WEBVTT 00:00:00.000 --> 00:00:02.839 Welcome back to the Deep Dive. So today we're 00:00:02.839 --> 00:00:05.719 pulling back the curtain on a company that it 00:00:05.719 --> 00:00:08.039 designs almost everything electronic you interact 00:00:08.039 --> 00:00:10.900 with every single day. And yet. And yet its name 00:00:10.900 --> 00:00:13.279 is probably one you've never seen on an actual 00:00:13.279 --> 00:00:16.760 product. We're talking about Cadence Design Systems 00:00:16.760 --> 00:00:21.079 Inc. or CDNS. Right. This isn't just another 00:00:21.079 --> 00:00:23.920 tech giant hidden in plain sight. It's really 00:00:23.920 --> 00:00:26.699 the invisible software architect of modern life. 00:00:26.879 --> 00:00:28.600 That's a great way to put it. They're the ones 00:00:28.600 --> 00:00:31.719 selling the shovels and picks to, well, to everyone 00:00:31.719 --> 00:00:34.119 digging for gold in the Silicon Rocks. It really 00:00:34.119 --> 00:00:36.579 is the infrastructure for innovation. And, you 00:00:36.579 --> 00:00:38.700 know, its obscurity is pretty deceptive. They're 00:00:38.700 --> 00:00:40.619 headquartered right here in San Jose, California, 00:00:40.939 --> 00:00:43.899 a multinational computational software powerhouse. 00:00:44.380 --> 00:00:46.479 And when you say powerhouse, the general public 00:00:46.479 --> 00:00:48.799 rarely thinks about this category. They helped 00:00:48.799 --> 00:00:51.960 pioneer electronic design automation or ED. Oh, 00:00:51.979 --> 00:00:54.500 almost never. Yeah. But Cadence's status is a 00:00:54.500 --> 00:00:56.759 component of both the NASDAQ 100 and the S &P 00:00:56.759 --> 00:00:59.560 500. Well, that tells you exactly how essential 00:00:59.560 --> 00:01:01.539 they are. It tells the whole story. It really 00:01:01.539 --> 00:01:04.239 does. They're like a utility provider for the 00:01:04.239 --> 00:01:07.599 digital age, you know, like water or power. Only 00:01:07.599 --> 00:01:10.420 their product is the actual capability to design 00:01:10.420 --> 00:01:12.700 things. All right. So let's unpack this. We have 00:01:12.700 --> 00:01:15.540 a... a massive stack of source material here 00:01:15.540 --> 00:01:18.299 and it details not just their foundational history 00:01:18.299 --> 00:01:20.920 and semiconductors which is complex enough on 00:01:20.920 --> 00:01:24.269 its own it is but they're truly radical and I 00:01:24.269 --> 00:01:26.329 have to say, pretty unexpected expansion into 00:01:26.329 --> 00:01:28.750 fields we wouldn't normally associate with chip 00:01:28.750 --> 00:01:31.670 design. Not at all. Think aerospace, defense. 00:01:31.790 --> 00:01:34.430 Even pharmaceutical research. It's wild. Exactly. 00:01:34.769 --> 00:01:37.030 So our mission for this deep dive is to really 00:01:37.030 --> 00:01:39.950 define the scope of Cadence's evolution. We need 00:01:39.950 --> 00:01:43.430 to understand the strategic and the technical 00:01:43.430 --> 00:01:45.230 journey that allowed them to go from being this 00:01:45.230 --> 00:01:47.670 very specialized EDA firm. The folks who helped 00:01:47.670 --> 00:01:49.989 design integrated circuits, systems on chips, 00:01:50.129 --> 00:01:54.140 soCs. Right. From that. into this sprawling computational 00:01:54.140 --> 00:01:56.780 giant. And the question isn't just what they 00:01:56.780 --> 00:02:00.079 acquired, but how their core expertise in simulation 00:02:00.079 --> 00:02:03.060 and modeling proves so transferable to completely 00:02:03.060 --> 00:02:06.099 different physical domains. And the scale of 00:02:06.099 --> 00:02:08.740 this operation is just immense. I mean, the company's 00:02:08.740 --> 00:02:11.280 origins go back to 1983, but it was officially 00:02:11.280 --> 00:02:13.719 incorporated via that critical merger in 1988. 00:02:14.120 --> 00:02:17.400 A key moment. Since then, Cadence has just grown 00:02:17.400 --> 00:02:20.620 into this multibillion -dollar enterprise. The 00:02:20.620 --> 00:02:23.340 most recent figures we have show 2024 annual 00:02:23.340 --> 00:02:26.659 revenue clocking in at a massive U .S. $4 .64 00:02:26.659 --> 00:02:29.620 billion. And that's supported by, what, around 00:02:29.620 --> 00:02:33.900 11 ,700 employees worldwide? That's a very significant 00:02:33.900 --> 00:02:36.439 footprint. A huge footprint. And, you know, that 00:02:36.439 --> 00:02:39.819 U .S. $4 .64 billion is so much more diversified 00:02:39.819 --> 00:02:41.939 than it was even 10 years ago. While they still 00:02:41.939 --> 00:02:44.099 make the essential software and hardware for 00:02:44.099 --> 00:02:47.060 designing ICs, SOCs, and printed circuit boards. 00:02:47.180 --> 00:02:48.360 For bread and butter. For bread and butter, yeah. 00:02:48.500 --> 00:02:50.780 But a huge portion of their business now involves 00:02:50.780 --> 00:02:52.900 licensing specialized and intellectual property. 00:02:53.020 --> 00:02:55.280 These pre -designed performance guaranteed components 00:02:55.280 --> 00:02:58.240 we call SIP blocks to these core industries. 00:02:58.620 --> 00:03:00.740 So they're providing the optimized building blocks 00:03:00.740 --> 00:03:03.219 for... Everything. Everything. High frequency 00:03:03.219 --> 00:03:05.539 radar and defense, advanced sensors in the automotive 00:03:05.539 --> 00:03:08.219 sector, you name it. So, OK, what does this all 00:03:08.219 --> 00:03:10.759 mean for you, the listener? We're talking about 00:03:10.759 --> 00:03:13.319 a company that really dictates the pace and the 00:03:13.319 --> 00:03:16.340 possibility of modern high tech manufacturing. 00:03:16.599 --> 00:03:18.740 You really are. We'll be tracing a story defined 00:03:18.740 --> 00:03:21.930 by just. intense corporate rivalry, sometimes 00:03:21.930 --> 00:03:24.909 shocking legal battles. And we're talking literal 00:03:24.909 --> 00:03:27.810 corporate thriller territory here. Oh, yeah. 00:03:27.909 --> 00:03:31.210 And these profound high stakes global controversies 00:03:31.210 --> 00:03:33.789 around technology transfer. So get ready for 00:03:33.789 --> 00:03:36.229 a deep dive into the complex technology and the 00:03:36.229 --> 00:03:38.469 strategic moves that made one software company 00:03:38.469 --> 00:03:41.330 absolutely crucial to designing, well, almost 00:03:41.330 --> 00:03:44.210 every modern high tech system. To really understand 00:03:44.210 --> 00:03:46.870 where Cadence is going today. with, you know, 00:03:46.889 --> 00:03:49.030 AI and drug discovery, we have to look back at 00:03:49.030 --> 00:03:51.090 its foundational moment. Right. Which, as you 00:03:51.090 --> 00:03:53.789 said, was less a single startup and more a strategic 00:03:53.789 --> 00:03:56.449 synthesis of existing industry heavyweights. 00:03:56.509 --> 00:03:58.870 It's a classic Silicon Valley origin story, right? 00:03:58.930 --> 00:04:01.349 It all starts with academic rigor. The narrative 00:04:01.349 --> 00:04:04.469 really begins in 1983 with a company called Solomon 00:04:04.469 --> 00:04:07.610 Design Automation, or SDA. And this firm was 00:04:07.610 --> 00:04:10.090 co -founded by James Solomon, Richard Newton, 00:04:10.310 --> 00:04:13.250 and Alberto San Giovanni Vincentelli. I mean, 00:04:13.270 --> 00:04:16.189 these were absolute luminaries. in the burgeoning 00:04:16.189 --> 00:04:18.769 field of electronic design tools. In this trio, 00:04:18.970 --> 00:04:22.610 they immediately establish the deep academic 00:04:22.610 --> 00:04:27.250 and technical credibility that Cadence would 00:04:27.250 --> 00:04:30.730 just rely on for decades. SDA was focused on 00:04:30.730 --> 00:04:32.829 the next generation of design tools. But the 00:04:32.829 --> 00:04:35.089 true corporate formation, the point where the 00:04:35.089 --> 00:04:37.449 name Cadence Design Systems was actually born, 00:04:37.649 --> 00:04:40.810 that came in 1988 through a really critical merger. 00:04:41.069 --> 00:04:44.149 It did. SDA combined forces with ECAD, which 00:04:44.149 --> 00:04:46.269 was a public company co -founded back in 1982 00:04:46.269 --> 00:04:49.949 by Ping Chao, Glenn Antle, and Paul Huang. And 00:04:49.949 --> 00:04:52.149 ECAD brought something specific to the table. 00:04:52.329 --> 00:04:54.170 They did. They brought expertise in physical 00:04:54.170 --> 00:04:56.689 verification. So, you know, ensuring the designs 00:04:56.689 --> 00:04:58.370 were actually manufacturable. It's one thing 00:04:58.370 --> 00:04:59.810 to design a chip. It's another to be able to 00:04:59.810 --> 00:05:01.610 build it. And the leadership was immediately 00:05:01.610 --> 00:05:04.490 set with this mandate for just aggressive growth. 00:05:04.910 --> 00:05:07.730 Joseph Costello was appointed the first CEO and 00:05:07.730 --> 00:05:10.149 president of the newly combined company. And 00:05:10.149 --> 00:05:13.189 once they merged, Cadence began trading on the 00:05:13.189 --> 00:05:15.629 New York Stock Exchange, which really marked 00:05:15.629 --> 00:05:17.990 the start of a period defined not just by innovation, 00:05:18.110 --> 00:05:21.449 but by intense market consolidation. And here's 00:05:21.449 --> 00:05:23.759 where it gets really interesting. Costello, he 00:05:23.759 --> 00:05:25.959 immediately embraced this aggressive strategy 00:05:25.959 --> 00:05:30.790 of just intense growth through mergers and acquisitions. 