Catching up with Steve Sanghi who was Microchip CEO for 31 years, now Executive Chair.
Check out his new book, Up and to the Right.
How Microchip was built into one of the biggest and most pofitable chip companies in the world.
This is copied from The Amp Hour podcast, as it shuld be of great interest to EEVblog viewers.
In this episode we discussed:
The early microcontrollers, ROM, EPROM, EEPROM, FLASH.
Semiconductor Fab's and locations.
How the US Government CHIPS act is a golden handcuffs trap.
US/China rhetoric.
What really happened durign the covid supply chain crisis.
How 90% of medical devices use Microchip parts.
Lead times.
Respect for the chip manufacturers and supply chains.
Just-In-Time becomes Just-In-Case.
How the Toyota production system works and how it had to change.
The impacts of a potential war in Taiwan.
How does it make sense to make thousands of part variants?
The secret to making part varients.
Why Microchip chose MIPS vs ARM.
Atmel didn't make any money!
Pricing discipline.
How buying Atmel almost didn't happen, and how Dialog Semiconductor goofed it.
Draconian NDA terms.
Atmel was bloated.
RISC-V plans.
Customer Driven Obsolescence.
Foundry vs In-House limitations.
Open Source FPGA tools?
Third party tool support.
Why not offer free optimised compilers?
It's all about black swan events, down cycles, pointy haired bosses, and how All your Cost Bases Beyong To Me.
Hobbyist vs professionals.
How Microchip is the largest aerospace chip maker in the world. Nothing leaves earth without a Microchip part in it.
Radiation hardended parts.
Market value profitability.
Will Microchip ever get acquired.
Strained US vs China relations.
If you find my videos useful you may consider supporting the EEVblog on Patreon:
Web Site:
Other channels:
#ElectronicsCreators #Podcast #microchip

This is the Amp Hour Podcast Released May 15, 2023 Episode number 632 with Steve Sangney discussing being CEO of Microchip Corporation for 31 years welcome to the Amp Hour I'm Dave Jones from the Eev blog and I'm Steve Sangee for 31 years I was a CEO of Microchip now the Executive chair of the board Steve Great to talk to you again. How long's it been 13 years since we last had a chat? Yes, Dave that was a long time ago. I think I would have picked it, three had just come out and that's right. There are some things you didn't like about it and I think that no, yes and then you guys made a hilarious response video and uh, yeah, that was great.

That was good times. So 31 years as CEO of Microchip and you've only recently, Um stood down and now your Executive Chairman. Is that right? Yes, I stood down in March of 2021 two years ago, right? And why was that you just? you've had long enough there? You wanted to retire, hand the reins over I Was you know, almost 66 years old and uh, I needed to pass it on to my successor Ganesh Murti. and uh, you know? so I I'm still working for Microchip three days a week, right? But you have ample time.

So I wanted to. Really? You know my family made a lot of sacrifice over those 40 years or so. Yeah, yes, and I wanted to really spend more time with them and reduce my hours at Microchip a little bit. I was working seven days a week and traveling 100 nights every year and it was just oh yeah, that's tough.

Yeah, no I can understand. So you've now written a book. We might talk about a bit more at the end, but uh, is that what's that about? The title of the book is up and to the right. that's the main title and then the subtitle is my personal and business Journey Building the Microchip Technology Juggernug.

So the book is really my you know 31 year history of building microchip and also has a little bit of personal stuff. my growing up as a small boy in India How I came to United States and you know, early years I went through a lot of hardship like most immigrants and how I got into business and then microchip and just a tremendous success building a 43 billion dollar market value Juggernaut So that's really what the book is about that is amazing. Hey, can you tell us very briefly I Don't want to go through the book? Uh, but tell us very briefly. how do you become uh, the CEO of a large company like? well, it was.

It was it. Arizona Microchip. back in the day I do have a recollection that it was called Arizona Microchip. Is that right? The name of the company was never Arizona Microchip.

It was Microchip Technology Inc right? It's a European name and some of the Asian countries they branded Arizona Microchip. Ah, but Microchip was too generous Arizona to it. But here in the U.S or the main company's name was always Microchip Technology Inc got it. So how do you become CEO of a company like that? What's the path? What's the well? there are two ways to become CEO of a large company.
Uh, one is to become the CFA small company and then make it large, right? You know which is what I did I Joined the company when I was a 60 million dollar company losing money. you know? we turned it around, took it public, and just built it a Absolutely. We just completed a fiscal year with sales of 8.4 billion dollars. Nice.

So you didn't start out as CEO though you were working there as uh, something else at the time were you I was I came to Microchip as a senior Vice President of Operations and after one month of me being here, the board took out the CEO because the company wasn't doing well right and then there was a three months of search out. You know they talked to internal and external candidates and then I got the job. About four months after I came to Microchip. Oh fantastic Because Microchip almost went bankrupt in 1990, didn't they is that before or after you took the reins? So Microchip really never went bankrupt I almost almost did.

So after I took the job, we basically were losing money and didn't have enough money. Venture Capital was very hard to come by I Wrote a business plan and started raising money. Took me about nine months to raise money during those nine months. every other.

Tuesday I Wouldn't know whether the paychecks will clear on a Friday Oh ouch, that's that is rough. So if any of those Fridays the paychecks had not cleared then then we essentially will file bankruptcy. But we never did. We turned the company profitable I Raised new money two years later, took it public and up ever since.

Fantastic! So did you change our product strategy or something? Did you? Is that how you helped turn it around? or was it the just the Uh Capital raising that allowed you to buy time to Drive in different directions. Massive overhaul and um, you know. Actually the second chapter of the book talks about the transformation of Microchip from what we were to what we have become. Microchip originally was an Eprom company.

We used to make Eprom memory and you know the prices in E-pron memory products are very low. It's a commodity market and and a very very competitive prices from Japanese and Koreans and others. So the company was really losing money selling Eprons and the strategy we changed was to turn the company into a microcontroller company. We took the pick architecture which was you know, I had some two or three rom-based products about a million dollars a quarter business.

