. Yeah. The compute needed is a lot. Storage needed. It’s a very expensive proposition. So, what used to cost cost 1020 million. Yeah. now cost several hundred million. Are you trying to discourage people from starting a chip design company? No, no, but hold on. So this is where the fun one fun thing happens. So I recently met a team of people who think that what can globally be done with $250 million will be done at $100 million in India because where is the arbitrage? Yeah. So uh one you get very cheap engineers in India. Two, what is also happening and this is where the geopolitical shift is so fascinating. See many of the American companies who have been barred from entering China have lost a lot of money. Yeah. Or lost a lot of potential revenue. Sure. So it has also hurt their ability to do research. Yeah. Because see all of these are very technology heavy very very IP intensive companies. So the uh R&D to sales ratios are very high. Suppose you lose $10 billion of revenue you lost billion half dollars of R&D revenue. What they’re even more worried about is that in their absence the Chinese EDA companies and the Chinese equipment companies will start sprouting up and then five years later start hitting them. So they have to create an ecosystem which they are funding because their customer survival is very important for them. So what they are saying is that look guys if you’re in India till the time that you’re generating revenue I’ll give you massive discount on your licensing. Sure. Right. So so these are things that are starting to happen. This this is the ecosystem disruption that India can take advantage of. Same with equipment companies. Equipment companies had a huge business in China. Now because of geopolitical reasons they’ve been barred and they realized that Chinese equipment companies in 5 years time just like you know they are beating the stuffing out of everyone in ship building in cranes and you know heavy equipment machinery uh electric vehicles and all of that they will start doing the same in semiconductor capital. But look 100 million is also not easy for a startup. So what is that just if you have to paint a picture of you know let’s say you know some of these 100 million 100 million is is a critical chip which will be globally competitive. Understood. So let’s say I am a you know very competitive chip designer in one of these multinationals in India. Me and a group of my friends we want to start chip design company. Where do we start and what is the role of semiconductor mission in facilitating that? So remember that your eventual cost may be $50 to $100 million. If you want to do a reasonably size or address a reasonable size TAM see if you want to do a microcontroller which goes into say refrigerators and washing machines and there’s a captive Indian market uh you can actually sit with the equipment company or the OEM and you know do a custom design these are called AS6 you know application specific integrated circuit so you can do that you can I mean I think a couple of million dollars you can design a chip if you want to target a larger applicable like you know what you keep talking about in your book that you know your time has to be big. So uh so then you’re talking about something which which is designed at a at a at a 3 nanome 5 nanome level at least and uh and then targets like say mobile phone system on chip or some of some some of the big ones and uh those you need $50 to $200 million but you don’t need all that up front. So what you can do is and this is where the DI scheme the design linked incentive scheme the government today I think that amount is not large enough but our hope is that in the second scheme that quantum newspapers are saying it’ll be 10x what it was there in the first so 1,000 crores will become 10,000 crores there also like uh ministry of telecom also has a 250 cr competition to design chips that go into telecom equipment so there are many such things that are starting off so look design competency and incent is one part what what about the you know going back to TAM you know the generating demand is that also for a upstart chip design company to just generate enough design enough demand for what they’re going to build is I’m assuming it’s going to be sizable challenge uh it is so uh and and this is part of the reason why we didn’t have startups right so what do you design chips for so you need to have some visibility of what’s coming 3 4 years later because it’ll take you one two years to design do the test chips test batches then try to integrate it with final device which is made by someone else uh and then rectify and then do mass production. So you need to know what this phone company is going to launch in 3 years or four years. How do you do that sitting in India? So uh so that was uh that was I think the the key bottleneck. So this is where I was a big supporter of the the PLI schemes for mobile assembly that once you have the assembly and basically some visibility in the ecosystem of what devices are coming then some entrepreneurs will then start working you get some visibility you’ll start designing stuff and are you seeing like are people starting this that’s happening yeah oh yeah very exciting see I’m part of the TFAG and so of