00:05:30.850 --> 00:05:33.930 A feeding frenzy. He was essentially buying up 00:05:33.930 --> 00:05:37.430 the competition and consolidating all these disparate 00:05:37.430 --> 00:05:39.689 EDA technologies under the cadence umbrella. 00:05:40.009 --> 00:05:42.389 It was an attempt to create a single unified 00:05:42.389 --> 00:05:45.610 design flow that no rival could possibly match. 00:05:45.829 --> 00:05:48.670 And that aggressive M &A strategy is the absolute 00:05:48.670 --> 00:05:51.649 key to their early market dominance. They weren't 00:05:51.649 --> 00:05:53.509 just building a product. No, they were building 00:05:53.509 --> 00:05:56.069 an entire ecosystem by consuming other companies. 00:05:56.720 --> 00:05:59.300 think about the speed. Only one year after the 00:05:59.300 --> 00:06:02.500 merger in 1989, they acquired Gateway Design 00:06:02.500 --> 00:06:05.899 Automation for $72 million. And that was a crucial 00:06:05.899 --> 00:06:07.939 buy. It was crucial because it significantly 00:06:07.939 --> 00:06:10.480 boosted their simulation and verification software 00:06:10.480 --> 00:06:13.699 capabilities, which are essential for checking 00:06:13.699 --> 00:06:15.500 a chip's functionality before it's taped out. 00:06:15.639 --> 00:06:18.779 And they very quickly understood the need to 00:06:18.779 --> 00:06:22.040 move beyond just designing the chip itself. The 00:06:22.040 --> 00:06:25.920 very next year, 1990, Cadence acquired Automation 00:06:25.930 --> 00:06:29.910 Which was a hugely strategic move. Why so strategic? 00:06:30.209 --> 00:06:32.889 Because it added board design and layout capabilities 00:06:32.889 --> 00:06:35.310 to their existing line of chip design software. 00:06:35.790 --> 00:06:38.470 So suddenly they weren't just selling tools to 00:06:38.470 --> 00:06:40.389 design the components. They were building out 00:06:40.389 --> 00:06:42.889 the full ecosystem for electronic product development. 00:06:43.269 --> 00:06:45.449 Exactly. From the smallest transistor to the 00:06:45.449 --> 00:06:47.350 largest printed circuit board that holds all 00:06:47.350 --> 00:06:50.290 the chips. But the acquisition in 1991, that 00:06:50.290 --> 00:06:52.149 was the biggest statement they could have possibly 00:06:52.149 --> 00:06:54.589 made to the industry. Oh, absolutely. They purchased 00:06:54.589 --> 00:06:57.589 their major rival, Valid Logic Systems, for around 00:06:57.589 --> 00:07:00.790 $200 million. This was a massive consolidation 00:07:00.790 --> 00:07:03.610 move that just instantly reduced the competitive 00:07:03.610 --> 00:07:05.639 landscape. And the sources... confirmed that 00:07:05.639 --> 00:07:08.079 the revenues of the newly combined company immediately 00:07:08.079 --> 00:07:11.699 shot up to $390 million. It was a signal. It 00:07:11.699 --> 00:07:14.279 was Cadence signaling its intent to completely 00:07:14.279 --> 00:07:18.399 dominate the nascent high stakes EDA space. That 00:07:18.399 --> 00:07:20.540 was the moment they really stepped up as the 00:07:20.540 --> 00:07:24.180 market leader. And the mid to late 90s just saw 00:07:24.180 --> 00:07:27.259 this acquisition spree continue completely unabated. 00:07:27.259 --> 00:07:30.339 It was almost like a feeding frenzy. They acquired 00:07:30.339 --> 00:07:34.040 high -level design systems in 1996. At which 00:07:34.040 --> 00:07:36.819 point Cadence had grown significantly, right? 00:07:36.899 --> 00:07:39.759 They were boasting, what, $742 million in annual 00:07:39.759 --> 00:07:42.519 revenue? That's right. And they also snapped 00:07:42.519 --> 00:07:47.019 up Ambit Design Systems in 1998 for $260 million, 00:07:47.339 --> 00:07:50.779 a massive price tag. And that was focused specifically 00:07:50.779 --> 00:07:54.500 on acquiring superior synthesis tools for system 00:07:54.500 --> 00:07:56.339 -on -a -chip technology. Which showed a lot of 00:07:56.339 --> 00:07:58.620 foresight. A ton of it. They recognized that 00:07:58.620 --> 00:08:01.199 chip complexity was growing exponentially and 00:08:01.199 --> 00:08:03.660 that it would require much better tools to synthesize 00:08:03.660 --> 00:08:05.639 these massive designs. And they rounded out the 00:08:05.639 --> 00:08:08.000 decade by bringing in Orcad systems in 1999. 00:08:08.519 --> 00:08:10.740 Orcad was significant because it focused so heavily 00:08:10.740 --> 00:08:14.379 on PCB design and FBGA design tools. Again, these 00:08:14.379 --> 00:08:16.600 were not random purchases. They were systematically 00:08:16.600 --> 00:08:19.019 filling every niche. And building the most comprehensive 00:08:19.019 --> 00:08:21.500 and tightly integrated design tool chain possible. 00:08:21.560 --> 00:08:23.800 Making it incredibly difficult for customers 00:08:23.800 --> 00:08:26.040 to even think about switching to a rival suite. 00:08:26.399 --> 00:08:28.819 Which brings us to what I think is one of the 00:08:28.819 --> 00:08:31.120 most compelling stories of this era, the quick 00:08:31.120 --> 00:08:34.120 turn defense. Ah, yes. The 1999 acquisition of 00:08:34.120 --> 00:08:36.840 quick turn design systems was less about, you 00:08:36.840 --> 00:08:40.320 know, pure technological aspiration and more 00:08:40.320 --> 00:08:42.700 about a desperate high stakes corporate defense 00:08:42.700 --> 00:08:45.080 against a major competitor. That's precisely 00:08:45.080 --> 00:08:47.659 right. And it offers great insight into the absolutely 00:08:47.659 --> 00:08:50.679 ruthless nature of the EDA market back then. 00:08:51.080 --> 00:08:53.220 QuickTurn was the leading provider of hardware 00:08:53.220 --> 00:08:56.179 emulation technology. And this acquisition, it 00:08:56.179 --> 00:08:58.980 was explicitly done to prevent a hostile takeover 00:08:58.980 --> 00:09:02.340 attempt by their rival, Mentor Graphics. Who 00:09:02.340 --> 00:09:04.200 had tried to acquire QuickTurn themselves, yeah. 00:09:04.360 --> 00:09:07.000 Cadence just swooped in with a superior offer 00:09:07.000 --> 00:09:10.299 and a very clear strategic goal. Deny Mentor 00:09:10.299 --> 00:09:13.860 a critical technology edge. Okay, so. Why was 00:09:13.860 --> 00:09:16.659 this emulation technology so important beyond 00:09:16.659 --> 00:09:19.000 just the corporate rivalry? For our listener, 00:09:19.080 --> 00:09:21.299 can you clarify what emulation is and how it 00:09:21.299 --> 00:09:23.860 became so foundational? Certainly. So simulation, 00:09:24.179 --> 00:09:26.879 which Cadence also does, uses software to model 00:09:26.879 --> 00:09:29.559 a chip's behavior. But when a chip design reaches, 00:09:29.740 --> 00:09:32.580 say, a billion gates, simulation just becomes 00:09:32.580 --> 00:09:35.399 way too slow. It's not practical anymore. Not 00:09:35.399 --> 00:09:38.460 at all. Emulation uses specialized hardware. 00:09:38.940 --> 00:09:40.919 essentially these giant reconfigurable arrays 00:09:40.919 --> 00:09:44.620 of chips, to run the actual design logic at near 00:09:44.620 --> 00:09:47.059 real -time speeds. Which lets engineers verify 00:09:47.059 --> 00:09:49.419 that the chip actually works with the software 00:09:49.419 --> 00:09:52.659 it's designed to run long before the actual silicon 00:09:52.659 --> 00:09:55.740 is manufactured. Precisely. So the technology 00:09:55.740 --> 00:09:57.820 Cadence got from QuickTurn wasn't just useful, 00:09:57.960 --> 00:10:00.700 it was foundational. It later became the core 00:10:00.700 --> 00:10:03.460 of Cadence's highly profitable hardware products 00:10:03.460 --> 00:10:05.799 like the Palladium platform. Securing a revenue 00:10:05.799 --> 00:10:08.600 stream for years. For years to come. It was a 00:10:08.600 --> 00:10:10.720 defensive maneuver that yielded just immense 00:10:10.720 --> 00:10:13.720 long -term offensive market capabilities. But 00:10:13.720 --> 00:10:16.320 that period of intense growth and rivalry, it 00:10:16.320 --> 00:10:18.320 also came with significant leadership volatility. 00:10:19.159 --> 00:10:21.500 Joe Costello, the original CEO who drove this 00:10:21.500 --> 00:10:24.320 aggressive M &A strategy, he resigned in 1997. 00:10:24.700 --> 00:10:27.240 Right. And he was succeeded briefly by Jack Harding 00:10:27.240 --> 00:10:29.720 and then by Ray Bingham in 1999. And this change 00:10:29.720 --> 00:10:32.620 in leadership coincided with just intense competitive 00:10:32.620 --> 00:10:35.299 pressure, specifically from Synopsys. Who rapidly 00:10:35.299 --> 00:10:37.960 grew into their primary market rival. This rivalry 00:10:37.960 --> 00:10:40.039 really came to a head when Synopsys acquired 00:10:40.039 --> 00:10:42.820 Avanti in 2001. A company we'll be discussing 00:10:42.820 --> 00:10:44.980 in a lot more detail later because of its... 00:10:45.720 --> 00:10:48.879 infamous legal battles. Yes, we will. And this 00:10:48.879 --> 00:10:52.000 strategic move by Synopsys really put Cadence 00:10:52.000 --> 00:10:54.759 on the defensive, challenging their market dominance 00:10:54.759 --> 00:10:57.480 in these crucial implementation tool areas. And 00:10:57.480 --> 00:10:59.879 Cadence's strategic response to this competitive 00:10:59.879 --> 00:11:02.820 threat was, well, it was textbook for their history. 