So it was very small business and we applied our Eprom technology to the microcontroller architecture to make the field programmable microcontrollers which could be programmed by the customer rather than programming in the factory by sending your code to Motorola and waiting for 16 weeks. You know for the parts and during that time your code changes and you start over again. That was game changing wasn't it? I mean that was the was the Pix16 C54 that was the first one exactly. It was a game changer.
It was absolutely in-game changer. and I write eloquently in the book about you know how the field programmability changed the game. There's a story about it which is also in the book. I was visiting a customer in Europe in Germany there was building remote controls so they were making remote controls for every single TV audio video karaoke, gradual openers.

they they were making about 500 different remote control models in a given year. Wow and they used to make them for you know Sony and Seiko and all the you know all the video audio TV players so but they were only making remote controls for other people. So and they were using Motorola product I mean obviously rambased product. So I went to visit the customer and the President showed up in the meeting.

So I asked the president my favorite question. which is you know dear sir, you know what is it that you cannot do today but would love to do if you only knew that it could be done? It's a it's a powerful and profound question if you think about it. Totally did it a couple of times and I said what is it that you cannot do today but would love to do if you only knew that it could be done He said follow me to the production floor, put a smocks back and he said you know, look at all those pile of parks, there are obsolete codes and the parts are scrapped. Look at those 14 people behind the behind the glass.

they're all on the phone with Motorola trying to expedite the parts on which the demand got stronger and you know it's like we're going crazy. We have to forecast 16 weeks ahead of time in exact model mix of those 500 different codes. Can you can you help me I said yes sir Music to my ears I said change all those parts to a microchip microcontroller. then all you have to do is buy the number of remote controls you build in a month.

no model mix issue. Your night shift programs it and you do it. Day shift puts them into the circuit and you know, guess who got the business exactly? Oh well, yeah, because so you guys funds who had a reprogrammable micro controller Field field programmable. So I think I I think to be honest Motorola Intel I believe even Hitachi had a Eprom based field programmable part.

but there were about 10 extra price. Oh so they were not used for production. Everybody can program and reprogram. use it for production.

you use it for pro you know, prototyping or for very high value added products. Very expensive products that's right. So what we brought to the market was a field programmable product which was only I think about 20. 25 higher price than ROM but tremendous value in time to Market Wow! and then so when when did you make the switch from E-squared Prom to Flash? So the the So First we went from Eprom to E-square I think there was a Pix16 C84 which was a easy.

That's right. Yeah, all right and you can really byte program and change a single byte. And then around the 1995 time frame we introduced The Flash based products where you can Flash erase the whole part and Flash program the whole part right? And what was like how do you change your Fab like process node or whatever it is to change from Uh Eprom to then E Squared problem and then to flash. Is it like a major upgrade to your Fab Or how does that? How does that work? Those were all distinct.
Technologies So today we still make the original pix16c54. Yeah and the Reprim technology still runs in a verb. We still make the 16 C84 which is an E-square technology and we still make it. And we make whole bunch of flash parts that you can buy from 16f 505 to you know, even 16-bit parts and high-end 8-bit Parts.

You know they're two thousand, three thousand different variety of pit microcontrollers today. So all those parts run into the Fab. So each one was a different technology that we developed and then implemented and put the designs on those Technologies right? So how easy is it to change your fabric? Like to like suddenly spin up a flash part as opposed to a E-squared prom part. Is it like really easy to do that? You mean um, the designs are already there.

Process: Instead the designs are already there. How do you change the process On the manufacturing floor so to speak? So all those products, all those processes are in production volume and I can today you know start a hundred thousand parts of Uh 16 C54 Tomorrow I can start a mask set which runs on the E-square process and the day after I can start the mask set which runs on the flash process. Or on a given day you can make all three. There is enough for the Fab to really make all those parts in parallel.

Oh wow. Okay, Fantastic. So how many how many Fabs do you guys have and where are they? So Microchip owns three large Fabs the uh, the oldest one that we we have it in Arizona it's in Tempe and then the second one is in Oregon. It's right outside of Portland in a city called Gresham.

So the temp in Oregon Fab are eight inch Fabs and then we have a third Fab which is a six inch Fab which is in Colorado and we got it with when we purchase Atmel it was at Mel's fair. Oh right. Yep and then we use a you know we probably use 20 other Fabs around the world in Taiwan and Japan and China basically products that we buy from subcontractors like Tsmc Global, Foundry, UMC and others got it. So what percentage of your products would be made in your own Fabs Do you think so about 42 of the products that roughly made in our Fab Oh wow, that's when making Foundry so you know that number was almost 98 made inside.

prior to us starting you know the Acquisitions yeah and starting to build the products in Tsmc. So I would say in almost till about 2000 8 2009 we made most of the products ourselves right? Then two things happened. Number one, we started to do acquisition. so when we for example bought Smsc in 2012, they were 100 fabulous.
SSD In 2010 they were 100 fabulous. When we bought Micro Semi in 2018, you know they were largely fabulous. So so when we bought all those companies then fair amount of our business started to be from the foundries. and the second reason is as the technology migrated up especially to 300 Wafers and past the 90 nanometer, 90 nanometer, 65, 40 nanometer, 28 nanometer, all those are very Advanced lithographies and microchip doesn't have 300 millimeter capability inside so we're buying from foundries.

So as that business grew that also increased the portion of the business we get from outside. Got it? But you are looking at building your own. A new 300 millimeter Fabian Aggression aren't you? Is that still happening? Or so We we did a large study whether we could do a 300 millimeter Fab owned by us with a chip sacked Grant from the U.S government that was approved a few months ago that you could apply for the chips grant money. but we were doing that because our foundries had told us that they're not going to add any trailing 300 millimeter capacity.

so they were. Foundry has told us that they would not add any 300 millimeter capacity at the 90 nanometer, 65 nanometer, and 40 nanometer nodes. So we had decided to do ourselves right. We were still, you know, working on and getting the money and laying the plans.