course when the proposals have gone through a certain level of checking then they come to us and clearance but um but I when I go to semicon India I get to see all these entrepreneurs there’s some very fascinating uh ventures that are coming up but remember that very few of them will succeed so I remember showing this chart uh back in 2022 that uh you know for one DM controller chip which was worth 30 $40 billion coming out of China there were 110 of them which had started so so you need to have that that large cohort uh so I hear of uh some very talented uh engineers many of them and see and this is again where you you sense that things are starting to fall in place of course fingers crossed there’s 20 years of hard work right remember that we we are we are kind of where the Chinese were 25 years back so uh so so there’s a long distance for us to cover uh and we have to be patient but at the same time you realize that many of the engineers who left India 20 25 years back uh are now in their late 40s uh early 50s and their parents are now mid70s late 70s and India is very different from when they left it. So quality of housing, quality of cars, quality of schooling and pay packages and of course so they want to come back to India as well. So they just look looking for opportunities. So in fact one of the uh people who I think will be potentially applying in the second round when I he said oh you know we can set up a much larger project and I said but where will you find the engineers? They said oh there’s like a flood of engineers just waiting for for projects to come up in uh so yeah so I think the the the talent uh in the Indian diaspora even in India in fact when I asked this this equipment company uh the the packaging equipment memory packaging equipment company did you have trouble finding skilled people all all there and in this you know as the semiconductor ecosystem develops in India where does a semiconductor missions how is it thinking about the equivalence of GPUs etc because that’s Everyone has been watching and sick of where Nvidia stock is. The alternative perhaps Google’s TPU is emerging as a one competitor but again probably only two companies in the world and uh I think last 24 months it’s become clear that AI is going to be extremely integral to pretty anything that happens in our world and all the AI is going to be running on GPUs. So to what extent or is that is going to be third fourth phase of evolution to see I think see there are uh so we can we should aspire to get there uh I’m aware of uh design companies in India which are in 28 29 going to tape out chips which will take on Nvidia uh so obviously I mean you know so you need entrepreneurs VCs who are willing to back them not sure they’ll succeed I hope they do also many such attempts in India I mean okay not as I mean like not even 5% of the ones of what is happening in China or the US but clearly way way ahead of what we would have dreamt of 3 years back. So this is this is where I um I’d like to summarize where we are in in this right. So we I think in the last three years we’ve covered a long distance. I think what I see is uh the the quality of proposals coming to us uh just in terms of the seniority the seriousness the large groups large balance sheets that are backing them the the technology partners that they’ve managed to get the ambition of where they want to take the business. All of them are several notches higher than where we were in 2022. The maturity of India’s semiconductor mission in terms of you know all the questions that we used to ask. This is where learning curves exist, right? Humans are very fast and they’re very smart people. So all the questions that we would ask in TFAG meetings in the in the first time the reports came to us uh are already asked before they come to us. So now the the decision-m is much more strategic. Um we’ve also I think managed to get some excellent experts. So as as you can imagine as the government’s own network expands figure out people who are willing to contribute their time. Remember it’s a it’s a thankless job meaning that you are sticking or you’re you’re signing off on things where if things go wrong your reputation is at risk. Uh and of course there is no reward other than the the happiness that the industry is getting developed. But there are people who are willing to do that. Some super experts that we have on the technology side now. So we’ve come a long way but at the same time we need to stay at it for 20. We need the government funding to keep flowing. So we need to build our credibility. A number of large Indian groups initially when you know they were approached by the government that why don’t you they had zero trust in the government. Some of them said that look you’re talking about uh hundreds of millions of dollars or maybe thousands of crores. Do you know that to get 50 crores of payment out of the government how many uh pillar to post runs I have to do and so how can we be sure? So making sure that we build credibility that uh subsidy checks which are paribasu basically the entrepreneur puts in x amount state and central government put in y amount all that happens on time credibility builds people start saying oh you know what uh the rest of the government