00:11:03.059 --> 00:11:05.820 More aggressive, targeted acquisition. Exactly. 00:11:05.899 --> 00:11:09.179 Between 2001 and 2003, they purchased multiple 00:11:09.179 --> 00:11:11.860 implementation tools specifically to counter 00:11:11.860 --> 00:11:14.600 Synopsys' market strength. These included Silicon 00:11:14.600 --> 00:11:17.460 Perspective, Verplex, and Celestri Design. They 00:11:17.460 --> 00:11:19.879 were essentially buying speed and parity, making 00:11:19.879 --> 00:11:22.059 sure they had competitive tools for the physical 00:11:22.059 --> 00:11:24.539 design flow. The leadership carousel kept turning, 00:11:24.600 --> 00:11:27.220 though. Mike Pfister took over as CEO in 2004. 00:11:27.700 --> 00:11:30.019 The most important organizational shift, I think, 00:11:30.039 --> 00:11:33.019 came later. In 2008, after Pfister resigned, 00:11:33.600 --> 00:11:36.179 Lipu Tan, a really respected venture capitalist 00:11:36.179 --> 00:11:39.120 and board member, he stepped in as acting CEO. 00:11:39.419 --> 00:11:41.820 And he was confirmed to the role in January 2009. 00:11:42.259 --> 00:11:45.399 Right. And Tan's tenure, it really marked a stabilizing 00:11:45.399 --> 00:11:48.440 period after years of tumult. It set the stage 00:11:48.440 --> 00:11:51.799 for the dramatic and systematic diversification 00:11:51.799 --> 00:11:54.059 we're seeing today. He focused the company not 00:11:54.059 --> 00:11:56.440 just on tools, but on high margin intellectual 00:11:56.440 --> 00:11:59.539 property licensing. And that focus is crystal 00:11:59.539 --> 00:12:01.940 clear. Under Tan's leadership, the strategic 00:12:01.940 --> 00:12:05.080 acquisitions continued. But they became so much 00:12:05.080 --> 00:12:07.679 more focused on high -value IP blocks rather 00:12:07.679 --> 00:12:10.419 than just design tools. You saw that with Alto's 00:12:10.419 --> 00:12:13.379 design automation in 2011, Cosmic Circuits in 00:12:13.379 --> 00:12:16.759 2013. And crucially, Tensilica in 2013 for their 00:12:16.759 --> 00:12:19.240 highly specialized data plane processing IP. 00:12:19.600 --> 00:12:21.779 That decisive move into specialized intellectual 00:12:21.779 --> 00:12:25.679 property blocks, the SIP signaled a major shift 00:12:25.679 --> 00:12:27.980 toward high -margin licensing, which perfectly 00:12:27.980 --> 00:12:30.080 supplemented their core software revenue. So 00:12:30.080 --> 00:12:32.759 the current decade, 2020 onwards, hasn't just 00:12:32.759 --> 00:12:35.440 been about refining ED. Not at all. It has been 00:12:35.440 --> 00:12:37.480 about using the core expertise Cadence gained 00:12:37.480 --> 00:12:39.860 in computational modeling, simulation, and verification 00:12:39.860 --> 00:12:43.000 to completely redefine what Cadence is. This 00:12:43.000 --> 00:12:45.059 is the point where the company transforms from 00:12:45.059 --> 00:12:48.039 an EDA specialist into a comprehensive computational 00:12:48.039 --> 00:12:50.919 software giant. And the first major indication 00:12:50.919 --> 00:12:53.399 of this shift is just looking at who's cutting 00:12:53.399 --> 00:12:56.659 the checks. We know Cadence's customers were 00:12:56.659 --> 00:12:58.899 traditionally pure semiconductor firms, you know, 00:12:58.940 --> 00:13:02.860 Intel, AMD, NVIDIA. Right. But our sources note 00:13:02.860 --> 00:13:06.740 this profound change. By 2022, a substantial 00:13:06.740 --> 00:13:10.139 portion, about 40 % of their revenue, was coming 00:13:10.139 --> 00:13:13.600 from systems -oriented customers. This is a fundamental 00:13:13.600 --> 00:13:15.700 change in the business model that reflects a 00:13:15.700 --> 00:13:18.960 major industry trend. Systems customers are companies 00:13:18.960 --> 00:13:22.129 like Tesla, Apple Inc. and amazon web services 00:13:22.129 --> 00:13:24.970 so companies not in the business of selling chips 00:13:24.970 --> 00:13:27.370 on the open market exactly they're building cars 00:13:27.370 --> 00:13:30.149 phones or massive data centers and they want 00:13:30.149 --> 00:13:33.090 highly tailored customized silicon that perfectly 00:13:33.090 --> 00:13:35.820 integrates into their final product They realize 00:13:35.820 --> 00:13:37.840 that off -the -shelf chips often compromise on 00:13:37.840 --> 00:13:40.200 performance or power efficiency. So Cadence steps 00:13:40.200 --> 00:13:42.500 in to fill that gap. These clients aren't buying 00:13:42.500 --> 00:13:44.500 generic semiconductors. They want Cadence to 00:13:44.500 --> 00:13:46.559 help them design a custom chip, an application 00:13:46.559 --> 00:13:49.039 -specific integrated circuit that perfectly fits 00:13:49.039 --> 00:13:51.340 their product's niche requirements. And this 00:13:51.340 --> 00:13:53.240 became especially important during the global 00:13:53.240 --> 00:13:55.940 chip shortages, when supply chain control became 00:13:55.940 --> 00:13:58.720 absolutely paramount. So Cadence designs it, 00:13:58.759 --> 00:14:01.179 verifies it, and then outsources the manufacturing 00:14:01.179 --> 00:14:05.500 to a foundry like TSMC. They've become the bespoke 00:14:05.500 --> 00:14:08.629 tailors of silicon. And it's driving this massive 00:14:08.629 --> 00:14:11.690 internal chip design movement across industries. 00:14:11.990 --> 00:14:14.350 This systemic shift, though, it must have required 00:14:14.350 --> 00:14:17.549 Cadence to expand its toolkit far beyond just 00:14:17.549 --> 00:14:19.730 traditional electrical engineering simulation. 00:14:20.129 --> 00:14:22.070 Oh, absolutely. When you design a custom chip 00:14:22.070 --> 00:14:25.070 for a car, you need to simulate the car's electrical 00:14:25.070 --> 00:14:27.669 system, its thermal profile, its interaction 00:14:27.669 --> 00:14:29.690 with the environment. You have to simulate the 00:14:29.690 --> 00:14:32.429 entire system the chip operates within. Exactly. 00:14:32.750 --> 00:14:35.429 Which meant expansion into computational fluid 00:14:35.429 --> 00:14:39.419 dynamics. or CFD, complex thermal analysis and 00:14:39.419 --> 00:14:41.940 electromagnetics, often grouped under the umbrella 00:14:41.940 --> 00:14:44.399 of multi -physics simulation. And that brings 00:14:44.399 --> 00:14:46.419 us to the targeted wave of acquisitions that 00:14:46.419 --> 00:14:48.779 push them deep into these system analysis tools. 00:14:48.940 --> 00:14:50.960 This isn't just incremental improvement. This 00:14:50.960 --> 00:14:53.259 is them buying expertise in completely different 00:14:53.259 --> 00:14:55.720 scientific fields. That's right. I mean, look 00:14:55.720 --> 00:14:59.340 at 2021. They acquired Numeca, which specialized 00:14:59.340 --> 00:15:02.659 in CFD and multi -physics simulations for major 00:15:02.659 --> 00:15:05.080 industrial sectors. We're talking marine, automotive. 00:15:05.710 --> 00:15:08.409 power generation right and cfd for example is 00:15:08.409 --> 00:15:11.789 all about modeling air or fluid flow how do they 00:15:11.789 --> 00:15:14.590 connect that to ed that's the question well both 00:15:14.590 --> 00:15:17.769 are fundamentally about solving massive complex 00:15:17.769 --> 00:15:21.190 partial differential equations on a grid cadence 00:15:21.190 --> 00:15:23.250 realized their computational platform could be 00:15:23.250 --> 00:15:25.659 repurposed so they also bought point -wise that 00:15:25.659 --> 00:15:28.100 year. Which provided essential CFD mesh generation 00:15:28.100 --> 00:15:30.720 tools, ensuring they owned the critical front 00:15:30.720 --> 00:15:33.220 -end technology for these simulations. These 00:15:33.220 --> 00:15:35.779 were direct injections of deep physics expertise 00:15:35.779 --> 00:15:38.399 that just instantly broadened their addressable 00:15:38.399 --> 00:15:40.679 market. And the integration of this new expertise 00:15:40.679 --> 00:15:43.940 is what led to this massive, yet very quiet leap 00:15:43.940 --> 00:15:47.820 in 2024. Cadence unveiled the Millennium M1 supercomputer. 00:15:48.340 --> 00:15:50.519 Which is extraordinary. A software company building 00:15:50.519 --> 00:15:52.850 its own computational hardware. It's a decisive 00:15:52.850 --> 00:15:55.070 move toward vertical integration and computation. 00:15:55.549 --> 00:15:58.289 The M1 supercomputer isn't a general purpose 00:15:58.289 --> 00:16:01.149 machine. It's specifically engineered by Cadence 00:16:01.149 --> 00:16:04.590 to run their complex CFD calculations and multi 00:16:04.590 --> 00:16:07.190 -physics models faster than any general purpose 00:16:07.190 --> 00:16:10.889 hardware could. And crucially, it uses AI acceleration. 