Um, the the foundries changed their decision and they said they will add trailing 300 millimeter Fabs because apparently their original analysis was that the trailing Edge demand wouldn't be there. But you know a lot of the you know a lot of the equipment are com You know Industrial Equipment Automobiles You know all sorts of you know, products built. They're all built on trailing Edge technology. so all the analog is built on trailing Edge Technologies It's only the advanced processors by Intel and Nvidia and AMD and Qualcomm.

You know they're built on what we call bleeding edge technology. Yeah, all of Microchips microcontrollers are built on trailing Edge Technologies. So as The Foundry started to get that feedback from us and ADI and TI and Nxp and St micro and Renaissance and others, they change their mind and they said they would add more trailing Edge capacity. So we we scrapped our plans.

We'd rather have good scrap. Okay, so even if they even if the government gives you money through the chips act, you won't build a new Fab You don't think you will. Currently we are planning to expand our eight inch and six inch Fabs because the demand on them is tremendous and we'll get some chips help for that 300 millimeter Fab from chipset. Well you know the government did little bit bait and switch also right? So basically uh, they came up with the rules that if you take more than 150 million dollars help from the government for a given project and a 300 millimeter Fab will be a five billion dollar project so we'll get a lot of help from the government.
Then you have to put a you know, unionized Workforce You have to build a daycare center. You have to share your profits with the government stock. You cannot increase your dividend. Oh wow, Yeah, so it's like, you know, hey, you know that's so you're asking us to swim in a highly competitive world I Yeah, I Think the name for that is uh, golden Handcuffs.

Exactly. So so number of companies have decided to not take the government money and we're going to take the government money, but stay below the 150 million dollar per project so we don't have to do all those things And and use that money to expand our existing fats? Got it? That makes sense. Okay, so you wouldn't be able would. Would you actually be able to get Capital somewhere to build your own? Fab That wasn't government money? Would that even be possible? Well, Amino we are.

You know number one. We're very profitable. We can fund Affair ourselves if if we absolutely needed the Fab and if the if uh, you know the foundries had told us that they won't build a trailing Edge technology and they had not changed their decision, then we will build a 300 millimeter Fab and we probably will only take 150 million dollars from the government and rest week in cash flow ourselves. Uh, we're very profitable.

We can borrow money, we can sell our stock to raise money, we can sell bonds. We can do a lot of things. So it the the problem wasn't really being able to fund the Fab. The problem is, you know, building the Fab in U.S with all the regulations and all that is more expensive you know than building a Fab in some other countries.

So we wanted a government subsidies for the Fab can be cost effective. but if they give you a subsidy on one hand and raise the cost even higher on the other end, then it's a mute point. Got it? Are you making any chips in China at the moment through any Fabs in China We are making very few chips over the last several years as the rhetoric between us. And yeah, we have largely withdrawn from Manufacturing in China.

We own no assembly or test plant in China Some of the packaging we do, we have moved it outside. You know we probably do less than five percent of our packaging in China right away for Fabs. We probably you know do only maybe two or three percent of our Manufacturing in China and wafer fraps. Interesting.

So what about raw materials and things like that? are you Reliant upon China For those like, if say if trade with China stopped uh tomorrow would your Fabs not be able to get material Or so we we are aligned on China on doors. But even on those things, we have uh built second sources for most of the materials right? So if the if the supply line were to get totally cut off from China which nobody's expecting it no But then then you know the other sources which we have qualified you know will have a lot of constraints on them and we won't be able to get everything we want. So we're still relying on China for a lot of the you know, a lot of those rare earth materials and other things, but the sources are being developed outside of China to become self-independent there too. What what sort of rare earth materials are used in Chip? Uh, Fabs I wouldn't be able to name them, but Palladium is one of them for example.
And there are others. Okay, interesting, can we talk about the supply chain and covert? Here's a big. here's a big topic. Can you tell us what happened during covert? Because my, let's let's see if my story ties up with well, from what I've heard, ties up with what you know? Um, I The supply chain crisis happened because when covert came around all the major Automotive manufacturers and all the big manufacturing you know, all your big makers like Apple and everything they stopped.

They panicked and stopped ordering parts which then meant you guys if that changed all your pipeline and everything and then six months into it they realized that uh oh no, we better start making chips again and then you guys couldn't ramp up fast enough. Is that what happened or somewhere that happened? Well, you know it's more complex than that. So right, Tell us so you know. Kobe Started in about March of 2020 and uh, in June of 2020 the automotive demand had gone to about 20 of Norma.

So Automotive manufacturers cut back that much. not for the whole quarter, but it was sliding down and in the month of June, it was about 20 of normal. Wow. And and we, uh, you know, we begged Automotive manufacturers that that can't be a reality.

You're just responding to this short-term thing. Give us a longer term forecast. what happens in six months, what happens in a year, So we maintain capacity and you know and not leave you high and dry. And nobody was giving any forecast.

Nobody was talking, Nobody knew what's going to happen. So while the automotive demand was down, the demand in two areas skyrocketed right? One was in a work from home. Yes. So I would say large number of our employees have a desktop in their office and when they started working from home, we had to buy them a laptop.

right? Instant demand for laptops and monitors, external monitors and things. Yeah, to buy them a laptop and monitor it. You know, a dark docking station? Yeah, either. Broadband at house, in the house, a printer to be able to print things.

So the demand for things needed for work from home, you know, even headphones and others. It skyrocketed and the second thing that happened was medical. So I'll give you an example on medication. You know in in March of 2020 if somebody had asked me, you know what is a ventilator? Yeah, it must be some sort of fan.
maybe an attic fan in Phoenix to get the heat up. But then I learned that uh, ventilator you know was this pump that pumped oxygen into the lungs of the patients that couldn't breathe. and every design around the world I would say 90 plus of the designs around the world had a pick in it. Oh really and that demand went up a thousand? X Oh a thousand times.

Wow, Yeah, a thousand times. And it was not only ventilators also you know, digital thermometers, scanners, the glucose pumps and you know and and diagnostic equipment. And when the covert test kit started to come out, they all had our products in it too. So the demand for medical just skyrocketed and took us more than a year year and a half to supply that demand.

