WEBVTT 00:00:00.000 --> 00:00:02.500 Okay, let's unpack this. We are diving into a 00:00:02.500 --> 00:00:06.960 company that has fundamentally reshaped its identity 00:00:06.960 --> 00:00:09.039 over the last decade. Oh, absolutely. It's gone 00:00:09.039 --> 00:00:12.000 from a core semiconductor component supplier 00:00:12.000 --> 00:00:16.019 to, well, a hybrid trillion dollar powerhouse. 00:00:16.120 --> 00:00:18.800 Right. We're talking about Broadcom Inc. And 00:00:18.800 --> 00:00:21.940 when you look at Broadcom today, you are really 00:00:21.940 --> 00:00:25.579 observing a company that has mastered one of 00:00:25.579 --> 00:00:29.199 the most... I mean, high stakes corporate growth 00:00:29.199 --> 00:00:32.179 strategies in modern history. Sure. It's a story 00:00:32.179 --> 00:00:36.039 of meticulous M &A mergers and acquisitions orchestrated 00:00:36.039 --> 00:00:39.039 to deliver, you know, highly reliable, high margin 00:00:39.039 --> 00:00:41.479 revenue. And not just from hardware. Not just 00:00:41.479 --> 00:00:43.399 from the hardware they sell. No. Right. But from 00:00:43.399 --> 00:00:45.479 the essential software that, you know, runs the 00:00:45.479 --> 00:00:48.000 global digital economy. And the scale of this 00:00:48.000 --> 00:00:50.899 achievement is just massive, which immediately 00:00:50.899 --> 00:00:53.380 establishes the relevance of this deep dive for 00:00:53.380 --> 00:00:55.780 you, the learner. It does. The market perception 00:00:55.780 --> 00:00:58.119 has shifted so dramatically that as of 2025, 00:00:58.420 --> 00:01:01.520 right in the middle of this AI boom, Broadcom 00:01:01.520 --> 00:01:04.239 isn't just a big tech player. It has officially 00:01:04.239 --> 00:01:07.099 been elevated to the exclusive Magnificent Seven 00:01:07.099 --> 00:01:09.319 group. Replacing Tesla. Of all companies. Yeah, 00:01:09.340 --> 00:01:12.000 replacing Tesla. That kind of market reassessment 00:01:12.000 --> 00:01:15.000 is, well, it's rare. It really speaks volumes 00:01:15.000 --> 00:01:17.519 about where the true value lies in the infrastructure 00:01:17.519 --> 00:01:21.120 that's powering AI. Right. And the financial 00:01:21.120 --> 00:01:24.680 reality caught up. In December 2024, Broadcom 00:01:24.680 --> 00:01:27.840 became the 12th company ever to surpass a $1 00:01:27.840 --> 00:01:31.620 trillion market capitalization. A trillion. And 00:01:31.620 --> 00:01:35.299 as of September 2025, it remains the most recent 00:01:35.299 --> 00:01:37.900 entrant. to hit that milestone. And you have 00:01:37.900 --> 00:01:39.799 to appreciate that feat. I mean, you have to 00:01:39.799 --> 00:01:41.780 remember, this is a company that a decade ago 00:01:41.780 --> 00:01:44.439 was still focused on specialized components and 00:01:44.439 --> 00:01:46.299 was headquartered in Singapore. Which is a huge 00:01:46.299 --> 00:01:48.480 part of the story we have to get into. A huge 00:01:48.480 --> 00:01:50.959 part. So our mission today is to trace this incredibly 00:01:50.959 --> 00:01:54.219 complex, decades -long journey. We need to start 00:01:54.219 --> 00:01:57.879 from its very quiet roots in 1961 through the 00:01:57.879 --> 00:02:01.359 Avago era, the intense political battles in Washington, 00:02:01.519 --> 00:02:03.939 D .C., and then that subsequent highly calculated 00:02:03.939 --> 00:02:06.650 pivot into software. And our sources really paint 00:02:06.650 --> 00:02:08.550 a picture of a company driven by a relentless, 00:02:08.750 --> 00:02:11.750 disciplined M &A playbook. Executed by CEO Hochtan. 00:02:11.889 --> 00:02:14.949 And we absolutely have to understand the current 00:02:14.949 --> 00:02:17.270 business split because that is the key to this 00:02:17.270 --> 00:02:19.889 trillion dollar valuation. It's everything. The 00:02:19.889 --> 00:02:22.330 Broadcom Inc. of today is defined by its hybrid 00:02:22.330 --> 00:02:25.389 nature. Approximately 58 percent of their revenue 00:02:25.389 --> 00:02:27.990 still comes from their core semiconductor based 00:02:27.990 --> 00:02:29.789 products. The chips we'll get into later. What 00:02:29.789 --> 00:02:31.909 the other side of the coin. But a massive 42 00:02:31.909 --> 00:02:34.389 percent is now derived from infrastructure software. 00:02:34.509 --> 00:02:37.810 products and services. That 42 % software chunk 00:02:37.810 --> 00:02:41.150 is the foundation of the high margins and the 00:02:41.150 --> 00:02:42.909 recurring revenue that Wall Street just loves. 00:02:43.090 --> 00:02:45.650 Okay. So let's unpack the origins and see where 00:02:45.650 --> 00:02:49.229 this disciplined approach to M &A was born. So 00:02:49.229 --> 00:02:52.090 the journey to becoming a modern big tech powerhouse, 00:02:52.270 --> 00:02:54.990 it starts surprisingly in the early days of Silicon 00:02:54.990 --> 00:02:57.530 Valley. What were the first footsteps of what 00:02:57.530 --> 00:02:59.650 would eventually become Broadcom? Yeah, we have 00:02:59.650 --> 00:03:02.719 to go way back to 1961. The current corporate 00:03:02.719 --> 00:03:05.400 structure, it traces its roots to HP Associates. 00:03:05.500 --> 00:03:08.259 Hewlett Packard, yes. It was established as their 00:03:08.259 --> 00:03:10.879 semiconductor products division. So you have 00:03:10.879 --> 00:03:12.979 to think about the legacy there. This division 00:03:12.979 --> 00:03:17.300 was steeped in the pioneering DNA of HP, focusing 00:03:17.300 --> 00:03:20.280 on components and specialized technologies. That 00:03:20.280 --> 00:03:22.659 was the incubator for its initial engineering 00:03:22.659 --> 00:03:25.469 prowess. But the HP umbrella wouldn't hold it 00:03:25.469 --> 00:03:28.710 forever. The first major separation happens near 00:03:28.710 --> 00:03:31.330 the turn of the millennium. Precisely. In 1999, 00:03:31.590 --> 00:03:34.009 as part of a massive corporate shakeup at HP, 00:03:34.250 --> 00:03:36.569 the semiconductor division separated to become 00:03:36.569 --> 00:03:38.990 part of the Agilent Technologies spinoff. Right, 00:03:39.110 --> 00:03:41.729 Agilent. So Agilent Technologies then housed 00:03:41.729 --> 00:03:44.030 this semiconductor products group for about six 00:03:44.030 --> 00:03:47.030 years. This was, you could say, the calm before 00:03:47.030 --> 00:03:49.430 the storm, a necessary period of independent 00:03:49.430 --> 00:03:51.750 operation before the financial engineers stepped 00:03:51.750 --> 00:03:53.889 in. And this is where we meet the name that defines 00:03:53.889 --> 00:03:56.250 the mid -period of the company's life, Avago. 00:03:56.650 --> 00:03:59.289 This is where private equity steps in to essentially 00:03:59.289 --> 00:04:02.810 carve out a focused, profitable entity. And this 00:04:02.810 --> 00:04:06.030 is a critical strategic move in 2005. Two giants 00:04:06.030 --> 00:04:08.409 in private equity, KKR and Silver Lake Partners, 00:04:08.629 --> 00:04:10.629 they acquire the semiconductor products group 00:04:10.629 --> 00:04:13.289 from Agilent. For how much? They paid $2 .6 billion 00:04:13.289 --> 00:04:16.230 for it and renamed the resulting entity Avago 00:04:16.230 --> 00:04:18.449 Technologies. And you have to understand, this 00:04:18.449 --> 00:04:20.910 transaction wasn't just a simple change of ownership. 00:04:21.269 --> 00:04:24.300 No. It was the start of a real operational shift. 