00:16:11.370 --> 00:16:14.730 It utilizes AI to enhance the speed and the fidelity 00:16:14.730 --> 00:16:17.970 of the simulation process. This marks Cadence's 00:16:17.970 --> 00:16:20.389 definitive entry into the high -performance computing 00:16:20.389 --> 00:16:24.389 or supercomputer business, not as a vendor, but 00:16:24.389 --> 00:16:26.429 as a creator of specialized engines for their 00:16:26.429 --> 00:16:29.289 own software stack. This computational leap is 00:16:29.289 --> 00:16:31.389 just, it's wild. I mean, you mentioned aerospace, 00:16:31.549 --> 00:16:34.620 automotive. But the most surprising diversification, 00:16:34.759 --> 00:16:37.500 it has to be the move into life sciences. Absolutely. 00:16:37.980 --> 00:16:39.539 Pharmaceutical drug discovery, specifically. 00:16:39.759 --> 00:16:41.860 I mean, how on earth do you jump from designing 00:16:41.860 --> 00:16:45.799 a microprocessor to simulating a protein folding 00:16:45.799 --> 00:16:47.379 problem? Well, it sounds counterintuitive, I 00:16:47.379 --> 00:16:49.620 know. But only until you look past the application 00:16:49.620 --> 00:16:54.019 itself and you focus on the math. The math. Designing 00:16:54.019 --> 00:16:56.460 a new drug molecule. Predicting how it interacts 00:16:56.460 --> 00:17:00.070 with a biological target. At its core, it's a 00:17:00.070 --> 00:17:02.929 massive, complex molecular simulation problem. 00:17:03.129 --> 00:17:05.829 You're simulating the interactions, the energy 00:17:05.829 --> 00:17:08.390 states, the conformational changes of millions 00:17:08.390 --> 00:17:11.430 of atoms. Which is mathematically analogous to 00:17:11.430 --> 00:17:14.230 simulating complex electronic circuits or fluid 00:17:14.230 --> 00:17:18.029 flows in 3D space. Exactly. And in 2022, Cadence 00:17:18.029 --> 00:17:20.130 highlighted this insight by acquiring OpenEye 00:17:20.130 --> 00:17:23.710 Scientific Software for $500 million. Half a 00:17:23.710 --> 00:17:25.769 billion dollars for a company that immediately 00:17:25.769 --> 00:17:28.950 rebranded as OpenEye Cadence Molecular Sciences. 00:17:29.369 --> 00:17:32.430 So what specific problem do they solve for pharmaceutical 00:17:32.430 --> 00:17:35.130 companies? They focus on computational molecular 00:17:35.130 --> 00:17:37.990 modeling and simulation software tools like the 00:17:37.990 --> 00:17:40.470 Orion Software -as -a -Service platform. This 00:17:40.470 --> 00:17:42.910 platform uses cloud -native tools to run massive 00:17:42.910 --> 00:17:45.130 simulations for virtual screening, trying to 00:17:45.130 --> 00:17:47.589 rapidly identify and optimize potential drug 00:17:47.589 --> 00:17:49.609 candidates or antibodies. So it dramatically 00:17:49.609 --> 00:17:52.289 accelerates the early phases of discovery. Right, 00:17:52.329 --> 00:17:54.509 replacing expensive, time -consuming wet lab 00:17:54.509 --> 00:17:56.269 experiments with high -fidelity simulations. 00:17:56.829 --> 00:18:00.589 And this isn't just theory. By late 2023, OpenEye 00:18:00.589 --> 00:18:03.430 Cadence Molecular Sciences had secured Pfizer 00:18:03.430 --> 00:18:05.869 as a client for its molecular design software. 00:18:06.460 --> 00:18:08.880 Which confirms that Cadence is successfully leveraging 00:18:08.880 --> 00:18:11.700 its computational architecture expertise, the 00:18:11.700 --> 00:18:14.140 simulation and high -speed parallel processing 00:18:14.140 --> 00:18:17.099 know -how developed in silicon to solve these 00:18:17.099 --> 00:18:19.740 highly complex biological problems at the molecular 00:18:19.740 --> 00:18:22.279 level. The fidelity of their models is crossing 00:18:22.279 --> 00:18:24.660 the threshold of commercial viability in pharma. 00:18:24.779 --> 00:18:26.920 But the culmination of this radical diversification 00:18:26.920 --> 00:18:30.059 drive came with the massive September 2025 announcement, 00:18:30.359 --> 00:18:33.259 establishing Cadence as a true industrial software 00:18:33.259 --> 00:18:37.160 powerhouse. The 2025 mega deal. Cadence announced 00:18:37.160 --> 00:18:39.319 it would acquire the design and engineering business 00:18:39.319 --> 00:18:42.920 of Hexagon AB, a major Stockholm -based industrial 00:18:42.920 --> 00:18:46.359 technology company. For a staggering 2 .7 billion 00:18:46.359 --> 00:18:49.880 euros. Which translates to approximately 3 .16 00:18:49.880 --> 00:18:52.440 billion dollars. And this wasn't just another 00:18:52.440 --> 00:18:54.859 bolt -on acquisition. It fundamentally reshaped 00:18:54.859 --> 00:18:57.609 Cadence's profile in the industrial sector. The 00:18:57.609 --> 00:19:00.089 acquisition includes Hexagon's MSC software business. 00:19:00.369 --> 00:19:04.109 And MSC is a major established provider of engineering 00:19:04.109 --> 00:19:06.829 simulation and analysis software. Right, particularly 00:19:06.829 --> 00:19:10.150 in computer -aided engineering, or CAE. This 00:19:10.150 --> 00:19:12.309 includes things like structural analysis and 00:19:12.309 --> 00:19:15.109 dynamic system simulation. So by integrating 00:19:15.109 --> 00:19:18.849 MSC, Cadence instantly and significantly bolstered 00:19:18.849 --> 00:19:21.650 its multi -physics simulation capabilities. Making 00:19:21.650 --> 00:19:23.950 them a titan in simulating mechanical structures 00:19:23.950 --> 00:19:27.029 and complex physical systems, not just electronic 00:19:27.029 --> 00:19:29.970 ones. They can now offer a comprehensive digital 00:19:29.970 --> 00:19:33.230 twin platform for manufacturers, a complete virtual 00:19:33.230 --> 00:19:36.390 model of a product. A model that includes its 00:19:36.390 --> 00:19:39.230 electronics, its mechanical structure, its thermal 00:19:39.230 --> 00:19:42.309 performance and its fluid dynamics, all optimized 00:19:42.309 --> 00:19:45.140 together. This is the very definition of a computational 00:19:45.140 --> 00:19:47.720 giant. That radical diversification is fascinating, 00:19:47.859 --> 00:19:50.299 but it's so vital to remember that Cadence still 00:19:50.299 --> 00:19:52.779 remains the architect of modern silicon. Their 00:19:52.779 --> 00:19:55.579 product suite, the fundamental electronic design 00:19:55.579 --> 00:19:57.980 automation tools, that's the engine that allows 00:19:57.980 --> 00:20:00.460 nearly every advanced chip to be conceived, designed, 00:20:00.720 --> 00:20:03.930 and verified. So let's dive deep into the specific 00:20:03.930 --> 00:20:06.230 core tools that make integrated circuit design 00:20:06.230 --> 00:20:09.829 possible. Absolutely. The rigor of their EDA 00:20:09.829 --> 00:20:12.529 tools is what earned them the license to expand 00:20:12.529 --> 00:20:15.410 into these other computational domains. And we 00:20:15.410 --> 00:20:18.230 start with the core work, designing the integrated 00:20:18.230 --> 00:20:21.210 circuit itself. The flagship tool here used by 00:20:21.210 --> 00:20:23.950 analog and mixed signal engineers globally is 00:20:23.950 --> 00:20:27.089 Virtuoso Studio. Virtuoso, that's the classic 00:20:27.089 --> 00:20:29.910 full custom IC design environment. The bread 00:20:29.910 --> 00:20:31.670 and butter for creating high performance circuits 00:20:31.670 --> 00:20:33.910 that can't be automatically generated. It's where 00:20:33.910 --> 00:20:36.569 engineers spend years perfecting circuits. And 00:20:36.569 --> 00:20:38.849 within that environment, simulation speed is 00:20:38.849 --> 00:20:41.170 king. They are constantly pushing the envelope 00:20:41.170 --> 00:20:43.670 on performance, and a great example is Spectre 00:20:43.670 --> 00:20:46.730 X, introduced back in 2019. And Spectre X is 00:20:46.730 --> 00:20:49.750 a parallel circuit simulator. Exactly. To simplify, 00:20:50.069 --> 00:20:52.250 traditional simulators ran sequentially, testing 00:20:52.250 --> 00:20:54.730 one thing after another. Spectre X allows users 00:20:54.730 --> 00:20:57.309 to distribute extremely complex time and frequency 00:20:57.309 --> 00:21:00.230 domain simulations across hundreds of CPUs at 00:21:00.230 --> 00:21:02.369 the same time. And that computational horsepower 00:21:02.369 --> 00:21:04.809 is critical. Oh, it's critical because a single 00:21:04.809 --> 00:21:07.690 simulation run can often take days. And reducing 00:21:07.690 --> 00:21:09.769 that time means dramatically accelerating the 00:21:09.769 --> 00:21:12.680 entire design cycle. and for specialized communications 00:21:12.680 --> 00:21:14.960 which are increasingly important with 5g and 00:21:14.960 --> 00:21:18.779 radar they have awr right awr provides the design 00:21:18.779 --> 00:21:20.859 environment that's dedicated to radio frequency 00:21:20.859 --> 00:21:24.960 or rf to millimeter wave products it's crucial 00:21:24.960 --> 00:21:27.400 for modern connectivity and essential for defense 00:21:27.400 --> 00:21:30.319 applications satellite communications the dense 00:21:30.319 --> 00:21:33.359 network needed for autonomous vehicles and 5g 00:21:33.359 --> 00:21:36.779 6g wireless as signal frequencies get incredibly 00:21:36.779 --> 00:21:39.799 high the electromagnetic effects become dominant 00:21:40.400 --> 00:21:42.140 and have to be modeled with extreme precision. 00:21:42.920 --> 00:21:45.640 AWR provides the sophisticated tools necessary 00:21:45.640 --> 00:21:48.319 to design those sensitive circuits that operate 00:21:48.319 --> 00:21:50.880 at the absolute cutting edge of wireless communication. 