So so the automotive capacity got shifted to medical and work from home. and then four or five months later when the automotive came back and saying hey, we want a part we had no other capacity. Wow and then everybody got constrained and still are I'm an elite Times are still, you know, 26 to 40 weeks on most of the products. Yeah, that was common back in like the late 90s wasn't it? I Can remember like 40 week lead times being very common and then and then it seemed to be plentiful in the last 15 years or something and now it's like 40 weeks again.

Yeah so you know we are on record to say that we bring the we'll bring the lead times on all of our products you know down below 26 weeks by end of September away and I think and then continue to bring down. Beyond that other companies are doing the same thing. so things are normalizing from covert. You know the medical demand has been met, the PC demand has been met.

Actually PCS are soft. um Automotive demand is rising but now the switch is happening. You know the the capacity we had given it to PCS and worked from home in medical. Now we can give that capacity back to Automotive So we're catching up us.

Engineers We think we foolishly think that the that the chips are the bottom of the supply chain like oh I I can't get my chips therefore I can't build my products but to you it must be different. It's like well if there's a supply chain crisis I can't get my Wafers I can't get my chemicals I can't get you know all the things I need to build those chips. It's like uh turtles all the way down. Did you guys have any um, supply chain problems for stuff to make the chips? Yeah we had huge problems to make the chips so you know we were expanding our Fab and we're adding equipment in our fair and when we will order the equipment you know the companies like Applied Material and clearly and others we were buying the equipment from will give us a delivery date when the equipment will be shipped and let's say they told us it will arrive in September 2020.

and it didn't arrive in September 2028 Arrived six months later. What happened because many of them couldn't get semiconductors to build the equipment. so in many cases we we found out if it was our chip. Then we preferentially gave the tip to those people who will help us give the equipment but we were not the only chip in those equipment even if he gave ours.
maybe they couldn't get it from TI they couldn't get it from somebody else. So Eventually everybody figured out to help the gear manufacturer so they can build enough gear to truly so that we can improve production. but that was not the only thing then our stuff let's say was coming from Japan and it said outside of Los Angeles Harbor on the ships for eight weeks. Oh right, because they they couldn't get the port workers or whatever.

the The Dock Workers is that right? Yeah and you know they couldn't get. uh, you know enough, they couldn't get the porch to be working efficiently enough. They were union workers. They were working.

You know, one shift and you know it was overloaded. So ships set out in the ocean for six to eight weeks. So by the time they were able to offload our stuff and they put it on the ground, our truckers had built out saying we got other jobs to do. You know, we've been waiting for a week to wait for your equipment and we're leaving.

So then it took us four weeks to get the truckers back. They took the equipment to Oregon where we needed to install it. By that time, the installers had bailed out and saying, well, you know we got other work we got to make money. We've been waiting for a week for the equipment to arrive.

So there were delays After delays. After delays after delays. So therefore the commitments that we had made to the customer a year out based on our capacity growth plans. We had to break those commitments because that capacity growth didn't happen because we didn't get the equipment.

Wow. That is like it's it. It's It's hilarious that you were able to supply actually prioritize your chips to make the equipment. but then they got stuck in the in the on the docks or sitting on the ships and you couldn't get to them.

That's wow. That is crazy. Did you have any problems with like uh, you know chemicals and things like that because Fabs use lots of chemicals and whatnot. We we did have uh, some issues with chemicals but I think we're hand them out right? We never shut down because of chemicals because there was a fair amount of chemical Supply available on Shore From you know, lots of people that make chemicals.

So we did come you know hand to mouth few times where you know ordinarily we'll have you know, six weeks of supply of chemical in-house and you know we never want to shut the Fab for a stupid chemical. Yeah and then we came within a week of shutting down. We never shut down because of that. Another thing point I wanted to make is uh, you don't wear uh Historically over the last 40 50 years customers have dealt with the chips as jelly beans.
Yes no whenever you want them they will be there. You know Automotive Suppliers never paid enough respect to the semiconductor suppliers because you know I got lots of suppliers and chips are available and you know whenever I need them I will have them and they for for a decade. Plus they were driving these Jit program just in time I don't want any inventory? Yeah all the inventory give me what I need and an automotive manufacturers got a big lesson in the last two three years three years. Basically what happened is for the shortage of a two dollar chip they couldn't ship 80 000 car Yes! and they lost so much money billions of dollars that most auto manufacturers have now changed their strategy from jit to Jic.

Jit was just in time. Jic is just in case. now they're all planning to start an inventory a product. So if this kind of thing happens again, they will not have billions of dollars of cars that they cannot ship.

Yes! I Heard that even Toyota did that and Toyota basically invented just in time didn't they with the Toyota production system? Absolutely. I Mean if you go, you know, look at in Japan Wherever the Toyota's factories are the, you know the big trucks and big rigs are lined up in the alleys behind Toyota's plans and they are supposed to deliver the product to Toyota every four hours Toyota keeps no inventory, truck pulls up and supplies it and gets on the back of the line. That's how it worked and and you know now all those trucks were empty. There were no parts in it.

So Toyota had to shut down factories too. Oh wow, yeah, that's a lesson hard learned. really. it worked for what 30 years though, didn't it? It worked for a long time.

So I think now you know semiconductor industry is getting healthy respect, getting respect from Automotive guys getting respect from the rest of the market, getting the expect from the governments and they're basically you know semiconductors are the you know the new oil of the economy right? Yes, you know the crude oil was. You know the Whoever has the oil, has the power because energy was needed and now nothing happens without semiconductors. So they're becoming the new oil, right? Can you? I I Don't really want to get into this, but it's what. I probably have to ask.

What do you think would happen if China invade? Taiwan I mean how big a disruption would that be with Taiwan being what? 70 of the chip Market globally is that right? Does Taiwan make 70 of its chips? I think that number is high? Okay, yeah so but it's like you know it's probably 70 of the advanced manufacturers of the advanced ones. Yeah, that's probably what it is. Yeah, you know if China language Taiwan that's a huge problem and um, you know everybody, number one is hoping that it wouldn't. You know wisdom will prevail.

