Aart de Geus
March 7, 2016
RB:†††† Hello again. This is Robert Blair for Silicon Genesis at the Stanford University Libraries. Today is Monday, March the 7th, 2016. Weíll be talking to Aart de Geus, the founder and now chairman and CEO of the industry leading electronics design automation company, Synopsys. The semiconductor industry has changed the world over the past five decades, but it would also be fair to say that the EDA industry on which the semiconductor industry relies has changed the semiconductor industry itself over the last four decades. They are really a matched pair. Aart de Geus arrived in the US from Europe with an engineering degree from the Federal Polytechnic in Lausanne, Switzerland. He completed his PhD in Texas and took his passion and vision to found a company in California that became Synopsys, raising both industrial and venture funding in the booming 1980s electronics era. Synopsys chose to focus on logic synthesis, an IC design technique that would become a fundamental requirement as the size of the chips grew exponentially under Mooreís Law. With over thirty years as Synopsysí leader, and with well over sixty acquisitions under his belt, Aart has established himself as one of, if not the, industryís renowned figureheads. His hands-on style, his high energy and enthusiasm for his craft and his company make him an industry icon who has been recognized by his peers with many prestigious awards. We caught up with Aart at the Synopsys headquarters in Sunnyvale, California. So letís now hear from Aartís perspective on the history of the EDA industry and its impact on the world of the integrated circuit.
RB:†††† So Aart, welcome. Welcome to the Silicon Genesis at Stanford Libraries. Today weíd like to hear your personal story of the rise of the semiconductor industry. And, of course, being the founder of Synopsys with a strong EDA bias. So welcome to the interview, Aart.
AdG:† Thank you.
RB:†††† Letís start off by just talking about your early days if we can - your background. And I understand you were born in the Netherlands in í54. Tell us a little bit about your parents and how you were brought up and what actually steered you to such a technical career.
AdG:† Well as you probably realize so many things are accidental. I was born in the Netherlands but my parents immigrated to Switzerland when I was four years old. And although at home we spoke fluently Dutch and I still speak fluent in Dutch and we went back literally every vacation to the Netherlands. I spent the first half of my youth in the French speaking part along the Lake of Geneva, grew up there. And then they moved again to the German speaking part, to Basel, where I learned both the Swiss-German dialect and the so-called High German and, you know, went to high school there. My father was an economist by training and worked at Nestlť. That was the reason he went to Switzerland, which was a big deal. They were immigrants literally. Right. And then later on, he worked at what is today Novartis in Basel. And in the process as a kid, you know, you just sort of went for the ride, not that there was much choice. Maybe the other thing that I still look back to as very relevant is that during my youth, going back literally every summer to the Netherlands, I spent every summer on a farm. And I think that Iíve learned more on the farm in terms of how things work together, what teamwork is, what ownership of a job is like, than almost any other place because a farm is a fabulous ecosystem. Now youíre seven years old, you have to go get the eggs. And man those chickens, they are dangerous, right. They have a vicious eye on you and as a little kid getting the eyes is sort of a shaky business. Now, at the same time, when youíre 77 and you have to go get the eggs, you know, you the shaky person, you shouldnít break them either. My point is itís an ecosystem where at any point in time, no matter what your age is, you are treated completely like an adult by the definition of whatever you can do, thatís your job. And, as a kid, that is fabulous because youíre with the big guys. You get to drive the truck at twelve. You get to, you know, Iím well versed in driving tractors and the cool stuff for a fourteen-year-old. But, more importantly, it was really the attitude of everybody counts on you all the time. And, in many ways, great companies are like that too. And it doesnít matter if itís cleaning the windows before the customer arrives or if it is actually being responsible for some deep algorithm, excellence always matters, and accountability is part of the game.
RB:†††† Yep, I agree. My mother was from the farming side and I had my share of tractors early on as well. So what we are sort of interested this time, were they tended to be mostly technical or how did you sort of become a very technical guy ending up with a PhD? I mean, thatís a direction that you chose to go.
AdG:† I think it is as accidental as choice. You know, naturally I was well-equipped for math, science, and physics. And in school, that came easy and I liked it and loved it. But, you know, as a kid, playing in the neighborhood was just as cool. And I think there, there was sort of a natural leadership that was already evident because, you know, when you grow with kids, the questions alóso what are we doing to do? What are we going to do? Well I was the one that said weíre going to do this. And then, of course, then it turned out to be a great idea, not because it was a great idea, but because everybody participated. And so that leadership was somehow already apparent when I was seven or eight years old. And then the whole scholastic thing was sort of oh how do I make it through school without too many accidents [laughter] was mostly my attitude. And then somewhat by accident, maybe because of the math and so on, I ended up in electrical engineering going to undergrad back in the French-speaking part in Lausanne, which turned out to be a very good school and then decided, you know, to go to the US, not so much because of, you know, studying there, but because I was sort of getting out of everything and trying something different. And the studying was just an elegant way of doing that. So I worked for a year to get money and then ultimately landed at SMU in Dallas by accident. And that was definitely a different world then than the quaint Swiss environment.
RB:†††† Right, absolutely. I noticed at the institute in Lausanne where you were, there was a famous guy there about the same time as you, the inventor of Velcro.
AdG:† Oh really? Sticky stuff.
RB:†††† Yeah. The guy called George de Mestral, so the inventor of Velcro. So I was just thinking as I was making a note here, itíd be nice to have a royalty on that.
AdG:† For sure. It is sticky stuff in every which way.
RB:†††† Exactly, exactly, yeah. So what caused you to come to the US to do your post-grad? Why didnít you stay in Europe or in Switzerland?
AdG:† You know, I think fundamentally thereís this feeling of getting into something different. You know, it is the going away from home for sure. It is the seeing rest of the world. At that time, initially, I thought Canada partially because, you know, we were sort of in the aftermath of the Vietnam War. And so the US had a mixed perspective as viewed from the European side. But, you know, by the time you submit your paperwork to a number of colleges, it was a bit random how that came out. And yet that randomness brought about an unbelievably lucky moment. And the lucky moment is literally as I walk into SMU and I walk up to the office of the Electrical Engineering Department because you have to start somewhere, that same minute the new chairman of the Electrical Engineering Department walks in, a fellow by the name Ron Rohrer.
RB:†††† Very famous name.