00:21:51.279 --> 00:21:53.380 Now let's move from that custom analog world 00:21:53.380 --> 00:21:56.039 into the highly automated world of digital silicon. 00:21:56.319 --> 00:21:58.539 This is where the sheer complexity of modern 00:21:58.539 --> 00:22:00.660 microprocessors is managed. And we enter the 00:22:00.660 --> 00:22:03.059 realm of digital implementation and sign -off, 00:22:03.200 --> 00:22:05.700 which are the unseen processes that turn abstract 00:22:05.700 --> 00:22:08.400 ideas into physical reality. The digital factory 00:22:08.400 --> 00:22:11.460 floor. That's a great term for it. The core set 00:22:11.460 --> 00:22:14.940 of tools, like Genus, Innovus, Tempus, and Voltus, 00:22:15.039 --> 00:22:18.440 are essential for the final phases of design. 00:22:18.900 --> 00:22:21.180 Their role is to take a high -level description 00:22:21.180 --> 00:22:23.240 of what the chip should do. And translate it 00:22:23.240 --> 00:22:25.859 into a physical, manufacturable layout on silicon. 00:22:26.200 --> 00:22:28.660 Making sure it meets all the timing, power, and 00:22:28.660 --> 00:22:31.140 physical constraints before it goes to the foundry. 00:22:31.480 --> 00:22:33.740 Okay, for the listener, can we break down what 00:22:33.740 --> 00:22:37.039 two of those tools, genus and innibus, actually 00:22:37.039 --> 00:22:41.000 do? Of course. Genus is the synthesis tool. It 00:22:41.000 --> 00:22:42.759 takes the high -level description of the chip's 00:22:42.759 --> 00:22:45.009 function. usually written in hardware description 00:22:45.009 --> 00:22:47.829 languages. And it translates that into a detailed 00:22:47.829 --> 00:22:50.690 list of standard cells, basic logic gates that 00:22:50.690 --> 00:22:52.730 will perform that function. And then Inovus takes 00:22:52.730 --> 00:22:55.329 that huge list of gates and physically places 00:22:55.329 --> 00:22:57.690 and routes them on the chip surface. It's the 00:22:57.690 --> 00:22:59.529 process of figuring out the optimal position 00:22:59.529 --> 00:23:02.170 for billions of transistors and how to wire them 00:23:02.170 --> 00:23:03.950 all together efficiently. And because a complex 00:23:03.950 --> 00:23:05.950 chip can contain hundreds of millions of gates, 00:23:06.150 --> 00:23:09.269 this step is just far too complex for human optimization. 00:23:09.509 --> 00:23:12.509 Making these automated tools absolutely indispensable. 00:23:12.880 --> 00:23:16.400 So why are Tempus and Voltus separate? They handle 00:23:16.400 --> 00:23:19.049 what's called sign -off. Tempus verifies timing 00:23:19.049 --> 00:23:21.390 making, absolutely sure that the signals arrive 00:23:21.390 --> 00:23:23.670 where they need to go fast enough to meet the 00:23:23.670 --> 00:23:26.269 operating frequency. And Voltus verifies power. 00:23:26.529 --> 00:23:29.589 Right. Analyzing the complex power network and 00:23:29.589 --> 00:23:32.089 ensuring the chip won't overheat or suffer from 00:23:32.089 --> 00:23:35.230 voltage drops. In advanced chip nodes, timing 00:23:35.230 --> 00:23:37.950 and power analysis are so sensitive and computationally 00:23:37.950 --> 00:23:40.190 demanding that they require these specialized 00:23:40.190 --> 00:23:43.150 high -capacity sign -off tools. Now let's talk 00:23:43.150 --> 00:23:45.630 about translating that initial design idea into 00:23:45.630 --> 00:23:47.680 something genus and innovative. can even work 00:23:47.680 --> 00:23:51.799 with using Stratus. Ah, Stratus. Stratus is Cadence's 00:23:51.799 --> 00:23:54.940 high -level synthesis, or HLS, tool, and this 00:23:54.940 --> 00:23:57.579 is a critical productivity booster. How so? Well, 00:23:57.599 --> 00:23:59.740 instead of engineers manually writing detailed 00:23:59.740 --> 00:24:03.039 register transfer level code, the very low -level 00:24:03.039 --> 00:24:05.440 language that describes precisely how data moves 00:24:05.440 --> 00:24:08.259 and is stored in hardware registers. HLS allows 00:24:08.259 --> 00:24:10.240 them to use standard programming languages like 00:24:10.240 --> 00:24:14.500 C, C++, or System C. Exactly. And Stratus then 00:24:14.500 --> 00:24:17.019 automatically creates the... complex, detailed 00:24:17.019 --> 00:24:20.180 RTL implementation. This speeds up the initial 00:24:20.180 --> 00:24:22.940 design phase dramatically and it allows designers 00:24:22.940 --> 00:24:25.470 to focus on high -level architecture. rather 00:24:25.470 --> 00:24:28.170 than, you know, TD's manual coding. So once the 00:24:28.170 --> 00:24:30.730 design is synthesized and placed, we hit the 00:24:30.730 --> 00:24:33.569 true headache of every chip design project. Verification. 00:24:33.789 --> 00:24:36.549 Did we build the chip correctly? And Cadence 00:24:36.549 --> 00:24:40.109 has very robust system verification tools. Starting 00:24:40.109 --> 00:24:42.809 with formal verification, which sounds incredibly 00:24:42.809 --> 00:24:45.910 rigorous. It is. Formal verification is mathematically 00:24:45.910 --> 00:24:49.049 rigorous. It proves, using logic and algorithms, 00:24:49.269 --> 00:24:51.849 whether a design is correct based on its specification, 00:24:52.170 --> 00:24:54.309 rather than just simulating millions of tests. 00:24:54.509 --> 00:24:56.809 cases. And Jasper Gold is their flagship tool 00:24:56.809 --> 00:24:59.150 for this. It is. And it was upgraded in 2019 00:24:59.150 --> 00:25:01.329 with machine learning to enhance its efficiency, 00:25:01.789 --> 00:25:03.990 helping it navigate the massive state spaces 00:25:03.990 --> 00:25:06.730 of modern chips more quickly and identify obscure 00:25:06.730 --> 00:25:09.289 bugs that simulation might completely miss. And 00:25:09.289 --> 00:25:11.349 for standard simulation, which is still the primary 00:25:11.349 --> 00:25:13.690 way to check general functionality, they launched 00:25:13.690 --> 00:25:17.230 Excelium in 2017. Excelium really represents 00:25:17.230 --> 00:25:19.990 that relentless move to scale. It's a parallel 00:25:19.990 --> 00:25:23.059 simulator. based on a multi -core computing architecture 00:25:23.059 --> 00:25:25.559 designed to handle the overwhelming complexity 00:25:25.559 --> 00:25:28.039 of modern chips. By distributing the simulation 00:25:28.039 --> 00:25:30.380 workload across multiple processors simultaneously, 00:25:31.039 --> 00:25:33.599 it allows companies to shave months off verification 00:25:33.599 --> 00:25:37.039 schedules. And time to market is everything in 00:25:37.039 --> 00:25:39.329 the semiconductor industry. Verification isn't 00:25:39.329 --> 00:25:41.769 just about software simulation, though. For designs 00:25:41.769 --> 00:25:44.309 involving billions of gates, you need dedicated 00:25:44.309 --> 00:25:47.009 hardware. Which brings us back to that emulation 00:25:47.009 --> 00:25:50.230 technology, which, as we discussed, was so foundational 00:25:50.230 --> 00:25:53.009 to their early history via the QuickTurn acquisition. 00:25:53.470 --> 00:25:55.589 And this is where the hardware emulation and 00:25:55.589 --> 00:25:58.089 prototyping platforms Palladium and Prophium 00:25:58.089 --> 00:26:01.190 come in. Right. Palladium is the hardware emulation 00:26:01.190 --> 00:26:03.829 platform, a machine the size of a refrigerator 00:26:03.829 --> 00:26:06.349 that is essentially a specialized supercomputer 00:26:06.349 --> 00:26:08.950 designed specifically specifically for verifying 00:26:08.950 --> 00:26:12.869 massive billion -gate designs very quickly. The 00:26:12.869 --> 00:26:16.190 successor, Z2, announced in 2021, offered improved 00:26:16.190 --> 00:26:18.990 performance and capacity, keeping pace with the 00:26:18.990 --> 00:26:21.230 exponential growth of chips. And Protium's the 00:26:21.230 --> 00:26:23.470 complementary piece of hardware, isn't it? Correct. 00:26:24.170 --> 00:26:27.089 Complementing Palladium is Protium, the FPGA 00:26:27.089 --> 00:26:30.740 prototyping platform. Protium is usually faster 00:26:30.740 --> 00:26:33.000 and cheaper to operate than the Palladium emulator, 00:26:33.200 --> 00:26:35.779 and it allows designers to run massive amounts 00:26:35.779 --> 00:26:38.240 of application software on the hardware before 00:26:38.240 --> 00:26:40.599 manufacturing. Which lets software developers 00:26:40.599 --> 00:26:43.039 start testing their code much, much earlier. 00:26:43.259 --> 00:26:45.460 A huge advantage. But wait, for the listener 00:26:45.460 --> 00:26:48.339 who isn't familiar with this, why is having a 00:26:48.339 --> 00:26:51.640 single compilation flow between two very different 00:26:51.640 --> 00:26:54.799 hardware platforms, Palladium and Protium so 00:26:54.799 --> 00:26:57.319 revolutionary for verification engineers? What 00:26:57.319 --> 00:26:59.559 headache does it actually... eliminate. That 00:26:59.559 --> 00:27:02.759 single compilation flow is huge because compilation 00:27:02.759 --> 00:27:05.380 is the hardest part. It can take days or even 00:27:05.380 --> 00:27:07.839 weeks to translate a chip design into the specific 00:27:07.839 --> 00:27:11.220 format needed for an emulator or an FPGA prototype. 