00:04:24.620 --> 00:04:28.399 Private equity demands efficiency, focus, and 00:04:28.399 --> 00:04:31.600 aggressive growth. And that immediately infused 00:04:31.600 --> 00:04:34.160 the company's culture. Avago was the nominal 00:04:34.160 --> 00:04:36.699 corporate entity that would survive all the future 00:04:36.699 --> 00:04:39.959 mergers. When a company is taken over by private 00:04:39.959 --> 00:04:42.620 equity, especially KKR and Silver Lake, you expect 00:04:42.620 --> 00:04:45.879 immediate internal corporate activity, you know, 00:04:45.879 --> 00:04:48.279 shuffling the deck to find the most profitable 00:04:48.279 --> 00:04:50.639 configuration. Did that happen right away? It 00:04:50.639 --> 00:04:52.779 absolutely did. I mean, the goal of private equity 00:04:52.779 --> 00:04:54.819 is to optimize the portfolio for profitability 00:04:54.819 --> 00:04:58.269 and an eventual public offering. Of course. Avago 00:04:58.269 --> 00:05:00.509 was busy from day one. Just a few months after 00:05:00.509 --> 00:05:04.009 forming in October 2005, they sold their IO Solutions 00:05:04.009 --> 00:05:07.470 unit to PMC Sierra for about $42 .5 million. 00:05:07.769 --> 00:05:09.709 They were already pruning the less strategic 00:05:09.709 --> 00:05:11.790 components. And they were also adding strategically 00:05:11.790 --> 00:05:13.949 important niche technologies, right? Not just 00:05:13.949 --> 00:05:16.610 selling. Yes, that's a great point. In 2008, 00:05:16.769 --> 00:05:19.649 they acquired Infineon Technologies' Munich -based 00:05:19.649 --> 00:05:22.870 bulk acoustic wave business for about $21 .5 00:05:22.870 --> 00:05:26.170 million. Bulk Acoustic Wave. Bulk Acoustic Wave 00:05:26.170 --> 00:05:29.879 or BAW technology. It's crucial for modern RF 00:05:29.879 --> 00:05:31.980 filters used in wireless communication. Like 00:05:31.980 --> 00:05:34.660 in our phones. Exactly. So by acquiring that, 00:05:34.720 --> 00:05:36.480 they strengthened their position in specialized 00:05:36.480 --> 00:05:39.019 high -frequency components, a very high -margin 00:05:39.019 --> 00:05:42.019 specialized business area. And all the streamlining, 00:05:42.079 --> 00:05:44.779 all this optimization, led directly to the company 00:05:44.779 --> 00:05:47.899 going public? It did. Avago Technologies successfully 00:05:47.899 --> 00:05:50.680 completed its initial public offering on NASDAQ 00:05:50.680 --> 00:05:55.060 with the now -famous ticker symbol AVGO on August 00:05:55.060 --> 00:05:59.209 6, 2009. They initially aimed high, and the market 00:05:59.209 --> 00:06:02.110 clearly saw the value in this newly optimized 00:06:02.110 --> 00:06:05.009 specialized semiconductor firm. But at this point, 00:06:05.069 --> 00:06:07.250 they're still primarily a vendor of specialized 00:06:07.250 --> 00:06:09.050 components. That's right. They're not yet an 00:06:09.050 --> 00:06:11.149 infrastructure behemoth. Okay. Let's talk about 00:06:11.149 --> 00:06:13.490 the acquisitions that started moving Avago away 00:06:13.490 --> 00:06:15.709 from being just specialized components toward 00:06:15.709 --> 00:06:19.009 what we recognize today, a mainstream industry 00:06:19.009 --> 00:06:21.889 power player. This is where that M &A playbook 00:06:21.889 --> 00:06:24.329 really began to crystallize. The initial steps 00:06:24.329 --> 00:06:27.750 were foundational. In 2013, they spent about 00:06:27.750 --> 00:06:31.550 $400 million to acquire PsyOptics. PsyOptics. 00:06:31.629 --> 00:06:34.029 And that dramatically expanded their fiber optics 00:06:34.029 --> 00:06:37.050 portfolio. And, you know, fiber optics are non 00:06:37.050 --> 00:06:38.810 -negotiable for high -speed data transmission 00:06:38.810 --> 00:06:41.529 in data centers and telecom networks, so that 00:06:41.529 --> 00:06:44.290 this bolstered their core physical layer dominance. 00:06:44.810 --> 00:06:46.970 But the real inflection point in this era, the 00:06:46.970 --> 00:06:50.389 move that really changed their trajectory, was 00:06:50.389 --> 00:06:53.550 the LSI Corporation acquisition in 2013. Oh, 00:06:53.550 --> 00:06:56.029 absolutely. That felt like the moment Avago started 00:06:56.029 --> 00:06:57.930 playing in the major leagues. That's absolutely 00:06:57.930 --> 00:07:01.069 true. The LSI deal was massive for the time, 00:07:01.610 --> 00:07:04.610 $6 .6 billion. It fundamentally repositioned 00:07:04.610 --> 00:07:06.829 the company. Avago had been focused on components 00:07:06.829 --> 00:07:09.339 for the edges of the network. Right. LSI, however, 00:07:09.519 --> 00:07:12.000 was a major supplier of chips and storage technology 00:07:12.000 --> 00:07:14.339 critical for inside the data center. For enterprise 00:07:14.339 --> 00:07:16.139 storage. Particularly for enterprise storage 00:07:16.139 --> 00:07:19.120 and hard drive controllers. By bringing LSI in, 00:07:19.199 --> 00:07:21.259 Avago achieved two things. They gained scale, 00:07:21.399 --> 00:07:23.360 and they moved into the mainstream infrastructure 00:07:23.360 --> 00:07:26.079 market, securing a foothold in server storage. 00:07:26.519 --> 00:07:28.920 So LSI gave them the foundation to be a data 00:07:28.920 --> 00:07:30.860 center leader, which is where the real money 00:07:30.860 --> 00:07:32.720 is made today. Exactly. It moved them from a 00:07:32.720 --> 00:07:35.100 specialized supplier to a crucial infrastructure 00:07:35.100 --> 00:07:39.339 provider. This also helped finance the deal through 00:07:39.339 --> 00:07:41.899 strategic divestiture. The discipline we see 00:07:41.899 --> 00:07:45.500 today, buy big, keep the core, sell the rest, 00:07:45.680 --> 00:07:48.120 was really established here. Tell us more about 00:07:48.120 --> 00:07:51.579 that streamlining post -LSI. What did they shed 00:07:51.579 --> 00:07:53.639 to make that acquisition financially efficient? 00:07:54.089 --> 00:07:57.009 The cleanup was swift and very purposeful. They 00:07:57.009 --> 00:07:59.810 sold off LSI's acquired solid state drive or 00:07:59.810 --> 00:08:02.529 SSD controller business to Seagate Technology 00:08:02.529 --> 00:08:05.689 in May 2014. Okay. Then a few months later in 00:08:05.689 --> 00:08:09.069 August 2014, they sold LSI's Axia networking 00:08:09.069 --> 00:08:12.889 business to Intel for a significant $650 million. 00:08:13.250 --> 00:08:15.350 And the Axia chips were good, right? They were 00:08:15.350 --> 00:08:17.189 good, but they didn't align perfectly with the 00:08:17.189 --> 00:08:19.490 evolving data center and connectivity focus that 00:08:19.490 --> 00:08:21.970 Avago was prioritizing. They demonstrated willingness 00:08:21.970 --> 00:08:24.199 to offload billions of dollars. in assets if 00:08:24.199 --> 00:08:26.220 they didn't meet the absolute highest standards 00:08:26.220 --> 00:08:28.800 of profitability or strategic fit. That disciplined 00:08:28.800 --> 00:08:31.319 approach, buying an entire company, paying for 00:08:31.319 --> 00:08:33.960 it partly by selling off, say, 20 % of its parts. 00:08:34.720 --> 00:08:37.200 That's crucial to understanding their CEO's subsequent 00:08:37.200 --> 00:08:39.539 strategy. It is the essence of the playbook. 00:08:39.659 --> 00:08:42.080 They are not sentimental about maintaining a 00:08:42.080 --> 00:08:44.620 sprawling product catalog. If an asset is valuable 00:08:44.620 --> 00:08:46.960 but doesn't contribute optimally to the centralized 00:08:46.960 --> 00:08:49.759 strategy of high -margin enterprise infrastructure, 00:08:50.120 --> 00:08:53.360 it is very quickly converted to cash. So now 00:08:53.360 --> 00:08:55.519 we enter the era that gives the company its current 00:08:55.519 --> 00:08:58.460 identity. In 2015, Avago, the nominal survivor, 00:08:58.840 --> 00:09:01.340 makes an acquisition so large it takes the name 00:09:01.340 --> 00:09:03.279 of the company it bought. This was the deal that 00:09:03.279 --> 00:09:06.700 truly defined the modern firm. On May 28, 2015, 00:09:07.200 --> 00:09:09.659 Avago announced it was buying Broadcom Corporation, 00:09:10.000 --> 00:09:12.720 the original large semiconductor vendor. For 00:09:12.720 --> 00:09:15.279 a staggering amount. A staggering $37 billion. 