AdG:† Famous name. I did not know that. I had no clue and I sort of just chatted to him. And somewhat unbeknownst to me, he quickly changed the paperwork and became my advisor. I had no clue what an advisor was so it never really registered. [laughter] And essentially took me under his wing because they must have seen some spark or something. And then I started to study the coursework. And he enabled me to get ten little computers that he had just received from TI, Texas Instruments. TI99/4 was the model, 16k memory, right. So very small. And the idea was hey, can you see if we could instrument some coursework to teach electrical engineering. And so within that 16k, I did the first Ohms Law which is, you know, hereís a voltage source and hereís a resistor and now can you compute the current and does the computer tell you if youíre right or wrong? And these things were unbelievably successfully. And very quickly, I got a couple of other students to do work for me. And, again, it was this transition in a matter of weeks of having an idea and becoming a manager. And we had some cool things. You know, we had the thing that would put up some random gates, connect it, and then time you. How quickly does it take with these inputs to figure out the output? Itís a fabulously quick way to learn what †is an NAND gate versus an NOR gate versus an inverter. You know, today it would be viewed as utterly simplistic compared to the Ė what you can do on a computer. But, at that time, it was fun stuff that really taught something well. And, of course, that caught the attention of a number of people and ultimately TI even tried to market this. So it was an accidental interaction. And then by the time Ė the time came around to say okay, what are you going to do for a PhD thesis, Ron had meanwhile already moved to Virginia for a variety of reasons, said oh why donít you come over for couple days at my place and then weíll look at a good, suitable thesis topic? I ended up staying there three months. And, you know, it gives new meaning to the word mentoring thatís for sure. But even then, I didnít fully appreciate who he was. And so let me tell you who Ron Rohrer is. This is the father of the program SPICE. And still today, you know, all the SPICE versions that exist are really derived from the work that he did at Berkeley in the early years. And initially the name of the program was called differently but it ended up being SPICE. And, of course, he pushed pretty hard to go why donít you do more circuit simulation, which was interesting but was ultimately not my bailiwick. But I did end up doing a thesis on mixed simulation that, you know, thatíll be forgotten based on what I have been doing since. But what Ron did is he put me in contact with other people that were leaders in the field. And, again, I did not realize that until many years later. What it didóit set the bar different because, you know, itís a little bit like if youíre a musician in a little town or if youíre a musician in New York. Thereís a big difference who you compare yourself to. And if youíre not conscious, well thatís obviously what happens. And so he helped me be on the Program Committees for Design Automation Conference for reviewing papers, which was quite challenging initially but very quickly sort of get the hang of it.
RB:†††† So was he at Virginia Tech or - ?
AdG:† Ah, so Ron has gone to many places. And while he maintained his professorship at SMU, he had meanwhile already gone to another university and landed at General Electric. And so while Iím at his place he says, ďYou know, why donít you start working for GE and weíll do this thesis thing at night?Ē Yeah, okay, great idea, right. Well, in hindsight, maybe not a great idea but it turned out to work fine except that at GE who had entered the semiconductor world, it started a program that ultimately became first known as Socrates and later as Design Compiler, still in use today, which was essentially synthesizing circuits. And the challenge with this thesis thing was that, of course, Socrates was really exciting and interesting and was really where my heart was and I had still to work on the simulator to finish a PhD, which ultimately I did. And thank you for that but, you know, my heart had alreadyóI was a digital guy really at heart more than anything.
RB:†††† But at least you had a job and presumably some money and -
AdG:† Yes, and I worked four on there and he gave, you know, freedom to work on the thesis except that, you know, I used the freedom to work on Socrates, which was really exciting. But really I owe him a great, great debt of gratitude as in so many cases, you know, accidental interactions with somebody who turns out to be visionary, capable, supportive, motivational, and, at times, critical, they just happened in life. Right.
RB:†††† Yeah, so the stars align well on the project, the ideas, the personality. The geography maybe wasnít ideal but it served its purpose.
AdG:† But, you know, at that point in time, geography was not an issue for me. And, you know, I was freewheeling, you know, I wasnít a kid but I felt like a kid looking back.
RB:†††† So then once youíre into that then Aart, then GE actually pulls the plug on the business and which sort of disrupts your thinking a little bit so you have to decide whatís next.
AdG:† Yeah, so there are a couple preambles to that story but letís fast forward massively. You remember in 2001 was the biggest downturn in semiconductor history and it was a tech downturn, meltdown after Y2K and dotcom, right. At that time, of course, I was long into Synopsys and Iím starting to say, you know, how profound is this?. And so Iím starting to do all these graphs of looking at how has the semiconductor industry grown over time. And as I do this, I see ups and downs. And I twisted the graph and to say for each up and down, let me do an individual graph and now you can see sort of from the top down to the bottom and back up. So you get these curves and you discover quickly that there have been a number of downturns in semiconductors. But what I also discovered is that in 1985 was the worst downturn of the eighties and nineties in semiconductors. And suddenly the light went off. That is one of the key reasons why General Electric that had invested massively to get into semiconductors decided to get out because hereís GE, this very large company at that time, very stable, you know, aircraft engines and refrigerators, financial systems, financially managed. And then they have this one piece semiconductor thatís only goes ka-chunk and a massive amount of profitability is lost. And suddenly a large company has to explain because a small portion goes up and down like a yo-yo...what to do. So they decided to get out of semiconductors, which then becomes the beginning of the story of Synopsys.
RB:†††† Right. I believe at the time, from memory, I believe we had a license deal with GE on gate arrays from LSI.
AdG:† Ah, so now youíre coming to that preamble that I wanted to talk to. Perfect. Perfect lead-in. Thank you. Which is that GE had entered the gate array business Ė what weíre talking about is sort of early 1980s. And for four or five years, they assembled a very capable team, both in terms of manufacturing. They had a fab. So they invested big money and they had a very substantial EDA team that was formed with some people that were well known and then some kids like myself that were sort of the hope of the future. And so they were into gate arrays, fundamentally NANDS NOR gates assembled in a somewhat preassembled fashion. Right. And so I had read a paper that said well, you know, NAND NOR inverters are cool but, you know, if you used multiplexers, you could actually be denser. And this is a paper from a guy, Shelly Akers, well known for essentially multiplexer logic. And so a friend of mine who was in the design team and I said hey, why donít we create a gate array that has multiplexers built in and we called it the multiplexer logic array, MLA. And if you have the footprint for that, then the designerís going to use that because itís going to be more efficient. Well so he created the footprint for that.† They didnít use it because they didnít know how to design with multiplexers. And so I thought well why donít we try to automate that and with some summer students I had, we wrote a program called Socrates that actually would create the logic using multiplexers and gates. And lo and behold, it was able to actually create logic pretty well from high level descriptions. Now, as a side note, multiplexers had other problems which is, depending on how you built them, if you want to be really efficient, they donít restore logic. So itís not like we can hook those things in long chains because then you have no signal left. So there were other complications to it and ultimately the multiplexer logic array as such was not really viable. But the Socrates program was a form of synthesizer. Now to show how little I knew, I hadnít realized that other people had worked on synthesis at that time.
RB:†††† It had been around for ten or twenty years I hear.
AdG:† Well, †all kinds of attempts - and some were much more, you know, fractionary than others. But yes, there were so many people that were quite competent in that. And, again, is it luck or is it just the right intuition? The one thing that we did different pretty much from the start is that we not only synthesized the logic, but we also added timing to it. And the timing was relatively simple by number of layers of logic. So if you have a lot of stages, itís slower than if you have just one, sort of obvious stuff. But nonetheless, it had sort of this ability to optimize for timing simultaneously as it optimized for area. And area was really fewer gates is better, right. And of course did this on a computer and was very fast compared to what a human can do. So now these two stories converge.
RB:†††† Of course, gate counts were going through the roof.