Secondly, if China does invade Taiwan shutting down words chip Supply I'm not sure how it helps China because of old world and you know shuts off everything. the US you know shut down Huawei a few years ago and if all the chips apply and all the raw materials and everything else stop going to China then China become the third world country. So I think it's kind of a you know Mutual destruction and everybody's hoping that it will happen. Yeah, I suspect.
So so let's talk about the the sheer number of Parts This was a question that uh, some people asked you make over I think over 1200 variants of just the 8-Bit micro controllers. How does it make sense to make so many minor variants of you know, one family or one part? How does it make sense from a production point of view from a supply chain point of view, how does that work? So I think uh you know over the years you know for 30 years and let's say let's go back to you know when 8-bit microcontrollers were you know, six seven years old? So we started them in 1990 and let's say 1998 time frame? You know we would. We would analyze our design wins and design losses and we will find design losses because you know, uh, we didn't have a Pwm on the chip and somebody else did right? We had you know one 8-bit A to D and you know somebody wanted A to D and two Pwm channels or we had. You know what would happen is customer would say I love your architecture I love your product.

it's great but I wanted you know Um but I wanted a USB on board so we'll say well here's a part with Usbn board but that only has 40 pins I want 60. Okay here's the part with 68 and USB but that doesn't have enough Pwm channels so that really happened because you know customers wanted it and we wanted to win every design. So we kept it proliferating it and we had to invent a system in our Manufacturing where you can order any one of those 1200 variants and we will make them and make them cost effectively and keep our inventories in control and keep our business very profitable. You know which is, by the way is a you know is a skill in itself.

Now you know Costco knows how to do it Walmart knows how to do it but I don't think Intel knows how to do it and I don't think AMD Nividi I don't know how to do it right and and many of the other semiconductor companies Qualcomm and others may not know how to do it. Microchip does and obviously analog guys knew how to do it Max Yeah and then you how to do it ADI knows how to do it and we know how to do it. So we we did it because you know we would lose design because you know customer didn't have the exact part and many times we would then make a super chip that had a bunch of Pwms and A to Z USB and heavy internet and have all that stuff and then we will lose the design because our part wasn't cost effective because we put too much into it. Got it? Are you going to share with us what the secret is there? or is that A? or is that a trade secret? Are there any tricks to being able to produce so many parts on a whim? like just bang? That's probably a trade secret? I Think others have it too and we don't know if the others do it exactly how we do it, you know? But I mean we're not the only one who does it.
You know who's good at it. I I would say ADI is good at it I would say Tia is good at it. so there are other people who are good at it. but I I do not know whether how we do it and how they do it is exactly the same.

I think TI does it through inventory. Yeah, you know their inventory is a lot higher than us, so they kind of do it through inventory. We don't really do it through inventory and and therefore I think yeah, it's probably. you know some others would like to know what Microchip does the Mips versus Arm thing.

Can you ask the history behind that? Why did you go into Nips when everyone was doing? Um, so that was long time ago. We had essentially a bake off between Mips and Arm and we found that arm's proposition in terms of upfront license fee and royalties were much more. Draconian because of their market share than Mips were and and a lot of our low-cost competitors like Atmel. I Mean Ratmill will sell a chip for half the price we would and they won't make any money at it.

But it didn't matter. You know, when I bought Atmel in 2016, they were making zero profit. Wow! And today they're making over 40 operating profit under Microchip. so they didn't have the pricing discipline they you know absolutely did not.

So you know, if we had gone to Arm then competing with a lot of low-cost competition and having no differentiation, we thought we wouldn't be able to do what Microchip wants to do. Which means be differentiated, sell with profit, and you know, grow our business responsibly. We couldn't do it with arms. So we went to the Mips architecture and our Mips was quite successful.

And even after we bought Atmel which was armed, we looked at it. Three years after, our Mips business had grown as much as Arm had from the point of you know, acquisition and forward. So now you know we kind of eliminated some of the very low cost competition. So then after we bought Atmel, we kept doing both.

But eventually we decided that if it's a you know general purpose part, then we will build them on Arm. If it's a part where the core is not visible to the customer means we're building a seasick for a single customer, a single application and we put the program in it and the customer doesn't see the architecture then those ones we will still build with Mips because you know, financially, it's a better solution. Got it? So how did the Atmel? uh, buyout happen? Did they approach you? Did you guys like poached them Or and how did it go? How was the uh transition to find them So um, so Atmel's board had decided um to uh to sell the company. So they got investment bankers uh, which were, uh, you know I think catalyst is the name of the company that's uh, um, that's a Bankers who were trying to sell at mail.
So we were approached by the banker saying Atmel's board has hired them to seek for bidders for Atmel and uh, um, you know. And then you know lots of companies looked at Atmel, including us. There's an entire story about Admiral acquisition in the book, but originally what happened is uh, you know Atmel gave us a non-disclosure agreement to sign which I wouldn't sign. Oh, because it had some Draconian terms in it and I don't know whether they gave the same Draconian terms to other companies, but they did give it to us.

You may recall that we tried to buy it at Mel Hostile back in 2008. Oh, that's right. So their acquisition failed because it ran into the global financial crisis, but some of the bad blood at the management level stayed between Atmel and Microchip. So this was now 2015.

About seven years later when the board was trying to decide you know, sell it the the management CEO and and the bankers gave us a non-disclosure agreement that I wouldn't sign. So then, uh, everybody else was looking at it and we were not because we were outside, we wouldn't get any information So Eventually, Eventually, I reached out to their Chairman and I asked him to give me a meeting and that meeting happened in San Jose Where I impressed on the Chairman that Microchip is probably the most likely viable and qualified buyer for these products. You know, looking at a record of Acquisitions and how successfully we made them if he didn't include microchip in the process I think there'd be a lot of questions asked by investors and lawyers and others. And so he agreed to.

you know, essentially, you know, help us take a look at the company and he would soften the NDA a little bit, which he did. So subsequently we signed the NDA and then we made our offer to buy Atmel. So our original offer to buy was at nine dollars a share. Then chairman called me and said that we were not the highest offer.

there is a higher offer than us, but if I were to raise my offer to ten dollars then he'll be willing to swing the board to really go with us. I Said you know our offer was not pulled out of the air. it was done based on a lot of analysis and and all that and what we came up with in was a nine dollar value and I wouldn't raise it to ten dollars. So then we then we didn't hear anything from them and a few weeks later the news came that they signed the agreement to be bought by Dialogue Dialogue.