00:09:15.620 --> 00:09:18.820 That was composed of $17 billion in cash and 00:09:18.820 --> 00:09:22.399 $20 billion in Alvo shares. The sheer scale of 00:09:22.399 --> 00:09:24.299 that transaction instantly vaulted the combined 00:09:24.299 --> 00:09:26.840 entity into the absolute top tier of the global 00:09:26.840 --> 00:09:28.909 semiconductor market. And the naming choice is 00:09:28.909 --> 00:09:32.409 fascinating. Why keep the Avoticker, AVGO, but 00:09:32.409 --> 00:09:34.769 adopt the Broadcom name? It was a very strategic 00:09:34.769 --> 00:09:37.710 branding move. Avago was the legal and corporate 00:09:37.710 --> 00:09:40.090 survivor, so the ticker was maintained. That 00:09:40.090 --> 00:09:42.330 ensures continuity for investors. Right. But 00:09:42.330 --> 00:09:45.370 the Broadcom brand carried massive weight. recognition, 00:09:45.610 --> 00:09:48.590 and frankly, reputation in mobile, networking, 00:09:48.669 --> 00:09:51.710 and consumer electronics. So the combined company 00:09:51.710 --> 00:09:55.070 became Broadcom LTD, and the original Broadcom 00:09:55.070 --> 00:09:58.629 core ticker, BRCM, was retired. And overnight, 00:09:58.789 --> 00:10:00.850 they were a huge company. Overnight, they were 00:10:00.850 --> 00:10:04.570 a $77 billion market cap company with pro forma 00:10:04.570 --> 00:10:07.590 annual revenues of $15 billion. Beyond just the 00:10:07.590 --> 00:10:09.730 revenue, what did the original Broadcom core 00:10:09.730 --> 00:10:12.350 bring to the table that was so valuable? Intellectual 00:10:12.350 --> 00:10:14.690 property. That was the crown jewel. The original 00:10:14.690 --> 00:10:17.909 Broadcom had an enormous, deep patent portfolio 00:10:17.909 --> 00:10:21.529 across virtually every modern technology. Mobile 00:10:21.529 --> 00:10:24.190 connectivity, data center components, and the 00:10:24.190 --> 00:10:27.250 nascent Internet of Things, or IoT. So they became 00:10:27.250 --> 00:10:29.990 a patent powerhouse. The combined Broadcom LTD 00:10:29.990 --> 00:10:32.210 instantly became the ninth largest holder of 00:10:32.210 --> 00:10:34.629 patents among top semiconductor vendors globally. 00:10:35.070 --> 00:10:37.450 And that patent muscle is critical, not just 00:10:37.450 --> 00:10:39.710 for innovation, but for defending your market 00:10:39.710 --> 00:10:41.490 position against competitors through litigation. 00:10:42.029 --> 00:10:44.190 And even in this massive merger, the playbook 00:10:44.190 --> 00:10:47.429 held firm, immediately shedding non -core assets 00:10:47.429 --> 00:10:50.710 to focus the balance sheet. Exactly. In May 2016, 00:10:51.169 --> 00:10:54.029 Cypress Semiconductor acquired the legacy Broadcom 00:10:54.029 --> 00:10:57.509 Corporation's entire IoT product line. The whole 00:10:57.509 --> 00:11:00.669 thing. The whole thing. Wi -Fi, Bluetooth, ZigBee 00:11:00.669 --> 00:11:04.950 IP, and the WICD development platform for $550 00:11:04.950 --> 00:11:08.389 million. This again showed the leadership's focus. 00:11:09.079 --> 00:11:11.320 They weren't interested in lower margin consumer 00:11:11.320 --> 00:11:13.659 or broad developer platforms. No, they wanted. 00:11:13.700 --> 00:11:15.840 They wanted the high margin, high volume infrastructure 00:11:15.840 --> 00:11:18.480 and networking components. OK, so this is where 00:11:18.480 --> 00:11:22.500 the story pivots from pure M &A to geopolitical 00:11:22.500 --> 00:11:24.980 drama. Because of its Singapore headquarters, 00:11:25.379 --> 00:11:27.899 the now Broadcom LTD started running straight 00:11:27.899 --> 00:11:30.279 into the wall of U .S. national security scrutiny. 00:11:30.480 --> 00:11:32.639 Right. The first major obstacle was the proposed 00:11:32.639 --> 00:11:35.860 $5 .5 billion acquisition of Brocade Communications 00:11:35.860 --> 00:11:39.639 Systems in 2016. Brocade. Because Broadcom was 00:11:39.639 --> 00:11:41.740 legally domiciled in Singapore at the time, this 00:11:41.740 --> 00:11:43.580 transaction was flagged and immediately delayed 00:11:43.580 --> 00:11:45.720 for extensive review by the Committee on Foreign 00:11:45.720 --> 00:11:49.340 Investment in the United States, CFIUS. And CFIUS 00:11:49.340 --> 00:11:51.279 is the body that reviews foreign acquisitions 00:11:51.279 --> 00:11:54.539 for national security risks. How did Broadcom 00:11:54.539 --> 00:11:56.960 react to this growing resistance? I mean, they 00:11:56.960 --> 00:11:59.980 had a massive deal portfolio that they couldn't 00:11:59.980 --> 00:12:01.980 execute if they kept hitting this barrier. They 00:12:01.980 --> 00:12:05.179 executed a brilliant if, you know, controversial 00:12:05.179 --> 00:12:08.840 strategic maneuver, to legally circumvent future 00:12:08.840 --> 00:12:12.539 CFIUS reviews, which focus primarily on foreign 00:12:12.539 --> 00:12:14.899 entities acquiring American strategic assets. 00:12:15.019 --> 00:12:16.720 Which they wanted to keep doing. Which they wanted 00:12:16.720 --> 00:12:19.519 to keep doing, yes. Broadcom announced in November 00:12:19.519 --> 00:12:22.320 2017 that they were relocating their legal address 00:12:22.320 --> 00:12:25.399 from Singapore to Delaware. Wow. This was a clear, 00:12:25.460 --> 00:12:27.820 calculated move to redefine themselves as a U 00:12:27.820 --> 00:12:30.820 .S. domiciled entity. It was directly tied to 00:12:30.820 --> 00:12:32.659 the name change from Broadcom Limited to the 00:12:32.659 --> 00:12:35.340 current Broadcom Inc. It was a formal declaration. 00:12:35.659 --> 00:12:38.320 We are now an American company. And that sets 00:12:38.320 --> 00:12:40.779 the stage for the alternate drama, the blocked 00:12:40.779 --> 00:12:43.940 hostile takeover of Qualcomm. Even after redomiciling, 00:12:44.080 --> 00:12:46.039 the U .S. government stepped in to stop what 00:12:46.039 --> 00:12:47.639 would have been the largest tech acquisition 00:12:47.639 --> 00:12:49.840 in history at the time. This happened almost 00:12:49.840 --> 00:12:51.799 immediately after the Delaware move was announced. 00:12:52.500 --> 00:12:56.600 In late 2017, Broadcom proposed to purchase Qualcomm, 00:12:56.799 --> 00:12:58.740 one of the most critical players in mobile chip 00:12:58.740 --> 00:13:03.820 technology, for US $130 billion. $130 billion. 00:13:04.299 --> 00:13:07.000 Qualcomm initially rejected the unsolicited offer. 00:13:07.200 --> 00:13:10.980 So Broadcom revised the hostile bid down to $117 00:13:10.980 --> 00:13:13.720 billion. This wasn't just corporate maneuvering. 00:13:13.720 --> 00:13:16.480 This became a global political statement. Washington, 00:13:16.659 --> 00:13:18.840 D .C. slammed the brakes on the entire deal. 00:13:19.100 --> 00:13:21.779 What was the exact security concerns cited by 00:13:21.779 --> 00:13:24.019 the Trump administration when they issued that 00:13:24.019 --> 00:13:26.460 executive order to block it? The concerns were 00:13:26.460 --> 00:13:29.039 explicit and they focused on the future of 5G 00:13:29.039 --> 00:13:32.480 dominance. CFI US stated that Broadcom, even 00:13:32.480 --> 00:13:35.379 with its planned U .S. redomiciling, had historically 00:13:35.379 --> 00:13:37.659 been too close to China and chipmaker hallway. 