AdG:† Yes, as you know well because you were at LSI Logic doing exactly that.
RB:†††† And so it became more and more important to have your tool because doing it manually became unwieldy and impossible.
AdG:† Except people didnít know it yet, how important it was going to be. Right. And so this is where the next accidental thing happens, accidental at least from my perspective, which is suddenly GE is sort of hinting that, you know, things are going to an end in semiconductors. And through a set of other accidental circumstances, I had the opportunity to then end up making a proposal to GE Ė if youíre going to lay us off anyway, why donít you let us go with the technology, maybe some investment, and we will try to set up a company to commercialize this and with a return to GE. And it was a very important decision that I decided, and this was with two people in my team, to approach GE because for weeks I was torn, you know, was it even legit to propose a business plan. This is technology thatís owned by GE, an outstanding company that treated us very, very well. And Iím happy to have done that because I think that is the way to start something. Now you donít walk out with somebody elseís technology and start something. It is their technology. And the happy circumstances is that I had been involved in a task force to help GE figure out what to do with an acquisition that they had made in all that which was the acquisition of Calma. And, you know, Jack Welch, the CEO of GE had done many things unbelievably well. You know, this is a guy who walked on water, whatever he touched was gold, except the acquisition of Calma, which, of course, was a West Coast EDA company that had seen its prime and was heading down, acquired by some big company that really didnít have the wherewithal to manage it. And so there was task force after task force how to say this because every Fortune magazine that said, you know, Jack Welch walks on water said yeah, but Calma really doesnít work. [laughter] And so you can just see him wanting to fix this. Long story short, I got a one-hour opportunity to present to the vice chairman of GE, you know, some guy 67 levels above where I am, which is a first level manager, to propose this plan and, you know, with some caveats that turned out to be actually pretty big. But he said fundamentally yes. And so that is how Synopsys started ... by accident, circumstances understood much later as to why it happened but all on the premise of cool stuff, can we do something with it?
RB:†††† But itís good on-the-job-training for you to be instant president and raising money. GE I guess put a million in.
RB:†††† And then you had to go out and get some more.
AdG:† And the next million was Harris Semiconductor who had done something like that before and who -
RB:†††† Became a customer.
AdG:† And became a customer even before we formally closed as a company. And ultimately we brought in some venture capitalists. And I learned some interesting lesson there which is that when forming something, you know, we were all about what it could do, the upside and so on. Contrary to what people say, venture capitalists are actually unbelievably skilled at something else which is risk management. The risk management element in this was that the GE and Harris commitment for the VCs was that technology must do something right, check, right, whereas for Harris and GE, the VCs putting in money was well they must see some opportunity. Check. And so essentially it was a balance of who was assessing what risk. And it took a little while to get that together. Took actually about nine months, part of it being cold-calling, hi, Iím so and so, you know, million. But it was learning that still to this day, risk management in big techonomic endeavors are absolutely very important to manage well.
RB:†††† So with things unfolding favorably in Carolina, what caused you to move to California for the next step here? So we know youíve got a couple million bucks from GE and Harris. What caused you to land in Silicon Valley?
AdG:† Almost. We didnít have the million bucks yet. We had the promise of a couple million bucks. We still needed to get the VC guys. And so that first phase of funding, the first nine months, we got a 250k loan and I still shutter today on the word ďloanĒ because thatís a thing youíre supposed to pay back, right. My net value Ė what was, you know, a used car. And secondly, we got a little bit of layoff benefits and thatís really what funded that first stage. But still, we felt we had the money virtually in our pocket, right. And so what became very quickly clear is that the center of gravity of Ė for starters, academia. We had some key professors that were engaged, that believed in what we were doing at Berkeley. And Berkeley was the center of gravity of EDA from a certain critical mass point of view. There were some other very good universities that did excellent work. It just Berkeley was the center of gravity. Secondly, the VCs they were all here. And so talking to them very quickly, they were not planning to travel a lot to California Ė to North Carolina, thatís for sure. Then the recruiting opportunity of people that knew technology and so on were here and then customers were here. And so that was actually a, in hindsight, obvious decision in the moment nontrivial, moving a group of seven people to the other side of the country of completely unknown. But we did. And so by September of that first year, which is now í87, we have settled here and weíre starting to recruit and sell to customers.
RB:†††† So thatís a major milestone obviously and, as you say, little bit more luck than judgment sometimes. So now youíre here in Silicon Valley with money in your pocket. Youíve crystalized the product idea. It happens to resonate with you personally. You love the technology. And then, interestingly enough, you hire an external CEO rather than yourself being CEO. Most founders classically want to be the CEO whether theyíve got experience or not. So you took a different tact and hired our friend Harvey.
AdG:† Harvey Jones, yeah.
RB:†††† Jones from the industry, from Daisy at the time. And that was a pretty big decision for you, was it not, to bring in a CEO when, in fact, you are the founder of the company?
AdG:† Yes and no. Yes because itís a big change and it is very material who you pick because, you know, the chemistry, the complementarity and the alignment are all very, very important. No, because I never felt that I came with a ticket with all the hole punched or whatever the terminology is of experience. And hopefully still today, Iíve always been in the mode of how much can you learn with another people because, at the end of the day, the success of whatever we do is what matters more than the accomplishment of the individual even if we realize that weíre in fields where the accomplishments of individuals is often the make or break of a company. And, by the way, I always represented that in such a fashion to the venture capitalists that bringing in talents in whatever role was most likely to make us successful was what I was willing to do. As I did bring in Harvey, I think another one of those being unbelievably lucky because he had left Daily already a bit before. He was sort of hanging around venture capitalist circles looking at opportunities and he had discovered synthesis on his own actually with another company, looking at other companies. And then here was this team that, you know, was actually hopefully pretty much on the ball and had something. But more importantly, he brought something that was maybe intuitively clear but I had no notion of, which is marketing and sales. And that is really where his background and pedigree was most strongly from, whereas I clearly came as an engineering contributor and manager and the combination of those two things was very, very important in terms of balancing what the company needed.
RB:†††† It was almost the perfect marriage.
AdG:† It was almost the perfect marriage and more than that is, you know, over time, one starts to adopt some of the learnings and trades of the other person. And so thereís an enormous debt of gratitude that I owe Harvey just from what I learned. You know, and just the hour before we talked, I talked to some people Ė all kinds of marketing concepts and some of that, you know, I can sort of still hear Harveyís voice in that.
RB:†††† Yeah. Now that was really instinct and the fact that Daisy Mentor and Valid or DMV as we used to call them, already existed at that time. So you walked into an EDA space where there were three, you know, brand names. And you were lucky enough to get the CEO of one of them and hit the ground running. So that was a great start. So then I think at that...
AdG:† Well then comes customers, right?
RB:†††† Right. So around that time, there started to be a paradigm shift. So, you know, gate counts were rising rapidly by í86. People wanted to design chips that werenít just 300 gates or 600 gates.
AdG:† Iím talking five micron, go to three micron.