That's right, Semiconductor. But the offer from Dialogue was in the form of large amount of Dialogue stock and some cash. So mostly stock, but some cash. When that offer was announced, the Dialogue stock went down precipitously.

So when the Dialogue stock goes down, the value of Atmos offer goes down. Yeah and every day every week Dialogue stock would go down and the value of Atmel offer will go down. The original offer started with ten dollars and forty cents. When the deal was announced the value of the offer was 10.40 and it kept going down down down and it went below nine dollars when it below nine dollars and I wrote to the Chairman again and I said mine now is a better offer and then that resulted into US engaging again.
But now at Mill's business was also starting to fall so we said we need some new number, new information where the business is and we got those. Based on that we eventually were able to buy Atmel at eight dollars and fifteen cents. Wow! and how's the transition been in terms of uh, you know, merging their products in? have you discontinued Admiral Parts at all? I Don't know. Yeah.

so we did not discontinue any Admiral pods. We usually don't. The strategy is that we build all the parts essentially forever. We uh, merge the Atmel businesses, their microcontroller business, merging our microcontroller business their you know 8-bit merge into a 8-bit the 32-bit merge together.

Uh, their E-square business C release per business merged with a city list Square business. so we did a fair amount of combination. There you know their RAF and wireless business merged into a wireless business and together we you know started to make rapid Improvement in products and yields and pricing and cost. And you know Atmel had a very bloated you know overhead structure.

There were spending huge amounts of money and you know finance and legal and HR and other disciplines that don't really add value to the product as much. So we did a significant restructuring. You know, Atmel in their headquarters where all the overhead was in San Jose had 400 people and we kept only 80. Oh wow, that's amazing.

So these days do you still more microchip microcontrollers? or Atmel microcontrollers? So they're all Microchips microcontrollers. We don't care what the customer wants to buy, we promote them equally and we are still building AVR products. and we're still building thick products. We're still building.

you know, Arm-based microcontrollers. Some of the new parts we build on arm they have a pick nomenclature on it. Some of the old parts that came from Atmel had a Sam but you know nomenclature on it? Got it? But which one sells more because people want to know? Are there more Avrs sold or picks? when we bought Atmel? Um, picks were selling 2x the AVR cells and in that distance it's probably even larger now. Oh okay, right.

Well, because you know for years Atwell had done, had not done any more AVR products. They have switched their entire energy to build a 32-bit arm. So therefore, Avrs were atrophene. So after Microchip came in, we said, you know we want to re-graduate you know AVR So a year later we introduced four new hard-hitting products with a lot of microchip features on AVR and we also made the AVR products work on Microchips a development tool set.
Um, so we re-energized AVR and they started growing again. But Picks are much, much larger business in India I Know that you have the Risk Five in the Polar Fire Fpgas. Do you have any plans on having Risk Five? Uh, like just chips on their own? We currently don't on a standalone microcontroller. We don't.

Um, because I think, uh, you know our customer base, which is, you know, largely automotive industrial. You know, you know those kind of customer base. their uh, they're really, very, uh, risk-averse right? and uh, you know they wouldn't unintended? Yeah, and we talked to them all the time. They wouldn't take the you know risk-based part today because they don't believe there is enough ecosystem.

So we are really using Risk Five where the architecture is not necessarily visible to the customer you know, or in Fpga Maybe you know in future, some of our data center products and others slowly walk the line and eventually we may do that someday. but not in the next two three years. So have you guys ever discontinued a part? you said that's that is not your thing And and I don't recall a Pic Micro ever being discontinued. So the the name we have there for that is, uh, customer driven Obsolescence.

So so as long as customer wants to buy the part, we will continue to make them. Now that one was a customer base. Obsolescence was perfect when we made all those parts internally because we could keep all those processes running with. You know, so much of a business now really gone to foundries.

You know we are not perfect in customer driven obsolescence. So Foundry word. Once in a while give us a notice that they're going to Absolute a process. Then we work with the foundries to go build five, ten years of inventory and still continue to supply the parts for a long time and create them in the minds of the customer.

It's a customer driven obsolescence, but we're not as perfect anymore. We've got I've got some questions from the audience. Um, so we'll rapid fire these ones. Do you have any plans for open source Fpga tool chains? I Do not believe.

So okay right because they're getting very popular now and a lot of people won't touch them if they don't have open source uh, platform support. I will ask that question but to my knowledge uh I think the answer probably right now is no. Okay, that leads into the next question. Any plans to support the Uh Platform IO IDE or any other third party because Microchips kind of famous for having their own ecosystem.

Yeah, so you know we support our AAR We support Kyle We support some others, but largely yes, We you know we prefer our own because I think we have got an inconsistent support from time to time. See what we could do with our products is we can have a development tool group line up their priorities with a microcontroller group. When the part comes out, the tool chain is there. If you rely on the outside guy and you go to them and say I you know I have this product coming in and I need the tool chain ready by June of 2024, he says I don't have resources I have other priorities I'll work on it a year from now.
Got it? So that's a problem. Well in terms of Uh platform, IO that's just an integrated development environment. It's not actually a like a compiler tool chain. So any plans to support third party development environments like that that are just like editor environments that are becoming very popular I I would probably have to find out and then let you know.

Okay, and I can really, you know, give that answer in a future session. Now this one comes up a lot. Why do microchip charge for compilers? or at least optimized uh compilers when most other manufacturers are now offering free compilers? So there is a there's a story behind it. Okay, you recall probably over the last uh, 35 years.

you know, Motorola or free Scale or whoever has spinned out the tools group two or three times. They spin it out and they you know, rebuild it I don't know if you if you know that or not. So what happens is if you run the tools business as a cross-center only and don't sell them and don't make money at it then when the down cycle comes like the 2008, you know, Financial Crisis 911. When these Black Swan events happen, some General manager says hey, you know I can't handle all this cost.