00:13:37.940 --> 00:13:40.820 The core worry was that a massive, dead -laden 00:13:40.820 --> 00:13:43.059 acquisition of Qualcomm would force Broadcom 00:13:43.059 --> 00:13:45.539 to cut research and development spending, specifically 00:13:45.539 --> 00:13:48.320 in Qualcomm's 5G programs. And why was that R 00:13:48.320 --> 00:13:50.679 &D cut a national security risk? Because the 00:13:50.679 --> 00:13:53.039 U .S. government believed any resulting slowdown 00:13:53.039 --> 00:13:56.019 in 5G R &D by the American entity would create 00:13:56.019 --> 00:13:58.919 a massive competitive vacuum. A vacuum that China 00:13:58.919 --> 00:14:01.639 would fill. A vacuum which CFIUS specifically 00:14:01.639 --> 00:14:04.960 warned China would likely compete robustly to 00:14:04.960 --> 00:14:08.120 fill. They argued, and I'm quoting from their 00:14:08.120 --> 00:14:11.039 statement, that a shift to Chinese dominance 00:14:11.039 --> 00:14:13.559 in 5G would have substantial negative national 00:14:13.559 --> 00:14:15.720 security consequences for the United States. 00:14:16.200 --> 00:14:19.279 Wow. So it was less about who owned the chips 00:14:19.279 --> 00:14:21.919 today and more about preventing China from gaining 00:14:21.919 --> 00:14:25.019 a strategic technological lead in the next generation 00:14:25.019 --> 00:14:28.450 of wireless standards. They viewed Qualcomm as 00:14:28.450 --> 00:14:30.909 a strategic national asset that could not fall 00:14:30.909 --> 00:14:33.669 under the control, even indirectly, of an entity 00:14:33.669 --> 00:14:36.269 with perceived deep ties to a geopolitical rival. 00:14:36.779 --> 00:14:39.559 So was this strictly a security issue or was 00:14:39.559 --> 00:14:42.039 there an element of, say, economic protectionism 00:14:42.039 --> 00:14:44.279 wrapped up in that executive order? It was a 00:14:44.279 --> 00:14:46.360 powerful confluence. I mean, while the administration 00:14:46.360 --> 00:14:48.899 focused on security, many critics at the time 00:14:48.899 --> 00:14:50.740 pointed out that the decision was also extremely 00:14:50.740 --> 00:14:52.779 consistent with maintaining trade balance objectives 00:14:52.779 --> 00:14:55.200 and ensuring the U .S. retained competitiveness 00:14:55.200 --> 00:14:57.879 in future tech. Either way, the political message 00:14:57.879 --> 00:15:01.419 was clear. Certain high -spakes hardware M &A 00:15:01.419 --> 00:15:04.580 was off the table for Broadcom. Two days after 00:15:04.580 --> 00:15:07.419 the executive order, Broadcom withdrew the takeover 00:15:07.419 --> 00:15:10.159 bid. That is a fundamental turning point. You 00:15:10.159 --> 00:15:12.039 spend months trying to pull off the biggest deal 00:15:12.039 --> 00:15:13.940 in tech history, and the government just steps 00:15:13.940 --> 00:15:16.519 in and blocks you. What does a company do next? 00:15:16.820 --> 00:15:19.860 You pivot. And you pivot hard. The failure to 00:15:19.860 --> 00:15:22.320 secure Qualcomm was an inflection point that 00:15:22.320 --> 00:15:25.220 forced CEO Hawk Tan to rethink the entire growth 00:15:25.220 --> 00:15:27.899 strategy. If you can't buy the biggest hardware 00:15:27.899 --> 00:15:30.799 company, where do you find the scale and the 00:15:30.799 --> 00:15:33.549 margins to satisfy Wall Street? The answer was 00:15:33.549 --> 00:15:35.929 infrastructure software. The internal strategy 00:15:35.929 --> 00:15:38.809 memo must have been crystal clear. Focus on infrastructure 00:15:38.809 --> 00:15:41.570 assets that generate recurring revenue and have, 00:15:41.649 --> 00:15:43.950 you know, much higher gross margins than commodity 00:15:43.950 --> 00:15:46.149 chips. A completely different business model. 00:15:46.289 --> 00:15:48.750 Completely. Broadcom had historically been a 00:15:48.750 --> 00:15:51.289 semiconductor -based company. But the failure 00:15:51.289 --> 00:15:54.169 of the Qualcomm bid directly led management to 00:15:54.169 --> 00:15:56.669 view infrastructure software acquisitions as 00:15:56.669 --> 00:15:58.909 the primary alternative path to massive scale. 00:15:59.279 --> 00:16:02.639 And the first loud signal of this new strategy 00:16:02.639 --> 00:16:06.360 arrived in 2018 with the CA Technologies purchase. 00:16:06.879 --> 00:16:11.539 $18 .9 billion. A massive move into a sector 00:16:11.539 --> 00:16:13.879 where Broadcom had almost no historical presence. 00:16:14.179 --> 00:16:16.659 Right. And CA Technologies, formerly Computer 00:16:16.659 --> 00:16:20.279 Associates, was a heritage giant. A true behemoth. 00:16:20.320 --> 00:16:22.779 In mainframes. They were deeply entrenched in 00:16:22.779 --> 00:16:25.059 software for mainframe computers, the systems 00:16:25.059 --> 00:16:27.100 that still run major banks, airlines, and insurance 00:16:27.100 --> 00:16:29.919 companies. But they also had significant newer 00:16:29.919 --> 00:16:32.539 cloud computing offerings. So what Broadcom really 00:16:32.539 --> 00:16:35.279 bought was? They bought entrenched infrastructure 00:16:35.279 --> 00:16:38.059 and a customer base that requires long -term 00:16:38.059 --> 00:16:40.000 expensive contracts for maintenance and upgrades. 00:16:40.399 --> 00:16:42.320 Let's dive into the financial motivation there. 00:16:42.399 --> 00:16:44.360 Why is mainframe software so much more attractive 00:16:44.360 --> 00:16:47.159 than, say, selling a Tomahawk chip? Well, it's 00:16:47.159 --> 00:16:48.860 all about the quality of the revenue. When you 00:16:48.860 --> 00:16:51.519 sell a chip, it's a one -time transaction. Even 00:16:51.519 --> 00:16:54.120 a high -end Tomahawk switch chip, while profitable, 00:16:54.320 --> 00:16:56.639 eventually sees its price eroded by competitors, 00:16:56.820 --> 00:16:58.879 and you have to constantly innovate just to stay 00:16:58.879 --> 00:17:01.980 ahead. Software, especially mainframe and enterprise 00:17:01.980 --> 00:17:05.200 infrastructure software, offers recurring, subscription 00:17:05.200 --> 00:17:07.900 -based, or high -maintenance annuity revenue. 00:17:08.740 --> 00:17:12.700 These contracts are incredibly sticky. Sticky. 00:17:13.000 --> 00:17:15.819 It's extremely expensive and complex for a Fortune 00:17:15.819 --> 00:17:19.220 500 company to rip out its core mainframe or 00:17:19.220 --> 00:17:22.619 data center management tools, meaning CA's cash 00:17:22.619 --> 00:17:25.420 flow was highly predictable and had much higher 00:17:25.420 --> 00:17:27.680 gross margins than hardware. So they weren't 00:17:27.680 --> 00:17:29.660 just buying software. They were buying financial 00:17:29.660 --> 00:17:32.480 predictability and high profit margins. Precisely. 00:17:32.480 --> 00:17:34.140 They were buying the stability of the enterprise 00:17:34.140 --> 00:17:36.960 customer relationship. That appetite for infrastructure 00:17:36.960 --> 00:17:40.059 software only grew. The very next year, they 00:17:40.059 --> 00:17:42.400 doubled down on security infrastructure with 00:17:42.400 --> 00:17:45.319 Symantec Enterprise Security. That 2019 deal 00:17:45.319 --> 00:17:49.039 for $10 .7 billion in cash was a logical continuation 00:17:49.039 --> 00:17:50.960 of the strategy. Makes sense. After acquiring 00:17:50.960 --> 00:17:53.299 the software that runs the corporate IT infrastructure 00:17:53.299 --> 00:17:55.640 with CA, you next need the software that protects 00:17:55.640 --> 00:17:58.799 it. Symantec gave Broadcom instant dominance 00:17:58.799 --> 00:18:01.220 in enterprise security products. These products 00:18:01.220 --> 00:18:03.599 are now operated under the Symantec brand within 00:18:03.599 --> 00:18:05.890 Broadcom. And just to clarify for everyone listening, 00:18:06.049 --> 00:18:08.230 this was strictly the business facing side of 00:18:08.230 --> 00:18:11.170 Symantec, not the consumer antivirus products 00:18:11.170 --> 00:18:13.430 most people remember. That's a crucial distinction. 