RB:†††† Exactly, yeah. And we were talking about ten thousand gates on a chip, wow, and even twenty, wow, not designable by hand in a sensible timeframe, of course. And so the concept of an RTL rather than schematic capture started to take hold and the market started to recognize that. And you guys found yourselves exactly in that point to -
AdG:† Thatís the right time.
RB:†††† So the timing Ė perfect timing if you will.
AdG:† Yeah, yeah, yeah. Well, you know, and so as we approached customers, initially, the first hurdle was Ė the fundamental question is, is whatís coming out good enough? And good enough was measured truly in two dimensions which is the area Ė but really think of it as gate count and then speed. And I still remember going to a number of customers and essentially taking the circuit that, you know, some engineer had slaved on for months and taking the circuit, reading it in and optimizing it. And in a matter of an hour, it came out as, you know, a third smaller and a third faster. At which point in time, they looked at it and said thereís no way. Itís just not possible. So they would go away and for two weeks, they would verify that there were no errors. And then they would come back and say this is amazing. This is a fabulous tool. This is magic. And, of course, now you have the other problem which is the expectations are way higher than what it can do because the initial impact is so big. Well you gradually work through this but suddenly you realize you have the nucleus of a technology that can do things. And this is particularly relevant because then a new angle appeared which is the angle of mapping from a high level description or an existing circuit, not just better into a technology, but being able to map into multiple technologies. And this is, in some ways, where your and our Ė your story and mine intersects which is I think, at that time, you were at LSI Logic. And, of course, LSI Logic was probably the king of the hill of gate arrays. There was Toshiba I recall was also there.
RB:†††† We licensed it to Toshiba.
AdG:† Oh you did? Okay.
RB:†††† Yes, yes.
AdG:† And so there was only an interest of well what if you could map into the technology, the library of gates of choice, which was the LSI Logic library? And so that is the only library we actually did by hand, taking your technology and then describing it so that the synthesizer could read the library and optimize for it. So if you had, letís say really great NAND gates but lousy NOR gates, it would know oh that one is big, let me optimize more toward the NAND gates than the NOR gates, right, for argumentís sake. And the reason this is so relevant is because this is a seminal capability that ultimately make the fabless semiconductor model possible, which is people do design, maybe RTL down two gates and then something maps it independently of the fab. And not that there were that many fabs but, you know, even within LSI, there were multiple types of libraries, right. And then we created a thing called the ďlibrary compilerĒ so that it would become easier to do in two different technologies. And there were lot of people that had their own manufacturing. And so essentially it became the intersection point that ultimately brought about names that had started before us like TSMC, which today obviously is the largest foundry in the world. And it is predicated on this notion of can you make things technology independent because if they are too dependent, everything becomes just too complicated. Right. If you have to solve all problems at the same time, you never can solve problems. And this notion of layering was key.
RB:†††† I think youíre right. I think the gate array, in many ways, once they forced that it wasnít a decision. It enabled people to see through the mist, which ultimately led to the foundry existing because, I mean, you know, we sat there at LSI and we did these designs and we sent a computer tape somewhere and back came a chip.
AdG:† Right and you have a fab in Oregon or so, right?
RB:†††† Yeah, but years later. I mean, LSI was fabless for many, many years. Finally Wilf went in the direction of, like Jerry Sanders said; real men have fabs, and then ultimately back again. LSI ended up being fabless, of course. So I wonder if Jack Welch realized he missed that opportunity.
AdG:† No, you know, the person I had to give the pitch to was Ed Hood who was the vice chairman. So, you know, sort of the number two in command. And many years later when we went public, I wrote him a little note and he wrote very nicely back.
RB:†††† Whatíd he say?
AdG:† That, you know, their return on that million was twenty-six million. So I think it, you know, we paid our dues and happy for every piece of that money that went there because it was a great company that made it possible.
RB:†††† Great story. So now you have essentially ushered in, let me call it EDA 2.2, which is the beginning of the real automation using the computer to design rather than just capture what engineers are sitting around doing manually.
AdG:† Yeah, in all fairness, in parallel to that, another automation thatís happening which is place and route, which is the layout automation and you can just see how these things are going to start working hand in hand.
RB:†††† Exactly, exactly, exactly. So you founded in í86 and then almost perfectly in í92, time to go public. And you went public as I can see on the wall here, quite successfully.
AdG:† I guess weíre still public so it must have worked.
RB:†††† Still public. Doing very well. Now youíre two billion in revenue plus. So a big day for you and Harvey. And, I mean, an IPO six years after founding a company for essentially an engineer with no Ė very little commercial experience must have been very interesting for you.
AdG:† Yeah, but even there, you know, itís a little bit youíre so much in the run of the race of doing things that I donít recall that there was much time to reflect. It was all looking forward. Now, you know, the going public process was all new, certainly to me, not to Harvey who had done that with Daisy. And I remember 54 pitches in two weeks. I remember something that few people experience which is yes, it is possible to give a presentation and fall asleep while speaking and itís a weird feeling, I tell you, because your head goes like this and you have no idea how long it was down. But, you know, all fun and games, it was just, you know, a very rapid moving thing. And then overnight we were public. And with that came all the learnings of what itís like to be a public company and how you speak to that or not speak to that and so on.
RB:†††† So with a successful IP under your belt, the two of you, and the company well positioned product wise in the market and so on, ignoring the ups and downs, you look like you were set for the strong long term. Itís very interesting that you embarked on a very aggressive over two decades' acquisition strategy. So my quick note suggests that youíve acquired forty companies in twenty years.
AdG:† Itís more like eighty.
RB:†††† More like eighty. Well thatís the forty I saw.
AdG:† Well some were very small.
RB:†††† - which is very aggressive. And so that can be multiple reasons - to grab the revenue, to keep other people off the market because the technology was complementary. Just tell me why you embarked on such an aggressive acquisition strategy.