00:27:11.559 --> 00:27:13.700 So if you have two different platforms, you often 00:27:13.700 --> 00:27:15.579 have to maintain two separate tool chains and 00:27:15.579 --> 00:27:17.480 run two separate time -consuming compilation 00:27:17.480 --> 00:27:20.480 efforts. By sharing a single flow, Cadence ensures 00:27:20.480 --> 00:27:22.480 that once the design is compiled for Palladium, 00:27:22.619 --> 00:27:26.829 it can be seamlessly used on Procium, Which minimizes 00:27:26.829 --> 00:27:29.069 errors, saves massive amounts of engineering 00:27:29.069 --> 00:27:32.019 time. and dramatically accelerates the entire 00:27:32.019 --> 00:27:34.380 hardware -software co -verification process. 00:27:34.799 --> 00:27:37.599 It provides an uninterrupted, highly efficient 00:27:37.599 --> 00:27:40.940 environment for the industry's most complex Sox 00:27:40.940 --> 00:27:43.859 Procium X2, for example, claimed to handle capacity 00:27:43.859 --> 00:27:47.720 exceeding that of 1 .2 billion gate Sox. Okay, 00:27:47.759 --> 00:27:51.480 moving now to the SIP blocks, the Silicon Intellectual 00:27:51.480 --> 00:27:53.880 Property. This is what I called the secret sauce 00:27:53.880 --> 00:27:56.799 inside your phone, a very strategic revenue stream. 00:27:57.019 --> 00:27:59.730 It's high -value IP that cadence licenses. and 00:27:59.730 --> 00:28:02.569 it saves customers massive amounts of R &D time 00:28:02.569 --> 00:28:05.170 while guaranteeing high performance. The most 00:28:05.170 --> 00:28:07.829 famous example is the Tensilica DSP processors, 00:28:08.130 --> 00:28:10.990 which they acquired in 2013 for $380 million. 00:28:11.410 --> 00:28:14.049 And these are specialized processor cores designed 00:28:14.049 --> 00:28:17.089 for specific high -performance tasks that a general 00:28:17.089 --> 00:28:20.170 CPU core handles poorly. They do. So why is licensing 00:28:20.170 --> 00:28:22.769 specialized IP like a Tensilica core superior 00:28:22.769 --> 00:28:25.309 to a customer just designing it internally? Well, 00:28:25.309 --> 00:28:27.710 there are three huge advantages. First, R &D 00:28:27.710 --> 00:28:30.349 time. Designing a cutting -edge DSP takes years 00:28:30.349 --> 00:28:32.769 and costs hundreds of millions. Licensing saves 00:28:32.769 --> 00:28:34.569 all of that. Second, guaranteed performance. 00:28:35.009 --> 00:28:38.269 Exactly. Cadence has spent years optimizing these 00:28:38.269 --> 00:28:40.470 cores for extremely low power and high speed 00:28:40.470 --> 00:28:43.809 for a specific function like audio or vision. 00:28:44.009 --> 00:28:47.650 And third, the ecosystem. Consilica cores come 00:28:47.650 --> 00:28:50.430 with a stable software tools, drivers, and debugging 00:28:50.430 --> 00:28:52.900 support. which accelerates the customer's time 00:28:52.900 --> 00:28:55.019 to market dramatically. It's like buying a high 00:28:55.019 --> 00:28:57.039 -performance engine for your car, rather than 00:28:57.039 --> 00:28:58.880 trying to invent the engine yourself. And that 00:28:58.880 --> 00:29:01.359 engine is used everywhere. Can you run down some 00:29:01.359 --> 00:29:03.339 of the specific applications of these specialized 00:29:03.339 --> 00:29:06.380 ESPs? Sure. They are deeply specialized. For 00:29:06.380 --> 00:29:09.500 example, Tensilica Vision DSPs are used extensively 00:29:09.500 --> 00:29:12.019 for imaging, computer vision, and AI processing 00:29:12.019 --> 00:29:15.460 in edge devices. Essential in modern smartphones, 00:29:15.819 --> 00:29:18.299 drones, automotive sensing applications where 00:29:18.299 --> 00:29:21.019 latency is critical. And the Hi -Fi DSPs. The 00:29:21.019 --> 00:29:23.319 Tensilica Hi -Fi DSPs are the industry standard 00:29:23.319 --> 00:29:26.059 for high -quality, low -power audio processing. 00:29:26.339 --> 00:29:28.180 You'll find them in everything from wireless 00:29:28.180 --> 00:29:31.420 earbuds to smart TVs. They process complex audio 00:29:31.420 --> 00:29:33.920 codecs far more efficiently than a generic... 00:29:33.900 --> 00:29:37.599 CPU. And there are others. Many others. Tensilica 00:29:37.599 --> 00:29:41.539 Fusion DSPs for low -power, connected IoT applications. 00:29:42.259 --> 00:29:45.339 Tensilica Connex DSPs for extremely high -speed 00:29:45.339 --> 00:29:48.880 communications, processing radar, LIAR, 5G base 00:29:48.880 --> 00:29:52.039 stations. And the Tensilica DNA Processor family, 00:29:52.240 --> 00:29:55.180 their most recent addition, designed purely for 00:29:55.180 --> 00:29:58.269 extreme AI acceleration at the chip level. They 00:29:58.269 --> 00:30:00.829 clearly recognize the coming AI wave very early 00:30:00.829 --> 00:30:03.329 on. You can see that with the 2021 launch of 00:30:03.329 --> 00:30:06.009 the Tensilica AI platform, designed specifically 00:30:06.009 --> 00:30:08.569 to integrate and accelerate AI development into 00:30:08.569 --> 00:30:10.910 their customers' system -on -chip designs. And 00:30:10.910 --> 00:30:12.809 that leads directly to the core of their future 00:30:12.809 --> 00:30:16.630 strategy, the AI design revolution itself. Cadence 00:30:16.630 --> 00:30:18.740 is working on a dual front. First, providing 00:30:18.740 --> 00:30:21.200 tools to design AI chips. And second, adding 00:30:21.200 --> 00:30:23.680 AI into its own software to optimize the design 00:30:23.680 --> 00:30:26.079 process itself. So they're using AI to build 00:30:26.079 --> 00:30:28.359 better chips and using AI to build the software 00:30:28.359 --> 00:30:30.279 that builds those chips. You got it. The first 00:30:30.279 --> 00:30:32.559 major release here, focusing on the latter, was 00:30:32.559 --> 00:30:35.859 Cerebrus in 2021. Cerebrus is a profound example 00:30:35.859 --> 00:30:38.859 of Cabin's shifting its engineering focus. It's 00:30:38.859 --> 00:30:40.519 machine learning -based chip design software 00:30:40.519 --> 00:30:42.920 that uses reinforcement learning. The same technique 00:30:42.920 --> 00:30:45.960 used to train AIs to play complex games. Right. 00:30:46.079 --> 00:30:48.880 It uses that to automatically optimize the digital 00:30:48.880 --> 00:30:51.619 design flow. Instead of a human engineer manually 00:30:51.619 --> 00:30:54.299 tweaking parameters for weeks to get a 10 % performance 00:30:54.299 --> 00:30:57.200 gain, Cerebrus can iterate thousands of times 00:30:57.200 --> 00:31:00.799 in hours, often achieving 20 or 30 % better results 00:31:00.799 --> 00:31:04.220 in terms of power, performance, and area optimization. 00:31:04.660 --> 00:31:07.200 It radically changes the economics and speed 00:31:07.200 --> 00:31:10.259 of chip design. It does. Then came the Optimality 00:31:10.259 --> 00:31:14.140 Intelligence System Explorer in 2022. This moves 00:31:14.140 --> 00:31:17.079 the AI focus outward, from the chip to the entire 00:31:17.079 --> 00:31:20.359 system. Exactly. This tool integrates the multiphysics 00:31:20.359 --> 00:31:26.200 analysis for electromagnetics and Sigridiak's 00:31:26.200 --> 00:31:29.319 analysis for power integrity, and uses AI to 00:31:29.319 --> 00:31:31.519 optimize the entire system design at the board, 00:31:31.660 --> 00:31:33.920 package, and chassis level. It's designed to 00:31:33.920 --> 00:31:35.900 allow engineers to explore millions of design 00:31:35.900 --> 00:31:38.740 variations simultaneously. Microsoft was an early 00:31:38.740 --> 00:31:41.759 adopter, recognizing the necessity of this system 00:31:41.759 --> 00:31:44.519 -level optimization for their complex data center 00:31:44.519 --> 00:31:48.019 infrastructure. And then in 2023, they introduced 00:31:48.019 --> 00:31:50.819 what seems like the ultimate automation tool 00:31:50.819 --> 00:31:55.299 for specification to silicon creation, ChipGPT. 00:31:55.710 --> 00:31:58.329 That name really encapsulates the trend, doesn't 00:31:58.329 --> 00:32:01.269 it? This software uses generative AI and large 00:32:01.269 --> 00:32:03.349 language models to allow companies to create 00:32:03.349 --> 00:32:06.130 custom silicon designs starting from high -level 00:32:06.130 --> 00:32:09.529 natural language or simple specifications. ChipGPT 00:32:09.529 --> 00:32:12.470 streamlines the entire pipeline, automating much 00:32:12.470 --> 00:32:15.210 of the tedious front -end design work. And that 00:32:15.210 --> 00:32:19.349 trio of tools, Cerebrus, Optimality, and ChipGPT, 00:32:19.470 --> 00:32:22.730 it shows Cadence's strategy very clearly, leveraging 00:32:22.730 --> 00:32:25.009 AI at every single layer to make the design. 00:32:31.279 --> 00:32:33.359 We've talked about the impressive technology 00:32:33.359 --> 00:32:35.640 and the massive growth, but this wouldn't be 00:32:35.640 --> 00:32:37.579 the deep dive without revealing the high stakes 00:32:37.579 --> 00:32:39.819 drama and corporate intrigue that have defined 00:32:39.819 --> 00:32:42.079 Cadence's history? Oh, absolutely not. And these 00:32:42.079 --> 00:32:44.200 events often sound less like business competition 00:32:44.200 --> 00:32:46.859 and more like a corporate thriller. They certainly 00:32:46.859 --> 00:32:49.579 do. And the most infamous chapter in their history. 