You know, putting a third party guys build our tools and you know this, shut it down or they spin it out. Just spin it out and it's fine. It looks like a great decision because you'll lower your cost and then two years later you're introducing your product and the other party doesn't get the tool chain ready because they tell you my priorities are different. You know.

same answer I had given you prior to that. Why We you know? Um, you know why we do our own. So uh, so we have seen other companies suffer from not having their own tool chain and sometimes, uh, their priorities and the priorities of the two chin supplier don't line up. So then Motorola at one time bought it back and you know you know, the buttered back and then few years later they sold it again.

you know, with a different management, different general manager, a different business cycle. So I I've learned this over time and in the last 33 years I've been at Microchip. I Said we will never shut down or sell our tools group. We will keep the priority of our tools for our microcontroller business.

Anybody else who provides support to our tool chain is in addition but not instead of us and we're gonna charge for it so we can afford to continue to build it. If we don't charge for it, then it becomes a cost center and then next General Manager or next CEO shuts it down. Got it? Yeah? I Think the problem here is that a lot of our audience are either hobbyists or they're like small business or something like that. And to them like free tool chains is a big business.
but that's probably a very small percentage of your Market is it? Yeah, well, but you know if they're Hobbies then they don't really need an optimized compiler. you know you're really your small volume customer. Then you can really use the version that is available for free and you don't really have to have an optimized compiler. And I think some of our compilers now you know you can rent them by month, right? Oh okay, fair enough.

Uh one last uh technical question on the Um chips and Fabs you guys do radiation hardened chips? How what sort of process is it for making those for like Aerospace Industry? Yeah so we are the largest Aerospace and defense supplier semiconductor supplier in the world really 2x larger than the nearest person. So we have very large number of chips for Aerospace and defense and many of them are red hard and red heart. Also is you know various different levels. It can be radiation tolerant and their radiation hardened.

and there's a category where they are good only in the lower orbit and there's another one which has to be good in the upper orbit. You know there are various ways to make them hardened. One way is where you design them with a larger tolerance, you know when the radiation hits, yeah it you know creates a lot of free electrons that move around and and really short the part. So if you build a part with you know larger variants then the part is more tolerant and the second way which almost always the case is that you Shield the device with you know these shields which are in processing on the wafer as well as Packaging which essentially doesn't let the radiation go through those Shields onto the dike Got it Interesting so there like very expensive parts to produce.

They are expensive parts to produce. Yeah, so they you know if a non-hardened part ships sells for two dollars, a radiation hardened part will sell for 200. Wow. but you know it in the shuttle, it's going in a satellite.

Yeah, Yeah, exactly. there's going to be up there for a long time. I Mean look at this, you know SpaceX They send a five billion dollar satellite and it blows up, right? Yeah, right? So you know there's a lot of money that's going into it and you know we get that pound of Flesh Oh my. God But I would say this that there is nobody who leaves Earth's orbit without microchip on board.

Wow. No shuttle, No satellite, No vehicle. No, you know anything. No.

Mars Lander No margin. There is no one who leaves Earth's atmosphere without microchip on board. Wow, That's incredible. Yeah, So where do you guys sit in the like? A list of largest chip makers both worldwide and in the US.

So I think in uh, in U.S Probably we're within top 10. Uh, you know could be. you know somewhere in the eight nine. I I'm really exactly seeing the ranking lately and if you add Global then you add some more like Samsung and others and you know we could be in the low teens area.
So that's what we said. You know that would be based on just the revenue, but another thing that's important is the market value because of profitability. You know, you know if you look at it that way, we're well within. uh, you know, top 10.

Yeah, you guys had a hundred quarters of profitability. 100 continuous 124 Now I think 124. Yeah, wow, is there any risk of microchip being acquired by anyone like because in theory somebody could just come and uh, hostile take over and buy all your shares? So so I think you know For example, this question was asked to a CEO of TI at one of the investor conferences you know about three years ago where investors said you know, hey, you know everybody's buying companies. look at ADI look at microchip, look at Renaissance TI Why don't you go buy microchip and the CEO's answer was I buy microchip and then I improve What? So so you know we are a very well-oiled machine.

extremely profitable, high gross margin you know, High operating margin, and a lot of different Industries running very well. So that's one issue that we're running very well so people don't see how they're going to improve. And to buy a company, you have to pay a significant premium and you have to improve them to earn the premium and like to actually make a profit out of it. Yeah, so like we did with that mail so they don't find what they can improve.

And second thing I think is right now trying to get Anti-Trust approval especially from China would be largely impossible because of the you know, strained trade relations between us. and China China is not approving any of those large deals. All right. So Microchip's not going to get bought out anytime soon.

Good to hear No. I Think you know if I could spend a minute on the book? Oh yes. Well I was going to finish up on that right now because our hour is up. So yes, please tell us about the book.

Yeah, so the title of the book is uh, up into the right and uh, the subtitle is my personal and business Journey Building the Microchip Technology Juggernaut and I think I you know, recommend the book to your audience. it's available on Amazon You know the book starts with you know Maya bringing growing up in India and then how I came to America I went through very difficult times I came to us with 150 in my pocket with nothing to fall back on. No place to stay, just admission to a Master's College and went through a lot of early struggles but you know. Then graduated with a 4.0 GPA from University of Massachusetts and went to work for Intel worked there for 10 years and at another small startup and then finally came to Microchip in 1990 and and then on.

After first chapter, it's largely building the Microchip Technology Juggernaut. You know how we transform the company, how we took it public, you know, went through them significant growth phase. You know in uh, in 19 You know in 1993 when we public we were 85 million dollar Revenue and you know by two thousand six. seven years later we were about 750 800 million dollar company You know had grown really just tremendous.
And uh, you know. then we started buying companies. and there are three chapters on negotiating the deals, buying the companies financing them, how we improve them, how we portrait our culture to microchip. There's a chapter on crisis management.