AdG:† Well thereís two lines of thinking that sort of merge. The first thing is when youíre in technology, doing anything different is always risky to some degree. And so you have to make fundamental changes Ė sorry, fundamental choices Ė do you invest in your own development and it has a certain timeline and you have a high degree of control versus do you look at other peopleís development and see if acquisitions can bring them in where you get the negative of it being much more expensive to acquire than develop your own. You get the positive that youíre not going to acquire the losers. Youíre going to acquire the ones that have something either technically or economically. And so early on, we realized that a sound balance between strong development efforts and continually looking at other developments that could be brought into the fold was actually a form of portfolio risk management but also of time management because if you buy something that exists, you get it the next day whereas if you say Iím going to invest, then you have a timeline. That may be okay or may not be okay. The second notion was this notion of how do you exploit your capabilities while minimizing risk. And I think the key minimization to risk is the notion of adjacency. What I mean with that is if something is technically adjacent, that reduces the risk because that means you probably understand it, you may be able to leverage things and so on. If itís channel adjacent, that means you can use the same people to bring it to market Ė your sales people, your support people, your global infrastructure. That reduces the risk. If itís customer adjacent, that means they can even go to the same customer they know already. And as we all know, acquiring new customers is actually difficult whereas if you can go to people where hopefully youíre trusted, you have at least an open door. And so many of the acquisitions fit all three criteria. And this was very much on purpose but that also determined how far can you reach. I mean, if I look at the acquisitions, there are few that in hindsight I wouldnít do because they were what I would call one bridge too far meaning that yes it was the right vision except that there was a piece in between missing. And so you could not leverage it until you had it. For example, we invested in a library company who did Ė a physical library company but we didnít have a place in route system. I donít know who thought about that. Iím afraid it was me. And so you learn from some of that but fundamentally the adjacency concept is very key. Now we were very heavy in R&D ourselves. So, you know, still today we invest over thirty percent of our revenue in R&D because, you know, weíre following a mistress called Mooreís Law thatís, you know, a tough broad to follow, right. And so we followed it really with both the acquisitions and the R&D. Now the second thing to say is that EDA itself got stratified in terms of what became ultimately sort of the logic or the functional level and there we broadened toward simulation and other things and then underneath the physical level where we were not. And by the end of the 1990s, I was very worried that we did not have a presence in the physical (?) at a moment where these things were going to stop being independent. Remember earlier we said it was great that you could do the logic design and when youíre done, check, you know, sign off for that. Then you go to physical. And there was not much back and forth in that. Later as things became much tighter, the physics started to impact the logic and vice versa. And economically, more and more customers said hey, you know, we want to have more integrated solutions. And we were worried that we were not in a situation where we could provide both. And so either if thereís a downturn, you can be economically out or technically you canít satisfy the customer. And that was a bigger jump to say okay, now how do we get into the physical domain. And we did a number of acquisitions there ultimately Avanti was the largest one.
RB:†††† Yeah. I think that was a very good move. I mean, at LSI we sort of were already there because we had simulation and we had physical. So we -
AdG:† Because you also had manufacturers.
RB:†††† Exactly. We were an early vertically integrated guy. And so -
AdG:† And you had a very talented guy by the name Jim Koford who, you know, doesnít get enough credit for having been really one of the people that drove an imminently practical working system for many years.
RB:†††† Yeah. No Jim indeed was the crown jewel of that LSI strategy and so it didnít surprise me and us, I think, that, you know, Synopsys ended up being king of the front end and then having to have the back end and finally offering the customer, you know, the whole vertical which, of course, your competitors ended up doing as well. So, you know, Cadence has the front end and back end and so on. And ultimately Mentor as well. So yeah, so but thatís a very aggressive strategy. I would saying thinking back in time, everybody Ė most companies acquired somebody, a few here and there. But the number of companies youíve acquired, considering you started as a fairly small company. I mean, weíre not talking General Electric here. Weíre talking about Synopsys Ė was a very aggressive strategy and itís obviously worked well for you in 95 percent of the acquisitions you did which, of course, is great. So I noticed also and I find it interesting that you added a co CEO and my thinking was, correct me if Iím wrong, that with all of this going on, you just needed someone who was, you know, helping you on that level that could make decisions when you were traveling or something. And so a co CEO structure which often doesnít work for some companies, worked for you perfectly obviously between the two of you. But because you were acquiring at such a rate, it was probably beneficial.
AdG:† Well, you know, this started actually earlier than the numbers would indicate. You know, Chi-Foon Chan, thatís the person youíre talking about, you know, joined Synopsys 26 years ago. And so he was in from the beginning and he came in originally to build the support for the Arm company. Now realize that theyíre sort of selling this Ė sales and marketing, thereís the R&D. But the third leg of that stool is actually the support team that interacts with the customer. And the unbelievable accomplishment that he had is he built arguably the best support team in our industry. And, you know, survey after survey have reflected that over twenty years. And I think it is as much a differentiation for Synopsys as the technology has been. And, of course, the reality is the two multiply. Right. You have great tools but nobody can use them, you still have nothing. You have super user friendly stuff that gives you horrible results, you have nothing. And so the two are Ė itís not an addition, itís a multiplication. And the other thing is that Chi-Foon has a personality and a business sense that is fabulous. And I recognized that early on. I recognized it especially in light of feeling that I wasnít anywhere close to his abilities. And so very early on, I started to work closer and closer with him and frankly Iím not sure exactly when it happened, late nineties, so itís that long ago, he became COO of the company and we immediately worked two in the box for many years. And, at some point in time, it became sort of obvious well, you know, he or I, we equivalently run the company. We have been a team for a long time and thereís always the question well, what happens if you disagree? Well the first reaction is I think you know something I donít know because, you know, heís super smart and, in many cases, has a better judgment than I have. So then we formalized that in the co CEO thing. That raised a lot of question marks. You know, how does that work? And so but the fact is for us it was not really a change. And it allowed him to speak for the company on completely equivalent terms to what I could do and especially at the moment where we then have a globalization. Yeah, thatís the other thing that is really central to this story, of course, is starting late nineties, the degree of globalization in the world, period, in high tech specifically and in our company is enormous. Right. And so thatís only means that you have to be at so many places in the world. You know, youíre going to die if you actually try. And we all came close to that in some form or another. And so it allowed him to speak at wherever he needed to go. And I always like to think of Ė itís a little bit, you know, in Silicon Valley, itís the ying and yang him coming from the Far East and me coming from Europe - and hereís Silicon Valley, you know, the melting pot that made it happen. And so, you know, in very simple terms, without Chi-Foon, you know, Synopsys wouldnít be here in [talking over each other].
RB:†††† No, itís been a great strength obviously and itís nice to see two guys with the chemistry thatís that close, two in the box that really works rather than the board dictated it kind of thing.
AdG:† Oh yeah. No, no, the Ė we had to convince the board that it was a good idea.
RB:†††† Which sort of leads me to my next point talking about the explosion of the industry, you know, the arrival of the internet. And now, of course, I mean, I saw this back at LSI. We were moving files around at kilobits per second and, you know, and as the -
AdG:† Whole kilobits.
RB:†††† And as we get up to 50 and 100,000 gates, these files get really big. And so the arrival finally of a meaningful, useful broadband internet changes the way that, you know, you have to travel, where you put design centers and so on and so forth. So the internet had a big business on Ė sorry, big impact on EDA. How would you say it impacted Synopsys the most?