00:32:49.930 --> 00:32:52.730 is the six -year legal dispute with Avanti Corporation, 00:32:53.230 --> 00:32:57.390 which spanned from 1995 until 2002 and involved 00:32:57.390 --> 00:32:59.329 criminal charges against the competitor's leadership. 00:32:59.839 --> 00:33:02.539 Cadence accused Avanti of what really amounted 00:33:02.539 --> 00:33:04.859 to industrial espionage staining core design 00:33:04.859 --> 00:33:07.980 code for their flagship implementation tools. 00:33:08.240 --> 00:33:10.480 This wasn't just a simple patent spat over features, 00:33:10.599 --> 00:33:13.119 was it? No. Businessweek was quoted describing 00:33:13.119 --> 00:33:16.240 the Avanti case as, and I quote, probably the 00:33:16.240 --> 00:33:18.500 most dramatic tale of white -collar crime in 00:33:18.500 --> 00:33:21.039 the history of Silicon Valley. Can you provide 00:33:21.039 --> 00:33:23.339 some color on how this all unfolded? It was a 00:33:23.339 --> 00:33:26.359 huge scandal. The allegations centered on key 00:33:26.359 --> 00:33:28.799 Avanti executives, including his founder, Gerald 00:33:28.799 --> 00:33:31.940 Hsu, who had allegedly hired former Cadence engineers. 00:33:32.259 --> 00:33:35.240 And Cadence claimed, and later proved in court, 00:33:35.400 --> 00:33:37.339 that these employees had illicitly transferred 00:33:37.339 --> 00:33:40.119 Cadence source code specifically for placement 00:33:40.119 --> 00:33:42.079 and routing tools, those tools that determine 00:33:42.079 --> 00:33:45.220 how transistors are physically laid out to Avanti. 00:33:45.259 --> 00:33:47.259 Who then rebranded and sold them as their own. 00:33:47.400 --> 00:33:50.299 Exactly. Hugely undercutting Cadence's development 00:33:50.299 --> 00:33:52.839 costs and their market position. Well, that explains 00:33:52.839 --> 00:33:56.220 the aggression. I mean, if your competitor bypasses 00:33:56.220 --> 00:33:58.940 billions in R &D via theft, it threatens your 00:33:58.940 --> 00:34:01.880 entire business model. Precisely. The legal battle 00:34:01.880 --> 00:34:04.579 was intense. It involved both civil lawsuits 00:34:04.579 --> 00:34:07.619 and an unprecedented criminal prosecution by 00:34:07.619 --> 00:34:09.539 the Santa Clara County District Attorney. And 00:34:09.539 --> 00:34:12.849 the outcome was stark. It was. After years of 00:34:12.849 --> 00:34:15.329 legal maneuvering, the Avanti executives eventually 00:34:15.329 --> 00:34:18.469 pleaded no contest to the criminal charges, which 00:34:18.469 --> 00:34:20.809 is a move often interpreted as a practical admission 00:34:20.809 --> 00:34:23.369 of guilt without formally submitting to a trial. 00:34:23.469 --> 00:34:25.829 And Cadence received several hundred million 00:34:25.829 --> 00:34:28.699 dollars in restitution and damages. Right. The 00:34:28.699 --> 00:34:31.659 whole saga closed in 2002 when Avanti was purchased 00:34:31.659 --> 00:34:35.260 by their other major rival, Synopsys. But even 00:34:35.260 --> 00:34:38.119 Synopsys had to pay an additional $265 million 00:34:38.119 --> 00:34:41.000 to settle the remaining legal claims Cadence 00:34:41.000 --> 00:34:43.659 had against the acquired Avanti. The entire episode 00:34:43.659 --> 00:34:46.800 just solidified Cadence's reputation for vigorously 00:34:46.800 --> 00:34:49.159 defending its intellectual property. To put it 00:34:49.159 --> 00:34:51.780 mildly. But the drama didn't end there. There 00:34:51.780 --> 00:34:54.360 was another massive legal battle, equally bizarre, 00:34:54.659 --> 00:34:57.440 involving a company they acquired, Quickturn, 00:34:57.699 --> 00:35:00.599 and their rivals Mentor Graphics and Aptix Corporation. 00:35:01.099 --> 00:35:04.929 This specific legal saga. truly borders on unbelievable. 00:35:05.269 --> 00:35:07.989 It moves from intellectual property law into 00:35:07.989 --> 00:35:10.809 the realm of true crime. So Aptix, a small company, 00:35:11.010 --> 00:35:13.570 licensed a patent to Mentor Graphics and they 00:35:13.570 --> 00:35:15.869 jointly sued Quick Turn, which was now Cadence 00:35:15.869 --> 00:35:18.369 owned, for patent infringement related to emulation 00:35:18.369 --> 00:35:21.010 technology. Right. But during the discovery process, 00:35:21.170 --> 00:35:23.289 the integrity of the evidence presented by Aptix 00:35:23.289 --> 00:35:27.300 came under intense scrutiny. The Aptix CEO, Amr 00:35:27.300 --> 00:35:29.679 Mohsen, was subsequently charged with conspiracy, 00:35:30.139 --> 00:35:32.760 perjury, and obstruction of justice for allegedly 00:35:32.760 --> 00:35:35.099 forging and tampering with key legal evidence. 00:35:35.239 --> 00:35:37.699 And then it just escalated wildly, right? Far 00:35:37.699 --> 00:35:40.980 beyond simple corporate misconduct. It did. I 00:35:40.980 --> 00:35:43.559 mean, Mohsen was arrested trying to flee the 00:35:43.559 --> 00:35:45.940 country while he was out on bail. But the situation 00:35:45.940 --> 00:35:48.360 got so serious that while he was awaiting trial 00:35:48.360 --> 00:35:50.619 in jail... Wait, so he's in jail at this point? 00:35:50.800 --> 00:35:53.820 He's in jail. And he's further charged with plotting 00:35:53.820 --> 00:35:57.019 to intimidate witnesses. And this is just incredible. 00:35:57.639 --> 00:35:59.900 Attempting to arrange for the murder of the federal 00:35:59.900 --> 00:36:02.679 judge presiding over his case. The judge. He 00:36:02.679 --> 00:36:04.699 tried to kill the judge. I mean, I think we have 00:36:04.699 --> 00:36:07.500 to pause there. An EDA company CEO charged with 00:36:07.500 --> 00:36:10.340 plotting murder. That that really puts the stakes 00:36:10.340 --> 00:36:12.099 in this market into a whole new perspective. 00:36:12.460 --> 00:36:14.880 Absolutely. It became an extraordinary case where 00:36:14.880 --> 00:36:18.039 severe criminal misconduct completely derailed 00:36:18.039 --> 00:36:20.179 the civil litigation. Due to the overwhelming 00:36:20.179 --> 00:36:22.420 evidence of evidence tampering and obstruction, 00:36:22.760 --> 00:36:26.260 the judge ruled the original patent lawsuit unenforceable. 00:36:26.320 --> 00:36:28.619 And Mosen was eventually sentenced to 17 years 00:36:28.619 --> 00:36:31.059 in prison. Which is a rare outcome for a technology 00:36:31.059 --> 00:36:34.119 executive. Following this, Mentor Graphics and 00:36:34.119 --> 00:36:36.780 Cadence both sued Aptics and each other to recover 00:36:36.780 --> 00:36:39.380 their substantial legal costs incurred during 00:36:39.380 --> 00:36:41.860 this bizarre, drawn -out affair. It just illustrates 00:36:41.860 --> 00:36:44.000 the fiercely competitive, sometimes surprisingly 00:36:44.000 --> 00:36:46.739 ruthless, environment of the early EDA market. 00:36:47.190 --> 00:36:49.170 And finally, we should just briefly touch on 00:36:49.170 --> 00:36:52.130 the 2013 lawsuit against Berkeley Design Automation, 00:36:52.130 --> 00:36:55.070 or BDA. This was less criminal but highly strategic. 00:36:55.530 --> 00:36:58.329 That case focused on BDA's circumvention of a 00:36:58.329 --> 00:37:00.389 license scheme that was required to link their 00:37:00.389 --> 00:37:03.909 analog fast -spice simulator to Cadence's Virtuoso 00:37:03.909 --> 00:37:07.010 ADE, the core IC design platform we discussed 00:37:07.010 --> 00:37:09.789 earlier. That essential interoperability. Right. 00:37:10.139 --> 00:37:12.860 It settled less than a year later, with BDA making 00:37:12.860 --> 00:37:14.960 an undisclosed payment and agreeing to stop the 00:37:14.960 --> 00:37:17.780 circumvention, instead supporting interoperability 00:37:17.780 --> 00:37:21.000 through Cadence's official interface. It again 00:37:21.000 --> 00:37:23.340 just underscores Cadence's aggressive defense 00:37:23.340 --> 00:37:25.760 of its core intellectual property ecosystem, 00:37:26.079 --> 00:37:28.519 where controlling the interface can be as important 00:37:28.519 --> 00:37:31.360 as controlling the core technology. Now, we need 00:37:31.360 --> 00:37:33.760 to address a much more serious and recent compliance 00:37:33.760 --> 00:37:36.780 challenge, the export control violations. And 00:37:36.780 --> 00:37:38.659 this is the bigger picture concerning global 00:37:38.659 --> 00:37:41.440 security and technology transfer. And it carried 00:37:41.440 --> 00:37:43.900 a massive financial and legal penalty. This came 00:37:43.900 --> 00:37:47.360 to a head in July of 2025. Cadence Design Systems 00:37:47.360 --> 00:37:50.280 agreed to plead guilty to criminal charges, not 00:37:50.280 --> 00:37:53.460 just a civil fine, and pay over $140 million 00:37:53.460 --> 00:37:56.260 in combined penalties for illegally exporting 00:37:56.260 --> 00:37:59.239 high -tech semiconductor design technology. That 00:37:59.239 --> 00:38:01.139 is a significant marker of the gravity of the 00:38:01.139 --> 00:38:03.639 offense. It is. The offense involved their Chinese 00:38:03.639 --> 00:38:06.980 subsidiary. knowingly selling electronic design, 00:38:07.059 --> 00:38:09.679 automation hardware, software, and semiconductor 00:38:09.679 --> 00:38:13.159 design technology valued at over $45 million 00:38:13.159 --> 00:38:16.219 to a highly restricted entity. China's National 00:38:16.219 --> 00:38:19.659 University of Defense Technology, NUDT, a military 00:38:19.659 --> 00:38:21.480 university controlled by the Central Military 00:38:21.480 --> 00:38:24.320 Commission. And this wasn't accidental or an 00:38:24.320 --> 00:38:27.119 unknown customer, was it? NUDT was placed on 00:38:27.119 --> 00:38:29.920 the U .S. entity list way back in February 2015. 