00:18:13.809 --> 00:18:16.549 The consumer portion of Symantec was spun off 00:18:16.549 --> 00:18:19.390 and changed its name to Norton LifeLock, which 00:18:19.390 --> 00:18:22.369 is now GenDigital. Broadcom was strictly interested 00:18:22.369 --> 00:18:25.269 in the high margin, sticky contracts tied to 00:18:25.269 --> 00:18:27.230 corporate networks. And the playbook continues. 00:18:27.630 --> 00:18:29.569 And again, we see the M &A playbook in action. 00:18:29.890 --> 00:18:33.490 In January 2020. They sold Symantec's cybersecurity 00:18:33.490 --> 00:18:37.369 services division, the 300 employees doing consulting 00:18:37.369 --> 00:18:40.190 to Accenture PLC. Of course they care. They keep 00:18:40.190 --> 00:18:41.789 the intellectual property and the proprietary 00:18:41.789 --> 00:18:44.470 platform software, and they shed the expensive, 00:18:44.750 --> 00:18:47.430 lower -margin service component. This disciplined, 00:18:47.589 --> 00:18:50.150 almost surgical approach to divestiture is a 00:18:50.150 --> 00:18:52.750 hallmark. But doesn't this strategy, this constant 00:18:52.750 --> 00:18:55.150 acquisition and aggressive streamlining, which 00:18:55.150 --> 00:18:57.950 often involves deep cost -cutting in R &D and 00:18:57.950 --> 00:19:00.710 staffing at the acquired companies, doesn't it 00:19:00.710 --> 00:19:03.710 fundamentally risk destroying the acquired company's 00:19:03.710 --> 00:19:06.349 innovative culture and alienating key engineering 00:19:06.349 --> 00:19:10.009 talent? That is the perpetual, crucial tension 00:19:10.009 --> 00:19:12.759 surrounding the Broadcom strategy. sometimes 00:19:12.759 --> 00:19:15.480 referred to internally and externally as the 00:19:15.480 --> 00:19:18.660 Hoctan playbook. The Hoctan playbook. The company 00:19:18.660 --> 00:19:21.460 is often criticized for prioritizing short -term 00:19:21.460 --> 00:19:24.099 margin growth and integration efficiency over 00:19:24.099 --> 00:19:28.140 long -term fundamental R &D in the acquired software 00:19:28.140 --> 00:19:31.230 firms. What do we know about what happened after 00:19:31.230 --> 00:19:33.730 those deals? Reports suggest that after the CA 00:19:33.730 --> 00:19:37.109 and Symantec deals, there were significant product 00:19:37.109 --> 00:19:40.170 line consolidations and, importantly, price increases, 00:19:40.430 --> 00:19:43.009 especially for long -term customers. For Broadcom, 00:19:43.210 --> 00:19:45.650 the priority is extracting maximum cash flow 00:19:45.650 --> 00:19:47.589 from an already established critical infrastructure 00:19:47.589 --> 00:19:50.069 base. Rather than nurturing innovation. Rather 00:19:50.069 --> 00:19:51.930 than nurturing bleeding -edge innovation within 00:19:51.930 --> 00:19:54.029 those acquired software divisions, yes. Which 00:19:54.029 --> 00:19:55.549 brings us to the centerpiece of the software 00:19:55.549 --> 00:19:58.150 pivot, the deal that fundamentally cemented their 00:19:58.150 --> 00:20:00.539 hybrid identity and their trillion -dollar status, 00:20:00.880 --> 00:20:04.140 the VMware mega deal. Announced in May 2022, 00:20:04.500 --> 00:20:08.220 this was truly breathtaking. $61 billion in cash 00:20:08.220 --> 00:20:10.779 and stock, plus the assumption of $8 billion 00:20:10.779 --> 00:20:13.740 in debt, valuing the total transaction at $69 00:20:13.740 --> 00:20:17.779 billion upon closing. $69 billion. And VMware 00:20:17.779 --> 00:20:21.259 is, simply put... the king of virtualization 00:20:21.259 --> 00:20:24.099 and cloud computing software. If you run a data 00:20:24.099 --> 00:20:26.599 center or a private cloud anywhere in the world, 00:20:26.740 --> 00:20:30.640 you are running VMware software. So by acquiring 00:20:30.640 --> 00:20:33.339 VMware, Broadcom isn't just selling chips to 00:20:33.339 --> 00:20:35.920 data centers anymore. No. They now own the operating 00:20:35.920 --> 00:20:38.240 system for the data center itself. Precisely. 00:20:38.259 --> 00:20:40.940 It gives them unparalleled influence over how 00:20:40.940 --> 00:20:43.859 corporate customers build, manage, and scale 00:20:43.859 --> 00:20:46.519 their IT infrastructure. It's the ultimate sticky 00:20:46.519 --> 00:20:50.319 asset. However, a $69 billion deal involving 00:20:50.319 --> 00:20:53.599 such a central piece of global IT infrastructure 00:20:53.599 --> 00:20:55.980 was never going to be easy to get past regulators. 00:20:56.359 --> 00:20:59.680 This required a truly global, multi -year regulatory 00:20:59.680 --> 00:21:02.319 clearance process. Oh, the scrutiny was intense. 00:21:02.660 --> 00:21:04.940 It highlighted Broadcom's strategic importance 00:21:04.940 --> 00:21:08.019 and its growing geopolitical profile. The UK's 00:21:08.019 --> 00:21:11.019 Competition and Markets Authority, the CMA, investigated 00:21:11.019 --> 00:21:13.099 for potential substantial lessening of competition. 00:21:13.440 --> 00:21:14.759 They were worried about... They were worried 00:21:14.759 --> 00:21:16.779 about Broadcom potentially tying its hardware... 00:21:16.809 --> 00:21:19.009 sales to VMware's essential software. And the 00:21:19.009 --> 00:21:21.329 European Union also put the deal under the microscope, 00:21:21.589 --> 00:21:23.849 leading to multiple extensions. The European 00:21:23.849 --> 00:21:26.630 Commission's investigation was rigorous, specifically 00:21:26.630 --> 00:21:29.150 focusing on whether the merger would allow Broadcom 00:21:29.150 --> 00:21:32.470 to restrict competition in the supply of certain 00:21:32.470 --> 00:21:35.380 hardware components used in servers. They forced 00:21:35.380 --> 00:21:38.200 deadline extensions multiple times past May 2023, 00:21:38.519 --> 00:21:42.299 then August 2023, and finally settling on November 00:21:42.299 --> 00:21:45.119 2023. A huge amount of effort to get that done. 00:21:45.279 --> 00:21:48.359 A huge amount of time and money providing concessions 00:21:48.359 --> 00:21:50.559 and information to satisfy these concerns. And 00:21:50.559 --> 00:21:52.940 the final hurdle was perhaps the most crucial 00:21:52.940 --> 00:21:55.160 politically, given the ghost of the Qualcomm 00:21:55.160 --> 00:21:58.599 deal, approval from China. China's approval was 00:21:58.599 --> 00:22:01.779 the final non -negotiable step for closing. And 00:22:01.779 --> 00:22:04.339 amid rising U .S.-China tensions, the deal was 00:22:04.339 --> 00:22:07.559 approved in November 2023, but it came with specific 00:22:07.559 --> 00:22:10.200 crucial conditions imposed by Beijing. What were 00:22:10.200 --> 00:22:12.339 those conditions? These conditions required that 00:22:12.339 --> 00:22:14.660 VMware's server software must remain compatible 00:22:14.660 --> 00:22:16.819 with local Chinese hardware, and that Broadcom 00:22:16.819 --> 00:22:19.000 could not restrict Chinese customers from using 00:22:19.000 --> 00:22:21.460 its own hardware products. This was a clear measure 00:22:21.460 --> 00:22:24.000 by China to ensure market competition and prevent 00:22:24.000 --> 00:22:27.039 a potentially adverse tie -in sales situation 00:22:27.039 --> 00:22:30.049 impacting local tech ecosystems. The deal finally 00:22:30.049 --> 00:22:33.430 closed on November 22, 2023, and the corporate 00:22:33.430 --> 00:22:35.750 changes were immediate, underscoring the importance 00:22:35.750 --> 00:22:38.569 of VMware to the new Broadcom identity. Yes. 00:22:39.049 --> 00:22:41.329 CEO Hawk Tan announced the relocation of the 00:22:41.329 --> 00:22:43.470 company's headquarters from San Jose straight 00:22:43.470 --> 00:22:46.069 to the former VMware campus in Palo Alto, California. 