AdG:† Well, you know, really the internet first had a big impact on our customers more than on us as a solution. Of course, internet became the, you know, the mechanism for massive communication every which way. Thatís a given. But really, in my opinion, what it did, it was sort of the second killer app to the first phase of semiconductors. Let me take a second for that. You know, from my perspective, semiconductors obviously have been around since the invention of the transistor. But really what put it on steroids was computation. And computation started with the calculator but the killer app was the PC because the PC was this foundational platform that would make other people productive and creative, meaning software guys. And you could have multiple different software guys on the same machine do stuff and that stuff really, you know, grew productivity in fifteen years between the early eighties and mid-nineties, unbelievable for the world. The killer app was the PC and it had one key thing which is whatís better than a PC? A better PC. Well how do you do a better PC? Well you put more stuff in the chips. You make them faster, lower power and the price sort of stays the same. Go. And then the second multiplier to that was well how do we connect them? And thatís, of course, the internet became not only the physical mechanism but also the (inaudible) on how you do that. And that, of course, unleashed what today would be called ďthe cloudĒ. And it brought about, you know, distributed compute centers, many private, some openly available, many things. It also had its own downturn. And, you know, the dot com boom still remembered for the ultimate semiconductor party in í99 and 2000 was the dot com bust in í01 and five years later, semiconductors came back. Right. During that five years, the beginning of the second phase of semiconductors occurred. And the second phase was stupid phone becomes smart phone. And ultimately this was what was referred as the convergence of computation and communication but really today itís called the mobility phase. And what mobility did, it, for starters, adopted the same fundamentals as the PC which is whatís better than a smart phone? A smarter phone, of course. What do you need to do that? Put more in the chips. But the smart phone had one more challenge which is low power. It really drove the notion of low power as a dramatic optimization criterion. And so, of course, massive growth. It changed the world again now making it possible through the phone, but also the tablets, and so on, to get access via the internet of information, of communication virtually everywhere. It changed. Itís changed the social economics of the world. Right. I mean, this is unbelievable impact. And this has been the second massive both technical driver and economic growth for the semiconductor industry. Now, at some point in time, these things gradually mature. The price points start to be softer. And now the question so whatís next? And I think where weíre sitting right now this minute is that if you ask people oh IOT, Internet of Things. Weíre now going to connect not just to people but to everything. You know, measure, have sensors, get enormous amount of data which absolutely will make big data analysis unbelievably interesting. The challenge with IOT if you formulate it just as sensors plus is, you know, those chips are this big. Takes a lot of them before you fill a wafer. And so I think that the next phase is enabled by IOT but is not IOT. I think the next phase is smart everything, which is the minute you start to attach some smarts to this local data now smart by definition is the next version of software or hardware. And whatís better than smart? Smarter, of course. It has, again, that driving characteristics that if you find the killer app, making it better is a natural driver for demanding more from Mooreís Law. Now I think weíre right now preparing for this. And the evidence thatís interesting is that here is a relatively stodgy industry, the automotive industry, that overnight turns into the poster child for digital intelligence. We see them driving around in the neighborhood on their own and the guy sitting at the steering wheel sort of bored because it actually works. That is unbelievable. But it is also a perfect example of how technology suddenly becomes enter the space of could be good enough, itís almost good enough, itís good enough. Itís great now. And I think that is where weíre heading. So these key phases really drive what I like to call the techonomics of what we have left because the technology is always governed by the economics and the economics are always governed by the technology. And, in my opinion, Mooreís Law is just another word for exponential change. And I know thereís technical exponential change and there is economic exponential change. And itís sort of an intersection and at times when people ask is Mooreís Law not dead, theyíre never quite sure do they mean technology or economics? Of course, itís both. But the fact is weíve been riding an unbelievable exponential of change. And if you look at the impact of computation, we clearly see that the impact of mobility we see that. The impact of smart everything will be just as big again as what weíve seen in the last fifteen years. Thatís how I put this picture together.
RB:†††† While weíre on the Mooreís Law thing which is obviously an interesting topic wherever we are on the time curve, so with the world today looking at, you know, sub ten nanometer geometries, that obviously reflects heavily on you guys as well as the fab.
RB:†††† So, I mean, thereís the optics over here and thereís the physics of, you know, the libraries over here. Youíre obviously working sub ten at this point. How far do you think thatís going to go out? I mean, is there, in fact, a physics limit here or are we smart enough to† just keep pushing it out for another decade at least?
AdG:† You know, in 1978 at the undergrad school I was at in Lausanne, Switzerland, there was a conference of the semiconductor leaders of the world. And as a little student, I was at the conference watching with eyes this big because these guys had written papers. So they were gods right. And the gods all agreed on two things Ė electronics was going to be a very big deal, and it would not stop until the ultimate physical limit which was one micron but then it would be over. And Iíve never forgotten that because many, many years later, I had the opportunity to give an award to Bob Dutton who was one of the leaders of that crew. And, of course, I loved the moment to remind him of his prediction of one micron. But then told him I respect this group so much because the very people that predicted this impossibility were the engineers that then navigated around it. And so since then, of course, weíve had many phases of this question of is Mooreís Law dead. And, if nothing else, the caution I would give is never underestimate the degree of innovation because just in the last ten years, weíve seen this FinFET will never happen. Now a transistor flat already so small itís almost impossible to manufacture them. Itís almost impossible to design with them. And then Ė and FinFET for sure will never work. You know, imagine these things vertical. These are impossible to build. You cannot make this yield and so on. And here we are sixty nanometer, forty nanometer, ten nanometer, and some people moving to seven nanometer FinFET as we speak. Right. So our design system is used for these designs today. The IP is becoming available for the most advanced nodes. And so this is moving. Now does it mean the physics gets easier? No, they change. Obviously they change. By the time youíre looking at thickness of things in terms of the numbers of atoms, you know, thatís not called a material continuum. Secondly, the number of new materials is growing substantially because weíre pushing the edge on everything. But there are people right now looking at yet another evolution of FinFET, nanotubes, or tubal type forms of transistors. And the whole question is can one find ways so that the manufacturing doesnít go out of sight in complexity, but rather becomes more self-aligning, for example. So is this possible? I donít know. We will see. But donít underestimate it. Now thereís another reason to not underestimate it because if you buy a division of Ė and here comes smart everything. Whatís smart everything going to do? Itís going to put pressure of hey, can you just give me 5x more, 10x more of performance over power. If you could do that, my thing would be so smart. Itís well worth the effort. And so this is where with the opportunity push ultimately back on the technology will determine the level of investment that will get made. And certainly the people that understand this I think are watching very carefully of what could happen here because, frankly, does the cost of the chip really matter that much? Of course it matters. And, of course, we have been accustomed so, I mean, the early ranges of Mooreís Law are bringing the cost for transistor down massively. But if you could have a device that you can hook to your body that can tell you at any point in time here are twenty things that happen in your body and I can compute if youíre in danger or not. And somebody says itís not a hundred bucks. Itís two hundred bucks. Would you hesitate? Of course not. Iím not saying itís there. But the value of the opportunity ultimately determines the investment in the technology. In the past, the technology has been enabling the value. I think that balance with techonomically balance itself.
RB:†††† Well, I mean, today the smart phone, I mean, at six hundred dollars is not a trivial amount of money today to a lot of people.
RB:†††† But everybodyís got one including in some very poor countries. And I think in many case, the business model has adapted for the high cash price of the item.
AdG:† Which really is saying that value change is an appropriate word right, because the value is harvested at the end of the chain. And, you know, we sometimes feel that weíre in the boiler room shoveling coal in the ship and, you know, the apps guys are drinking martinis on the upper deck. But the reality is that the value chain that finds its own techonomic equilibrium and while thereís a lot one can do with advanced software for digital intelligence, it does rely on processors that will be, in my opinion, optimized for these type of opportunities.