You know every company sooner or later will have a crisis and you know I Talked about crisis of Y2K then 911, then 2008, 2009 Global Financial crisis, then covet what we did. how we came out of Crisis stronger than we went in. Um, and you know how we gained market share I Talk about a strategic planning process I Talk about how we sustain The Continuous Improvement for 31 years I Talk about microchip as a top training organization, how we acquired low-cost manufacturing capacity, how we Delight our customers. We talk about succession planning and promoting internally.

you know our compensation system. And the final chapter before summary is a chapter which is titled 31 years at a glance. So when I was writing that, chapter I had pages and pages of charts. You know Revenue You know, Growing Gross Margin Growing Operating Margin growing Market share, growing stock Price growing dividend growing and all these charts and all these charts had one thing in common.

They were all up and to the right and and that's where the title of the book came from. Till I wrote that chapter I didn't have a title for the book that right? So I think I Think anybody, your audience who use Microchips products and have used them for many years, if they've used them for, you know, five years or 30 years, it doesn't matter. We'll relate to things in the book regarding you know how we did to build this Juggernaut and how we financed it, how we took it public, how we transformed it, how we grew it. So I think I think people will enjoy it and if they buy the book and they enjoy the book, then also, please ask them to, you know, give me a review on Amazon or review.

Yes, most important I Will definitely. Uh, get the Kindle version of that and I will give it a read and a review. Sure, that sounds absolutely fantastic, so let's recommend that one even though I only found out about it yesterday. so I haven't read it yet unfortunately.

Yeah! so Kindle version you can download in seconds. Well thank you very much Steve for joining us. This has been fantastic. Thank you David It was very enjoyable.

It's nice to catch up with you after probably about 25 years or so. Yeah, maybe it's not that much, exactly. Let's not take another 20 years, let's let's Well we could do a part two because I've got more questions so but we'll talk later today. Yeah, okay, thank you very much thank you.

Avatar photo


19 thoughts on “Talking with steve sanghi, ceo of microchip for 31 years”
  1. Avataaar/Circle Created with python_avatars Atheatos says:

    Great interview!
    One of my favorite companies / products to use. Along with Lattice.

  2. Avataaar/Circle Created with python_avatars Mohamed Lanjri says:

    The book should be sold with a uController!!

  3. Avataaar/Circle Created with python_avatars 💰 Make $575 Per Day says:

    "As we look ahead into the next century, leaders will be those who empower others." —Bill Gates

  4. Avataaar/Circle Created with python_avatars 💰 Make $575 Per Day says:

    "A real entrepreneur is somebody who has no safety net underneath them." —Henry Kravis

  5. Avataaar/Circle Created with python_avatars EꞩꞓaꝓꬲꟻꞧꭴꝳꞒꭴꝴtꞧꭴꝇꝇꬲdꮢꬲaꝇꭵtꝩ says:

    I like to thank Microchip for all the free 16F* samples I received as a kid.

  6. Avataaar/Circle Created with python_avatars : ) SMILE says:

    One of my favorite podcasts from the Amp Hour. I listened on my drive into work and replayed again on my drive home.

  7. Avataaar/Circle Created with python_avatars Rick Armstrong says:

    You are a Brilliant Soul Steve Sanghi ! Great talk between You two Gentelmen I Enjoyed this ! Thanks Dave! Cheers!!! Rick Armstrong Aurora Ontario Canada

  8. Avataaar/Circle Created with python_avatars Nabil says:

    Good content and great questions 👍🏾

  9. Avataaar/Circle Created with python_avatars Samantha Kerridge says:

    The 16f84 got me into microcontrollers. I wanted a stepper controller and the only way I could get one to do what I wanted was to roll my own. Thanks to microchip I'm still obsessed with electronics with microcontrollers at the heart. This interview is amazing

  10. Avataaar/Circle Created with python_avatars Pebbles says:

    Many Thanks Steve Sanghi for your time and glimpse into the world of the semiconductor device industry.

  11. Avataaar/Circle Created with python_avatars Pebbles says:

    A fun fact – Microchip bought Aussie software company HITECH C! HITECH C was great for the Microbee’s Z80 and Applix1616’s 68000.

  12. Avataaar/Circle Created with python_avatars Hola! Zoe Nagy says:

    wow insight!

  13. Avataaar/Circle Created with python_avatars Ray says:

    Is that a library picture Dave 😉

  14. Avataaar/Circle Created with python_avatars Piotr Ludorowski says:

    Nice talk. Realized they bought atmel 2016. They are strong in eeproms. Will think to buy his book. But i'm shocked thah they didnt talk about Atmega328p(b) :p
    Also as pcb designer i want to ask what AVRxxDA\DB\DD\DE will be available for longest amount of time.

  15. Avataaar/Circle Created with python_avatars Roberto Gatti says:

    Just now some PIC16Fxx are 72+ weeks …

  16. Avataaar/Circle Created with python_avatars Roberto Gatti says:

    Government printed money out of thin air…

  17. Avataaar/Circle Created with python_avatars Martin says:

    Microchip could benefit immensely by adding support for their FPGAs in the open source F4PGA toolchain project. Most engineers that I know would love to use the same open source FPGA toolchain across various FPGA products instead of the proprietary and painful toolchain mess that we are suffering today. In short, we need that gcc toolchain equivalent for FPGAs.

    Also, Microchip should make a good single/dual/quad core RISC-V SoC for embedded Linux IoT etc. We can't wait 2-3 years for that!

  18. Avataaar/Circle Created with python_avatars floodo1 says:

    fascinating actually ty

  19. Avataaar/Circle Created with python_avatars BabbageBrassWorks says:

    Extraordinary interview, no BS. Had 20 years in Semiconductor business, this is one of the best interviews. Due to acquisitions there are not many left who know the history. Hard to believe they still make the 16C84, think how many microprocessors have vanished over that same time frame.

Leave a Reply

Your email address will not be published. Required fields are marked *