00:22:46.390 --> 00:22:49.069 It was a physical and symbolic merger. putting 00:22:49.069 --> 00:22:51.509 the software pillar front and center. And true 00:22:51.509 --> 00:22:53.630 to form. And staying true to the playbook, in 00:22:53.630 --> 00:22:57.130 February 2024, they sold the inherited end -user 00:22:57.130 --> 00:22:59.750 computing division, the desktop virtualization 00:22:59.750 --> 00:23:04.269 part, to KKR for $4 billion. Keep the core data 00:23:04.269 --> 00:23:06.910 center hypervisor, sell the client -facing lower 00:23:06.910 --> 00:23:09.430 margin pieces. Let's quantify this staggering 00:23:09.430 --> 00:23:12.069 growth. If we look at the financials, the M &A 00:23:12.069 --> 00:23:14.309 engine has resulted in geometric growth in both 00:23:14.309 --> 00:23:16.609 assets and personnel. The employee count tells 00:23:16.609 --> 00:23:19.660 a dramatic story of integration. Avago had about 00:23:19.660 --> 00:23:22.859 3 ,200 employees in 2009. That grew to 8 ,200 00:23:22.859 --> 00:23:24.700 by the time of the first Broadcom acquisition 00:23:24.700 --> 00:23:27.980 in 2015. After the CA and Symantec deals, they 00:23:27.980 --> 00:23:31.339 were at 19 ,000 employees in 2019. Post -VMware 00:23:31.339 --> 00:23:34.359 integration in 2024, the pound jumped to 37 ,000 00:23:34.359 --> 00:23:36.339 employees. It doubled in five years. They doubled 00:23:36.339 --> 00:23:38.339 the size of the workforce in just five years, 00:23:38.500 --> 00:23:40.980 purely through integration. And the financial 00:23:40.980 --> 00:23:43.539 statements, I imagine, reflect that massive influx 00:23:43.539 --> 00:23:46.279 of high -value assets. The numbers are staggering. 00:23:46.650 --> 00:23:49.289 In 2024, the company reported a colossal revenue 00:23:49.289 --> 00:23:54.170 of US $51 .57 billion. Operating income hit US 00:23:54.170 --> 00:23:58.000 $13 .46 billion. billion. Critically, their total 00:23:58.000 --> 00:24:01.900 assets ballooned to US $165 .6 billion in 2024, 00:24:02.299 --> 00:24:05.480 largely due to VMware being consolidated onto 00:24:05.480 --> 00:24:07.799 the balance sheet. So that's a doubling of assets 00:24:07.799 --> 00:24:10.359 from the previous year. A doubling, yes. This 00:24:10.359 --> 00:24:12.480 confirms that the M &A strategy, for all its 00:24:12.480 --> 00:24:14.720 controversy, has been incredibly effective in 00:24:14.720 --> 00:24:16.839 delivering unprecedented growth and scale. It's 00:24:16.839 --> 00:24:19.519 worth noting the key institutional players who 00:24:19.519 --> 00:24:22.140 hold the reins of this now trillion dollar company. 00:24:22.400 --> 00:24:24.920 Ownership is heavily institutional, which ensures 00:24:24.920 --> 00:24:27.140 a focus on long -term stability and index tracking. 00:24:27.299 --> 00:24:30.240 As of early 2024, the Vanguard Group is the largest 00:24:30.240 --> 00:24:32.960 shareholder at over 10%. BlackRock follows closely 00:24:32.960 --> 00:24:37.059 behind at 7 .63%. Other major investment groups 00:24:37.059 --> 00:24:39.039 like Capital World Investors and State Street 00:24:39.039 --> 00:24:42.480 Corporation hold significant positions. This 00:24:42.480 --> 00:24:45.039 institutional density reflects Broadcom's role 00:24:45.039 --> 00:24:48.180 as an essential foundational stock in global 00:24:48.180 --> 00:24:51.069 technology indices. Despite the massive pivot 00:24:51.069 --> 00:24:53.430 and all the drama of the software acquisitions, 00:24:53.470 --> 00:24:56.630 we must reiterate Broadcom remains a global powerhouse 00:24:56.630 --> 00:24:59.170 in hardware. They are the unseen plumbing of 00:24:59.170 --> 00:25:01.509 the digital world and their chips are everywhere. 00:25:01.849 --> 00:25:03.829 That hardware division is the engine that drives 00:25:03.829 --> 00:25:06.890 the connection economy. Broadcom's core product 00:25:06.890 --> 00:25:09.049 offerings cover data center networking, broadband 00:25:09.049 --> 00:25:12.549 infrastructure, wireless technology, data storage, 00:25:12.690 --> 00:25:14.609 and industrial control. I mean, their components 00:25:14.609 --> 00:25:17.430 are in data center racks, mobile phone base stations, 00:25:17.730 --> 00:25:20.150 home routers. Factory automation systems, and 00:25:20.150 --> 00:25:22.829 thanks to that LSI acquisition, a huge portion 00:25:22.829 --> 00:25:24.950 of the world's storage systems. And we are talking 00:25:24.950 --> 00:25:27.549 about chips used by major vendors, including 00:25:27.549 --> 00:25:30.190 Apple and their smartphones, Dell and their servers, 00:25:30.329 --> 00:25:32.529 and even small -scale computing like the Raspberry 00:25:32.529 --> 00:25:35.990 Pi boards. Right, which use the Broadcom. BCM2711 00:25:35.990 --> 00:25:39.349 chip. They are deeply structurally embedded in 00:25:39.349 --> 00:25:41.190 the global supply chain. They are essential. 00:25:41.509 --> 00:25:44.349 They are, but their true dominance is in high 00:25:44.349 --> 00:25:46.990 -speed networking chips, which enable the hyperscale 00:25:46.990 --> 00:25:49.210 data centers, the Googles and Metas of the world, 00:25:49.309 --> 00:25:51.970 to operate at mind -bending speeds. Let's focus 00:25:51.970 --> 00:25:54.170 on that networking hardware, specifically the 00:25:54.170 --> 00:25:57.009 switch chip portfolio, starting with the Jericho 00:25:57.009 --> 00:26:00.109 2. The Jericho 2 is a remarkable piece of engineering. 00:26:00.640 --> 00:26:04.519 It's a programmable Ethernet switch chip capable 00:26:04.519 --> 00:26:08.539 of achieving up to 10 terabits per second. 10 00:26:08.539 --> 00:26:10.940 terabits switching capacity per single device. 00:26:11.180 --> 00:26:14.160 10 terabits per second. And to put that 10 terabits 00:26:14.160 --> 00:26:17.019 in perspective for you. the learner, that means 00:26:17.019 --> 00:26:19.579 the chip can process the equivalent of transmitting 00:26:19.579 --> 00:26:22.579 over 1 ,000 high -definition movies every single 00:26:22.579 --> 00:26:25.059 second. That is the fundamental speed barrier 00:26:25.059 --> 00:26:27.240 they are breaking. But the Tomahawk series is 00:26:27.240 --> 00:26:29.160 the line that really dictates the speed limits 00:26:29.160 --> 00:26:31.880 for massive cloud networks, demonstrating a relentless, 00:26:32.079 --> 00:26:34.960 almost exponential increase in density. The Tomahawk 00:26:34.960 --> 00:26:37.420 series is the backbone of hyperscale cloud switching. 00:26:37.619 --> 00:26:40.059 The Tomahawk 3, for example, supports high -density 00:26:40.059 --> 00:26:45.640 400 -gigabit Ethernet standards. bits of total 00:26:45.640 --> 00:26:48.220 chip bandwidth which means which means it enables 00:26:48.220 --> 00:26:51.500 32 ports to simultaneously run at 400 gbits and 00:26:51.500 --> 00:26:54.039 the speed trajectory from there is just astounding 00:26:54.039 --> 00:26:56.980 they are essentially doubling the capacity with 00:26:56.980 --> 00:26:59.859 each iteration it's an incredible pace of innovation 00:26:59.859 --> 00:27:02.599 driven by the demands of ai and streaming video 00:27:03.369 --> 00:27:07.809 The Tomahawk 4 achieved 25 .6 T -bits, and the 00:27:07.809 --> 00:27:10.890 latest, the Tomahawk 5, reaches a maximum speed 00:27:10.890 --> 00:27:15.230 of 51 .2 T -bits. 51 .2. When a chip hits 51 00:27:15.230 --> 00:27:17.349 .2 T -bits, you are talking about processing 00:27:17.349 --> 00:27:19.690 the data equivalent of transferring roughly 5 00:27:19.690 --> 00:27:22.529 ,000 high -definition movies every second. These 00:27:22.529 --> 00:27:24.410 chips are non -negotiable infrastructure for 00:27:24.410 --> 00:27:26.390 the world's largest cloud builders. This speed 00:27:26.390 --> 00:27:28.349 and capacity bring us directly to the current 00:27:28.349 --> 00:27:30.630 AI boom and the reason for that trillion -dollar 00:27:30.630 --> 00:27:34.640 valuation. The XPU custom chips. What is an XPU 00:27:34.640 --> 00:27:37.500 in the context of Broadcom and why is it so critical? 