RB:†††† Thereís another example of that too out in the EDA industry that you well know, and that is that, you know, the industry went from this big check up front license fee with, you know, a trickle of maintenance to more of a SaaS model with, you know, renting a car rather than buying a car. And, you know, seventy percent of the cars sold in the US these days are leased not bought. So that was a big transition for the EDA industry but you went through that very quickly and surprisingly well.
AdG:† Actually we pioneered it. We were the company that drove this. And it was a very difficult decision because, you know, we faced, like any company that becomes of some size, the never-ending oh itís the end of the quarter and, you know, if you donít get the deal, youíre going to miss Wall Street expectations. And when that happens, you know, really bad things happen. And, of course, you know, the customer is completely trained on that as well. And so, you know, at 5:00 on Friday, you know, the purchasing agent says oh I forgot, I have to pick up my kids for soccer, which point then the sales guy gets completely nervous and off you go. But there was another reason behind this which is the notion that we had long moved from transactions to really a formal relationship management in both directions meaning a customer could not buy something for short period of time because they had a long, very complex project and needed the support that goes with that. And we needed the opposite which is the ongoing feedback and sometimes in very direct negative form of, you know, this is what you have to do to win. In other words, the push to stay on Mooreís Law. And so why not do the transition? Well the problem is when you have revenue thatís this high and you go to ratable, itís going to be great here but you start by going this far. And so we actually did it in two steps, mostly so that we would not completely tank the revenue and therefore tank the valuation and therefore be in danger of being acquired in an undesirable fashion. And then we moved forward to that and now quite a number of large software companies over the years have followed that model. And it brought a great stability and during the í08, í09 downturn which was very tough on the industry, we sailed through essentially flat.
RB:†††† Yeah. No, that was a great move for you and the industry has followed. †So thatís great. So just a couple things on the, you know, for those of us that have been in the semiconductor industry for a long while, as you have also, thereís been what I refer to as two cold wars that have been going on. One is, of course, real men have fabs. And that war is over. Thatís no longer true so now foundries are legitimate and control the business.
AdG:† That was never our war and a little too much for modern PC thinking.
RB:†††† From the EDA point of view, I lived through the era of you got to have your in-house design tools versus third party tools. And so for the guy with a fab, the guy with the fab felt like he had to have his own tools. If the customer started to use Synopsys tools, then we lost control of the customer, which ultimately, of course, was true. And so you won that war, meaning the EDA industry won that vertical, meaning that independent tools were the right way to go ultimately for the customer. Now everybodyís comfortable with that. So today weíve got, you know, semiconductor companies that use third party foundry fabs and third party tools. And thatís the norm. So thatís been two very large transitions over the last twenty years.
AdG:† Yeah. And so, you know, the first one, the real men own fabs and so on really became a business transition towards companies really managing a P&L versus companies managing capital expenditures because manufacturing obviously is a very, very big investment. And then critical mass matters because with every generation of new silicon, the cost to play went up rather rapidly. And if you were not large enough, you couldnít fill the fab and a fab thatís not full is a disaster now from a P&L point of view. Right. Heavy on capital. And so this is really what was one of the change factors towards the fabless industry. The change from internal tools to external tools was long in coming because all these companies internally Ė initially had internal tools. And there was at least some reason to stay attached to that as it pertains to the understanding of the physics of the manufacturing. And thatís actually where the longest, you know, the place and route was one of the things that stayed longest within companies.
RB:†††† And there was some, I mean, you know, at LSI, we had some very, very competent EDA guys as you mentioned. Kofordís crew was a very, very competent crew but ultimately there was sort of a little bit of a job security issue if the tools moved out of house.
AdG:† It was a very big issue, the de facto for understandable reasons. And one of the interesting side effects of that was that the companies that were best at it stayed longest on it and by the time they stayed longest on it, they missed Ė they became much less competitive because the commercial solutions had the leverage of a broad customer base and also much broader learning and so moved much faster. Now this is actually quite long behind us and now thereís still some customers that have some tools, mostly very strongly related to some specifics of their fabrication that they want to maintain, you know, proprietary for example.
RB:†††† Yeah, itíll always be a long tail but, I mean, but the war is over I think is -
AdG:† But, you know, thereís a third one in that same category which is and we started this already in the nineties, which is the notion of a reusable IP. And initially it started with simple things. Oh, can we just do the libraries add some adders and multipliers and then gradually these became more significant IP blocks. And today about, you know, a bit over twenty percent of our business is IP blocks. And I still remember going to some customers early nineties saying hey, you know, we can provide some of those IP blocks and some went like, you know, this. Thatís our business and, you know, we Ė for the same reasons except that, over time, it became clear that if you have really good engineers, you have to use them for things that are differentiated. Many engineers think that what is difficult is differentiated. Now whatís difficult may be differentiated. Whatís differentiated is whatís truly different, right, and adds value. And so as that happens, the notion of letís say doing a USB core which gradually became more and more difficult to do is difficult but itís not differentiated because you want to have the same as the other guys. And so those are good examples of IP blocks that became massively reused and a commercial market grew and ARM, of course, became the one that provided a process, of course, to the mobility wave and we are the second largest providing, you know, all the interface blocks and many of the other building blocks that configure around that.
RB:†††† So the EDA industry, Aart, now is in what I perceive as a nice, stable state, meaning thereís three big players. The three of you tend to acquire anybody thatís pretty good before they get too big. You know, whether thereíll be more consolidation in EDA, you know, beyond the three, you know, remains open. We wonít debate that. But of the EDA leaders in the US, I would suggest that, you know, with your thirty plus years at Synopsys you are the EDA leader at this point. I mean, Wallyís got twenty years into Mentor right now but heís got twenty years behind that at TI. And Cadence has had, you know, several changes. So youíre really the anchorman of the EDA industry and, as you say, you know, you lucked out on a few things. But I think you guys have done a fantastic job over the years strategically.
AdG:† As the saying goes, the harder you work, the luckier you get.
RB:†††† Yeah. So itís interesting. And so itís not surprising that youíve got a lot of awards that have accrued to you over time and rightly so. Youíre a fellow of the IEEE. Youíve got the Phil Kaufman Award which is, of course, often thought of as the Nobel Prize of the EDA industry.
AdG:† Yeah, I donít think that the Nobel Prize, you know, refers to itself as weíre the Kaufman Award of, you know, physics or so.
RB:†††† But thatís, you know, a fantastic award to have. And then a number of CEO of the Year, Entrepreneur of the Year and so on. So youíre clearly recognized as having gone from a brilliant engineer to first class CEO. And, you know, my congratulations on that one.
AdG:† Thank you.
RB:†††† So when youíre not at Synopsys, what do you do? Do you do anything else? I understand youíre a guitarist.
AdG:† Yes, I am actually.
RB:†††† I have a dozen guitars so Iím with you on that.
AdG:† Oh you do? Oh wow, you know, [talking over each other] this interview is just now going to take three more hours if weíre talking about that.
RB:†††† Mostly Spanish. I love the guitar.
AdG:† Oh you mean the Spanish guitar?
RB:†††† I made one when I was a kid.