may be what it is ISM works very differently India is very serious every time semicon India happens the leaders of global semiconductor companies who come to India actually are hosted by the honorable prime minister himself he takes notes comments on all the things that are mentioned as issues. So as of now the desire and the momentum I think is super strong and given where also niland we are on the uh geopolitics of the globe seems like this is there is no choice in some way has to work I think there is no version of economy where you don’t have at least some sense of self-reliance and also you know competence in house to be able to exist as a free independent nation in the world right so it’s some ways whole country needs to root for semiconductor mission yes and its objectives come true irrespective of political parties Exactly. And and uh and and and the whole ecosystem. So and this is what I was trying to say earlier that look the value add is in supplying gases, supplying clean water. Uh because you know even the layers of photoresist and uh insulator and metal that you put and then you have to clean it up. The chemicals need for cleaning. You know they need to be extremely high purity water you know there’s so many other suppliers that emerge. There is so much value add in the ecosystem. That is one. two once you have that that expertise can be deployed in several other industries. So it is it is also a part of just technical training at a giant scale. At the same time as much as I think we should try to maximize the value ad we can have in India. This is one sector where one shouldn’t even try to be completely atar because uh you know the reason why this industry is done so well it is that M’s law has progressed so nicely is that uh you know so of these 5070 blocks that I said that everything there is an oligopy so that oligopoly is not uniformly distributed geographically so you need to have you know the memory companies in Japan and Korea and GPUs in US designed in US and fabbed in Taiwan. So somewhere in this 50 60 center of expertise, India needs to also carve its place and exist and be part of those oligopali somewhere. Correct. So so right now uh I think we backed some excellent entrepreneurs with substantial amounts of capital. The hope is that in five, seven, 10 years, remember that when TSMC started and these are things that are much more detailed and much more colorfully described and repeat all that one uh industry move that Morris Chang took advantage of or TSMC took advantage of was the separation of IDMs into fabulous and fun. So they were independent device manufacturers. So basically the same company did design and manufacturing and he figured that it was getting starting to get so complicated. Some companies would have to focus on manufacturing and some companies would have to focus on design and he said you know what we are only going to do manufacturing we are not going to compete with design companies and therefore they can trust us and we can we will collaborate intensively with the cutting edge design companies and that’s how that is how uh they have out competed so I remember when I was in Taiwan UMC and TMC were considered competitors TSMC had started to establish establish but the fact that UMC will be a 4050$50 billion which is a lot but but TSMC is in trillions we wouldn’t have imagined that so it is possible through consistent innovation consistent quality consistent focus on research making big bets so Morris Chang coming back from retirement in 2019 10 and that famous deal with Apple those were strategic bets that played out phenomenally well for them I’m sure if you are part of the game as of now we are not even part of the game once you are part of the game you will start seeing those opportunities right so good news is through because of effort of semiconductor mission there is now at least hope and there’s something happening on ground the real plants are being set up and so I guess I can on the closing note I’ll just say that people who are interested in a deep tech uh ecosystem of India should very closely watch what happens this first set of companies as well as going to come in ISM2 and perhaps attend also this uh semicon India which happens but last once a year so yeah September uh right yeah so that’s probably become a big event on the industry calendar so this time this time the u you know the stalls there was a lot of attention yeah I’m almost tempted to kind of go into totally different you know segment of let’s walk through exactly how a chip is manufactured soup to nuts but we’ll park that for a separate episode later so thanks so much for giving us a full overview of you know where we are with the whole ISM story thank you thank At Sparks, we aim to bring to you stories of exponential impact. We share in-depth analysis of what goes behind success stories. If you find our conversations interesting, you can join us by subscribing to our YouTube channel. You can also listen to Sparks on Spotify, Apple Podcast or any other audio platform of your choice. If you have any suggestions on who we should invite or what topics we need to cover, just let us know in the comments. We are always listening, looking for ways to improve and keep getting better as we go