00:27:38.190 --> 00:27:41.430 So Broadcom uses the term XPU to describe custom 00:27:41.430 --> 00:27:43.710 application -specific integrated circuits, or 00:27:43.710 --> 00:27:46.269 ASICs, that they co -develop with major hyperscale 00:27:46.269 --> 00:27:48.589 clients specifically for high -intensity specialized 00:27:48.589 --> 00:27:51.529 AI workloads. So this is a collaboration business. 00:27:51.789 --> 00:27:53.970 It's a business based on collaboration, not off 00:27:53.970 --> 00:27:56.150 -the -shelf components. The prime example of 00:27:56.150 --> 00:27:58.450 this co -development model is their longstanding 00:27:58.450 --> 00:28:01.269 relationship with Google on the Tensor Processing 00:28:01.269 --> 00:28:05.210 Unit, or TPU. Broadcom is a critical, often uncredited 00:28:05.210 --> 00:28:08.670 partner in the AI race because they have co -developed 00:28:08.670 --> 00:28:11.269 every single generation of Google's TPU chips. 00:28:11.490 --> 00:28:14.670 Every generation? Every single one. Google designs 00:28:14.670 --> 00:28:17.130 the architecture, but Broadcom translates that 00:28:17.130 --> 00:28:20.109 architecture into manufacturable silicon. Their 00:28:20.109 --> 00:28:24.089 role is comprehensive. They provide crucial proprietary 00:28:24.089 --> 00:28:27.339 technologies. Like what? Like the Certi's high 00:28:27.339 --> 00:28:29.980 -speed interfaces, they handle the complex ASIC 00:28:29.980 --> 00:28:32.599 design implementation, and they manage the fabrication 00:28:32.599 --> 00:28:35.140 and packaging process through third -party foundries 00:28:35.140 --> 00:28:37.690 like TSMC. Wait, you mentioned SERDES high -speed 00:28:37.690 --> 00:28:40.029 interfaces. For the learner, can you explain 00:28:40.029 --> 00:28:42.289 what that is and why it's such a crucial proprietary 00:28:42.289 --> 00:28:44.690 technology that Broadcom brings to the table? 00:28:44.849 --> 00:28:47.150 Absolutely. SERDES, which stands for serializer 00:28:47.150 --> 00:28:49.730 to serializer, are essentially the ultra -high 00:28:49.730 --> 00:28:52.529 -speed internal highways that allow data to flow 00:28:52.529 --> 00:28:54.589 extremely quickly between different chips or 00:28:54.589 --> 00:28:57.630 within a large system, like a DPU cluster. In 00:28:57.630 --> 00:29:00.269 AI, you have massive amounts of data constantly 00:29:00.269 --> 00:29:02.829 needing to move between processors, memory, and 00:29:02.829 --> 00:29:05.890 networking components. Broadcom Certi's technology 00:29:05.890 --> 00:29:08.069 is considered world -class for its efficiency 00:29:08.069 --> 00:29:11.069 and speed. So without? Without robust, proprietary 00:29:11.069 --> 00:29:14.369 Certis, the sheer data flow necessary for modern 00:29:14.369 --> 00:29:17.569 AI models would be impossible. So Broadcom provides 00:29:17.569 --> 00:29:19.589 the high -speed plumbing that makes Google's 00:29:19.589 --> 00:29:22.589 architecture physically work. That explains why 00:29:22.589 --> 00:29:25.089 the relationship is so deep and sticky. They 00:29:25.089 --> 00:29:27.390 are providing the essential foundation. And this 00:29:27.390 --> 00:29:30.089 model is clearly expanding beyond just Google, 00:29:30.130 --> 00:29:33.019 right? It is the future growth engine. Broadcom 00:29:33.019 --> 00:29:35.359 also co -developed Meta's Training and Inference 00:29:35.359 --> 00:29:38.519 Accelerator, or MTIA, chips, which Meta rolled 00:29:38.519 --> 00:29:42.059 out in 2023. And most recently, in 2024, they 00:29:42.059 --> 00:29:45.180 initiated a collaboration with OpenAI to co -develop 00:29:45.180 --> 00:29:48.380 a custom inference chip. This move by OpenAI 00:29:48.380 --> 00:29:51.119 is highly strategic. It's part of a broader industry 00:29:51.119 --> 00:29:53.819 -wide push to reduce the heavy reliance on NVIDIA 00:29:53.819 --> 00:29:56.700 GPUs by developing purpose -built, highly efficient 00:29:56.700 --> 00:29:59.619 silicon for specific AI tasks. So Broadcom is 00:29:59.619 --> 00:30:01.680 positioning itself as... As the ASIC engine for 00:30:01.680 --> 00:30:04.799 the AI giants. Okay, we must now transition to 00:30:04.799 --> 00:30:06.980 the legal and security friction Broadcom has 00:30:06.980 --> 00:30:09.769 faced as it has grown. Despite their cutting 00:30:09.769 --> 00:30:11.849 -edge technology, their products have not been 00:30:11.849 --> 00:30:14.349 without significant, frankly frightening, security 00:30:14.349 --> 00:30:16.750 issues. This is a sharp contrast to their technical 00:30:16.750 --> 00:30:20.670 dominance. In April 2017, Google's famous Project 00:30:20.670 --> 00:30:24.109 Zero security research team investigated Broadcom's 00:30:24.109 --> 00:30:28.309 system on a chip, or SO, Wi -Fi stack. The findings 00:30:28.309 --> 00:30:30.150 were genuinely alarming. What did they find? 00:30:30.289 --> 00:30:33.410 The code for the Wi -Fi stack lacked all basic 00:30:33.410 --> 00:30:35.900 exploit mitigations. Explain what that means 00:30:35.900 --> 00:30:38.420 in layman's terms, what basic protections were 00:30:38.420 --> 00:30:40.980 missing. Well, basic exploit mitigations include 00:30:40.980 --> 00:30:43.319 fundamental software defenses like stack cookies, 00:30:43.460 --> 00:30:45.960 which prevent buffer overflow attacks, and proper 00:30:45.960 --> 00:30:48.339 memory access permission protection. Essentially, 00:30:48.460 --> 00:30:50.200 the software stack was operating without the 00:30:50.200 --> 00:30:52.799 basic safety locks that almost all modern operating 00:30:52.799 --> 00:30:55.440 systems and components utilize to prevent malicious 00:30:55.440 --> 00:30:58.150 intrusion. And what was the terrifying real -world 00:30:58.150 --> 00:31:00.750 risk stemming from that absence of basic defense? 00:31:01.089 --> 00:31:03.170 Project Zero demonstrated that this vulnerability 00:31:03.170 --> 00:31:06.549 allowed for full device takeover by Wi -Fi proximity 00:31:06.549 --> 00:31:09.730 alone, requiring no user interaction. No user 00:31:09.730 --> 00:31:12.569 interaction? None. Think about the drama of that. 00:31:12.670 --> 00:31:15.769 A malicious actor sitting in a cafe could potentially 00:31:15.769 --> 00:31:19.460 hijack your phone. or any affected device just 00:31:19.460 --> 00:31:22.559 by being nearby and exploiting the Wi -Fi vulnerability 00:31:22.559 --> 00:31:25.299 without you ever having to click a link or download 00:31:25.299 --> 00:31:28.180 anything. It was a catastrophic security hole 00:31:28.180 --> 00:31:31.420 that affected numerous major smartphones. Including 00:31:31.420 --> 00:31:34.240 phones. Including devices from Apple, Samsung, 00:31:34.460 --> 00:31:37.420 and Google itself, forcing widespread immediate 00:31:37.420 --> 00:31:40.450 patching. Moving to corporate behavior, Broadcom 00:31:40.450 --> 00:31:43.609 has repeatedly faced massive antitrust investigations 00:31:43.609 --> 00:31:46.089 on both sides of the Atlantic regarding its business 00:31:46.089 --> 00:31:48.849 tactics. The regulatory scrutiny has been consistent 00:31:48.849 --> 00:31:51.730 and severe. In the U .S., the Federal Trade Commission, 00:31:51.950 --> 00:31:55.210 the FTC, investigated Broadcom between 2018 and 00:31:55.210 --> 00:31:57.910 2021. The claim was that the company was engaging 00:31:57.910 --> 00:32:00.670 in anti -competitive tactics. How so? Specifically 00:32:00.670 --> 00:32:03.329 abusing its monopoly power in certain chip markets 00:32:03.329 --> 00:32:06.210 by employing restrictive contract terms. What 00:32:06.210 --> 00:32:07.890 exactly were the restrictive contract terms? 