AdG:† Oh really?
RB:†††† Thatís what got me into it so and I still have it.
AdG:† And lucky you because thatís got to be unbelievably precious to you.
RB:†††† Yeah, itís a fifty year old guitar which is interesting. So you -
AdG:† Yeah. I do play, you know, electric guitar in a blues band. And actually coincidentally, the weekend before last, we spent five days in a professional studio doing a recording of a CD. And I can tell you, you know, that is both exhilarating and as humbling and stressful as anything because once you know the recorderís on and everybody is watching, you know, itís like, you know, skiing on this unbelievably large slope and thereís one tree and you head straight for that tree because you become conscious of what you do. Right. But itís a very cool hobby and, you know, thereís so many great, great players that itís very humbling to see how far one can still go.
RB:†††† One of our mutual friends, of course, Brian Connors is into the same thing.
AdG:† Of course, yes, of course.
RB:†††† Heís heavily into that. So the question then leads to, Aart, whatís next for the semiconductor EDA industry? What do you see thatís Ė ten years out, are we going to see something thatís quite different or are we going to see an extension of Mooreís Law, a bigger and better Synopsys, a bigger and better Intel? Is there another big change coming maybe?
AdG:† Well, you know, earlier you sort of said well, you know, youíre in a very stable industry. I think the word ďstableĒ has never applied. I think when youíre riding an exponential and I think we are still riding an exponential; itís actually really a combination of many different exponentials all interacting. You know, an exponential is a tough cookie to follow, right, because, you know, the same level of difficulty of yesterday is tomorrow again. You know, the slope is always massively up to the right. And so just continuing that in its own right is a fulltime job for all the companies that are aspiring to this. What weíre now seeing is a number of discontinuities we talked about, you know, some of the silicon technology challenges. We didnít talk much about the intersection of hardware/software because I think that is really where a lot of the emphasis is going. You know, semiconductor companies today, a vast majority of the engineers are software engineers. And that means that theyíre embedding a lot of software that has to be highly tuned and toned before the silicon technology. And we have invested massively now in both the verification of the software and hardware simultaneously and we even have a business thatís aimed at trying to see if we can bring the notion of sign-off, which weíre well familiar with in semiconductor land, to software for both quality and also security because thatís the other new thing that has entered this picture for all these next waves of products, right. IOT is great but every IOT device is the equivalent of putting, you know, a kitchen window into a bank. This is bringing up some connectivity issues. And so I think the number of challenges ahead of us is as big as it always was. But the brain power thatís heading there is also very, very potent. And so I think no company for starters, not Synopsys, can take a single day as given looking at tomorrow. And so, you know, I rarely actually do what weíre doing right now which is sort of looking back a little bit at what happened because the focus is always on tomorrow and the next year. Now the other thing is I think in ten years, there will be another Bob Blair who says, so what do you think? You know, Mooreís Law, is it dying or [laughter] seen that movie before. And maybe itís not the same Mooreís Law as we grew up with, so to speak, but the notion of an exponential, super high impact change in ten years will actually be, in my opinion, an unbelievable full swing because suddenly many, many fields that have just started to use computation to change how they look at things will arrive at a point where they can model things that they couldnít model before. Human body for starters. Weíve lived, in the last fifteen years, the impact of DNA cracking. And having met both of the people that cracked the human DNA, you could see how much effort that took and how the price tag was in the billions moving to the millions, moving to the thousands, moving now to the hundreds of dollars to get a full DNA analysis of a human. Once that is possible, the next step is, of course, how do you use this data to understand what the fundamentals of being a human is. Already today, it is possible to have implants in somebodyís brain, read the brainwaves, interpret them, and control a mechanical arm to help a paraplegic person have some control.
RB:†††† Look at our professor at Cambridge, Dr. Ė whatís his name Ė Hawking, an example of, you know, he should have been dead years ago. But, you know, heís still active lecturer today even with that huge impediment.
AdG:† Yes, and he is not yet in the category of bionic person, which is an intersection between a human and a machine. And there are many people that are already in that category today. And, you know, artificial eyes, hearing and so on, the intersection between electronics and the brain is absolutely work in fast progress. But the understanding of what being a human is is also in super-fast progress. The understanding of what the world is, you know, take the notion of global warming and weather patterns. You know, weíre going to move pretty fast from conjecture to so-so models to good models, to unbelievably good models because the computer power and IOT, by the way, will change that landscape again. The notion of going to stores and people understanding already who you are, what youíre interested in, what youíre looking at and how you feel about it, you know, these are all interpretative skills that with some computing, facial recognition, feeling recognition, possible today Ė weíll all enter every day. Not all of that is good or bad. Thereís all kinds of questions and judgments there. But the impact in the next ten years of this wave of computation and big data and this wave of digital intelligence Ė and I use digital intelligence as both being less than artificial intelligence because it cannot mimic today anywhere close to human and being more than artificial intelligence because cars can actually see infrared. We canít. So already theyíve got something over us. And so that is a space that is proactively week by week now advancing. And, you know, that is not dissimilar to week by week when mobility came in, it changed the world.
RB:†††† So, Aart, I feel lucky to have, you know, lived in this fifty-year period and gone through all of this as we both have and so this little conversation will be retained obviously. So sort of a final question Ė looking back, what would you do differently? Would you do it all again? Are you glad you did everything you did?
AdG:† Well the, you know, for starters, Iím an active working stiff here, meaning all my focus is on tomorrow and not much looking back. Secondly, you know, Iím a little bit of the frame of mind that you make whatever decisions, good, bad, or indifferent that you make under the circumstance you made them then. And sure you can look back and say well, you know, if I could have done it differently, I should have. But, you know, thatís quickly going to the why donít you predict the right lottery number because now in hindsight, we know what it is. You donít. You do the best you can and hopefully you have the right motivation going forward and then you learn as quickly as possible. Now the other line of thinking is to say, you know, it would also have been cool to have been a professional musician and then read about all this exciting electronic stuff. And when I talk to musicians, they all think that they have a really, you know, boring life and how exciting it must be to be in advanced technology. Well, you know, there are many lives that one wished one could have had and Iím glad that, you know, have a couple of percent that are in music and, you know.
RB:†††† Maybe, maybe, must maybe weíll see the Aart de Geus Country Four at the local pub sometime in the next ten years.
AdG:† Yeah, yeah, weíre less into country music than blues and jazzy blues. But, frankly, it doesnít matter what it is as long as what you do, you do with great enthusiasm and with people that you really like and respect. Those are great gifts that life has.
RB:†††† Well, Aart, congratulations on the success of Synopsys. Itís been fantastic and it still is and looking forward, thereís a great future in continuing to design these chips. So Iíd just like to thank you finally for allowing us to have this conversation with you and inviting us here to Synopsys. Itís been valuable and weíre glad to have you in the Stanford Library as one of the Silicon Genesis folks. And congratulations again and thank you.
AdG:† Well thank you. Itís a privilege to be considered for this. Thank you.
[End of Interview with Aart de Geus Ė March 7, 2016]