00:38:30.360 --> 00:38:32.599 And why were they restricted? They were restricted 00:38:32.599 --> 00:38:35.280 precisely because of their use of American technology 00:38:35.280 --> 00:38:39.239 for supercomputers. NUDT's supercomputing capabilities 00:38:39.239 --> 00:38:42.840 are believed to support nuclear explosive simulation 00:38:42.840 --> 00:38:46.179 and military simulation activities. So the technology 00:38:46.179 --> 00:38:48.860 Cadence illegally provided advanced EDA tools 00:38:48.860 --> 00:38:51.880 for designing chips is absolutely critical for 00:38:51.880 --> 00:38:53.920 developing the cutting -edge microprocessors 00:38:53.920 --> 00:38:56.300 needed to enhance military supercomputing power. 00:38:56.579 --> 00:38:58.679 It was a direct contribution to restricted military 00:38:58.679 --> 00:39:01.239 technology development. And the violations spanned 00:39:01.239 --> 00:39:03.940 a long time, six years, between 2015 and 2021. 00:39:04.400 --> 00:39:06.840 The source document suggests the tactics used 00:39:06.840 --> 00:39:09.500 by Cadence's Chinese subsidiary were intentional. 00:39:09.769 --> 00:39:12.730 And that's why the penalties were so severe. 00:39:12.889 --> 00:39:16.150 The tactics used were explicitly designed to 00:39:16.150 --> 00:39:18.489 conceal the illegal exports from U .S. regulators. 00:39:18.869 --> 00:39:21.650 They used intermediary companies, specifically 00:39:21.650 --> 00:39:25.090 Central South Cad Center, or CSCC, and later 00:39:25.090 --> 00:39:27.889 Fidium Technology, to disguise the final end 00:39:27.889 --> 00:39:30.739 user of the controlled technology. Masking the 00:39:30.739 --> 00:39:33.880 fact that NUDT was the ultimate beneficiary. 00:39:33.960 --> 00:39:35.920 Exactly. And the internal communications you 00:39:35.920 --> 00:39:38.139 mentioned earlier are particularly damning. They 00:39:38.139 --> 00:39:40.780 show an internal awareness that what they were 00:39:40.780 --> 00:39:43.480 doing was illegal and required cover. Indeed. 00:39:43.920 --> 00:39:46.079 Internal communications instructed employees 00:39:46.079 --> 00:39:49.500 to refer to NUDT only in Chinese characters and 00:39:49.500 --> 00:39:52.000 to use CSCC in English correspondence because, 00:39:52.239 --> 00:39:55.039 and I quote, the subject was too sensitive. which 00:39:55.039 --> 00:39:57.340 strongly suggests a deliberate corporate policy 00:39:57.340 --> 00:40:00.519 to evade U .S. export controls. And to monetize 00:40:00.519 --> 00:40:02.239 advanced technology that the U .S. government 00:40:02.239 --> 00:40:05.000 had deemed crucial to national security and prohibited 00:40:05.000 --> 00:40:07.239 from reaching NUDT. So what were the specific 00:40:07.239 --> 00:40:10.639 consequences for the company beyond the $140 00:40:10.639 --> 00:40:13.440 million penalty? Well, as a consequence of pleading 00:40:13.440 --> 00:40:16.559 guilty, the Department of Justice imposed significant 00:40:16.559 --> 00:40:19.519 compliance requirements. Cadence was placed on 00:40:19.519 --> 00:40:21.469 three years of corporate probation. And this 00:40:21.469 --> 00:40:24.849 is a key detail. It is. They received only partial 00:40:24.849 --> 00:40:27.690 cooperation credit during the investigation because 00:40:27.690 --> 00:40:30.650 they failed to voluntarily disclose the misconduct 00:40:30.650 --> 00:40:33.989 initially and did not fully facilitate interviews 00:40:33.989 --> 00:40:36.969 of their China -based employees. And this case 00:40:36.969 --> 00:40:39.489 really highlights the profound risks and legal 00:40:39.489 --> 00:40:42.449 exposures facing all computational software companies 00:40:42.449 --> 00:40:44.929 operating globally when their technology has 00:40:44.929 --> 00:40:47.789 dual use or military applications. It's a constant 00:40:47.789 --> 00:40:50.150 tightrope walk between globalization and national 00:40:50.150 --> 00:40:53.239 security. Wow. From foundational mergers and 00:40:53.239 --> 00:40:55.179 corporate espionage straight out of a novel, 00:40:55.280 --> 00:40:58.460 to designing drug molecules, building supercomputers, 00:40:58.480 --> 00:41:01.860 and facing $140 million in federal penalties 00:41:01.860 --> 00:41:05.099 for global security violations. Cadence is truly 00:41:05.099 --> 00:41:07.559 a multifaceted company that shapes the modern 00:41:07.559 --> 00:41:10.159 world in ways we rarely appreciate. To summarize 00:41:10.159 --> 00:41:12.800 the key takeaways, Cadence has successfully executed 00:41:12.800 --> 00:41:15.039 a comprehensive, though sometimes controversial, 00:41:15.360 --> 00:41:18.059 transformation. They started as the essential 00:41:18.059 --> 00:41:21.760 EDA pioneer. providing the indispensable Virtuoso, 00:41:22.000 --> 00:41:24.920 Spectre, and Palladium tools. But they have now 00:41:24.920 --> 00:41:27.679 evolved into a diversified computational giant. 00:41:27.940 --> 00:41:30.380 We've seen them move into deep physics simulation 00:41:30.380 --> 00:41:33.340 through acquisitions like Numeca and PointWise, 00:41:33.340 --> 00:41:36.519 establishing capabilities in CFD and multi -physics. 00:41:36.679 --> 00:41:38.900 They moved into advanced chemical and drug discovery 00:41:38.900 --> 00:41:41.460 with OpenEye. leveraging computational rigor 00:41:41.460 --> 00:41:43.820 developed in silicon. And they solidified their 00:41:43.820 --> 00:41:46.360 system -level prowess with the colossal acquisition 00:41:46.360 --> 00:41:49.559 of Hexagon's design and engineering unit, positioning 00:41:49.559 --> 00:41:52.000 them to simulate the entire industrial lifecycle. 00:41:52.420 --> 00:41:54.900 And all of this is underpinned by their strategic, 00:41:55.079 --> 00:41:58.000 pervasive shift towards AI -enhanced design tools 00:41:58.000 --> 00:42:01.920 like Cerebrus and ChipGBT. So what, for you, 00:42:01.960 --> 00:42:04.820 the learner, is this? Cadence illustrates a fundamental 00:42:04.820 --> 00:42:07.219 industrial truth. That the core technology required 00:42:07.219 --> 00:42:09.599 to design cutting -edge semiconductors, namely 00:42:09.599 --> 00:42:11.880 computational modeling, simulation, and verification, 00:42:12.179 --> 00:42:15.139 is not limited to electronics. That expertise 00:42:15.139 --> 00:42:17.940 is now the fundamental engine being leveraged 00:42:17.940 --> 00:42:20.860 to solve complex multi -physics problems across 00:42:20.860 --> 00:42:23.820 entirely different industries. Whether it's optimizing 00:42:23.820 --> 00:42:26.400 the behavior of a new drug molecule, ensuring 00:42:26.400 --> 00:42:29.039 the thermal performance of a massive data center, 00:42:29.179 --> 00:42:32.099 or modeling the fluid dynamics of a Formula One 00:42:32.099 --> 00:42:34.599 car. as we've seen from their multi -year partnership 00:42:34.599 --> 00:42:36.960 with McLaren Racing. Computational software, 00:42:37.260 --> 00:42:40.420 now enhanced by AI, is becoming the new foundational 00:42:40.420 --> 00:42:43.239 layer of industrial innovation. They are selling 00:42:43.239 --> 00:42:45.380 the digital blueprints for physical reality. 00:42:45.619 --> 00:42:48.719 And Cadence's ongoing aggressive expansion, particularly 00:42:48.719 --> 00:42:51.320 the acquisition of Hexagon's unit for over $3 00:42:51.320 --> 00:42:54.219 billion, suggests a future where computational 00:42:54.219 --> 00:42:57.179 software companies will compete not just on designing 00:42:57.179 --> 00:42:59.739 microchips, but on simulating the entire physical 00:42:59.739 --> 00:43:01.920 world around those chips, integrating electronics, 00:43:02.059 --> 00:43:04.739 mechanics, and chemistry into unified digital 00:43:04.739 --> 00:43:07.719 models. Which raises our final provocative thought 00:43:07.719 --> 00:43:10.880 for you to consider. Cadence has moved from electronics 00:43:10.880 --> 00:43:14.239 into physics and pharma, leveraging the transferability 00:43:14.239 --> 00:43:17.139 of simulation expertise. Given the demonstrated 00:43:17.139 --> 00:43:19.739 power of AI -powered simulation in these diverse 00:43:19.739 --> 00:43:23.159 fields, what entirely new industries will CDNS, 00:43:23.280 --> 00:43:26.880 or their rivals like Synopsys, absorb or fundamentally 00:43:26.880 --> 00:43:29.460 transform next using this foundational computational 00:43:29.460 --> 00:43:32.019 power? Will we see them move into environmental 00:43:32.019 --> 00:43:34.659 modeling, climate science, or perhaps even urban 00:43:34.659 --> 00:43:36.519 planning? That's the question that will define 00:43:36.519 --> 00:43:38.300 the computational giants of the next decade.