00:32:07.880 --> 00:32:10.440 terms the FTC alleged. The allegation was that 00:32:10.440 --> 00:32:12.960 Broadcom would threaten retaliation against customers 00:32:12.960 --> 00:32:15.759 deemed disloyal if they dared to buy competing 00:32:15.759 --> 00:32:19.119 components from rival chip makers. The terms 00:32:19.119 --> 00:32:21.920 allegedly included effective exclusivity arrangements, 00:32:22.279 --> 00:32:24.559 making it practically impossible for smaller 00:32:24.559 --> 00:32:26.960 competitors to gain market share, even if they 00:32:26.960 --> 00:32:29.140 had better or cheaper components. And what was 00:32:29.140 --> 00:32:32.019 the outcome? Broadcom eventually agreed to a 00:32:32.019 --> 00:32:35.500 settlement with the FTC in 2021. And the European 00:32:35.500 --> 00:32:38.000 Union was running a similar investigation simultaneously. 00:32:38.039 --> 00:32:40.339 That's right. The European Commission's investigation 00:32:40.339 --> 00:32:44.160 ran from 2019 to 2021. They ordered Broadcom 00:32:44.160 --> 00:32:46.980 to cease these allegedly anti -competitive practices. 00:32:47.690 --> 00:32:49.970 with European Competition Commissioner Margrethe 00:32:49.970 --> 00:32:52.630 Vestager, stating that Broadcom's behavior could 00:32:52.630 --> 00:32:55.529 cause serious and irreversible harm to competition 00:32:55.529 --> 00:32:57.470 and innovation in the market. And they also settled. 00:32:57.799 --> 00:33:00.039 The company ultimately settled with the EC as 00:33:00.039 --> 00:33:03.299 well, agreeing to suspend exclusivity or quasi 00:33:03.299 --> 00:33:05.660 -exclusivity arrangements and committing not 00:33:05.660 --> 00:33:07.640 to enter into such restrictive agreements for 00:33:07.640 --> 00:33:10.680 a period of seven years. It shows a pattern of 00:33:10.680 --> 00:33:12.700 aggressive contracting that reflects a company 00:33:12.700 --> 00:33:15.140 determined to use its market power to maintain 00:33:15.140 --> 00:33:18.019 dominance, regardless of the regulatory pushback. 00:33:18.619 --> 00:33:21.220 Finally, let's look at the patent lawsuits. Broadcom 00:33:21.220 --> 00:33:24.039 files many, but the high -profile case against 00:33:24.039 --> 00:33:27.200 Netflix in 2020 highlights a deeper, more defensive 00:33:27.200 --> 00:33:29.839 strategy tied to protecting its old business 00:33:29.839 --> 00:33:32.519 model. The 2020 suit against Netflix alleged 00:33:32.519 --> 00:33:34.799 multiple patent infringements related to video 00:33:34.799 --> 00:33:37.900 streaming technology. Now, Broadcom often files 00:33:37.900 --> 00:33:40.539 suits against smart TV manufacturers and the 00:33:40.539 --> 00:33:43.220 USITC often ruled against Broadcom in those cases. 00:33:43.380 --> 00:33:45.839 But the Netflix case was different. Different 00:33:45.839 --> 00:33:47.940 how? Because the underlying motive wasn't purely 00:33:47.940 --> 00:33:50.500 about patents. It was about defending a legacy 00:33:50.500 --> 00:33:53.140 cash cow. Explain that underlying context. Why 00:33:53.140 --> 00:33:55.960 would Broadcom sue Netflix? Because Netflix's 00:33:55.960 --> 00:33:57.980 success in driving cord cutting was directly 00:33:57.980 --> 00:34:00.599 hurting Broadcom's traditional, highly profitable 00:34:00.599 --> 00:34:03.279 customer base, the cable industry. Of course. 00:34:03.500 --> 00:34:06.339 Broadcom is a major supplier of set -cop box 00:34:06.339 --> 00:34:08.860 chips and related components to cable and pay 00:34:08.860 --> 00:34:11.739 TV providers. So critics argued that Broadcom 00:34:11.739 --> 00:34:14.019 was effectively suing Netflix for its disruptive 00:34:14.019 --> 00:34:17.119 success. So the loss of pay TV subscribers wasn't 00:34:17.119 --> 00:34:18.980 just a headache for the cable companies. It was 00:34:18.980 --> 00:34:21.420 a revenue threat to Broadcom's hardware division. 00:34:21.679 --> 00:34:24.840 Exactly. And the data confirms the threat. Major 00:34:24.840 --> 00:34:28.019 U .S. pay TV providers lost nearly five million 00:34:28.019 --> 00:34:32.380 net video subscribers in 2019 alone. Those subscribers 00:34:32.380 --> 00:34:34.460 were moving to streaming platforms like Netflix. 00:34:34.699 --> 00:34:36.900 And the money in set -top boxes is significant. 00:34:37.219 --> 00:34:39.400 The cable industry generates billions of dollars 00:34:39.400 --> 00:34:41.840 in profit, approaching $20 billion annually, 00:34:42.179 --> 00:34:44.659 according to some estimates, primarily from the 00:34:44.659 --> 00:34:46.860 mandatory rental fees for those set -top boxes, 00:34:47.019 --> 00:34:50.219 which use Broadcom chips. By suing the streaming 00:34:50.219 --> 00:34:52.699 provider, Broadcom was essentially using its 00:34:52.699 --> 00:34:54.920 patent muscle defensively to slow down the erosion 00:34:54.920 --> 00:34:57.400 of the legacy cable business that provided stable 00:34:57.400 --> 00:35:00.000 hardware sales. It's a clear example of protecting 00:35:00.000 --> 00:35:02.539 the old income stream while simultaneously building 00:35:02.539 --> 00:35:05.039 the new one in software. This has been an incredibly 00:35:05.039 --> 00:35:07.960 complex journey. When we synthesize Broadcom's 00:35:07.960 --> 00:35:10.639 story from its HP roots and the political battles 00:35:10.639 --> 00:35:14.019 over Qualcomm to the strategic pivot into software 00:35:14.019 --> 00:35:17.440 and the acceleration of AI chips, the narrative 00:35:17.440 --> 00:35:21.519 is one of unrelenting discipline strategy. The 00:35:21.519 --> 00:35:24.309 synthesis is clear. Broadcom's modern dominance, 00:35:24.510 --> 00:35:26.750 its status in the Magnificent Seven, and its 00:35:26.750 --> 00:35:29.429 trillion -dollar valuations stem almost entirely 00:35:29.429 --> 00:35:32.969 from a strategic, high -stakes series of multibillion 00:35:32.969 --> 00:35:36.750 -dollar acquisitions. LSI, CA, Symantec. LSI. 00:35:37.150 --> 00:35:39.650 CA technology, Symantec, and especially VMware. 00:35:40.150 --> 00:35:42.829 This was the master plan to transition the firm 00:35:42.829 --> 00:35:45.449 from a specialized component supplier to an essential 00:35:45.449 --> 00:35:47.510 provider of both the high -speed underlying hardware 00:35:47.510 --> 00:35:50.210 and the critical entrenched enterprise software 00:35:50.210 --> 00:35:52.510 that powers the digital infrastructure. And here's 00:35:52.510 --> 00:35:54.150 where it gets really interesting for you, the 00:35:54.150 --> 00:35:56.869 learner. Broadcom's story demonstrates that in 00:35:56.869 --> 00:35:59.090 modern tech, sometimes the greatest growth is 00:35:59.090 --> 00:36:00.949 achieved not just by creating the next great 00:36:00.949 --> 00:36:03.730 chip, but by successfully acquiring and rigorously 00:36:03.730 --> 00:36:05.929 streamlining the entrenched infrastructure needed 00:36:05.929 --> 00:36:08.460 to run. the entire digital economy. Even if that 00:36:08.460 --> 00:36:11.860 means navigating massive regulatory hurdles and 00:36:11.860 --> 00:36:14.519 shifting the company's entire identity from hardware 00:36:14.519 --> 00:36:17.199 to hybrid. Exactly. And what's fascinating here 00:36:17.199 --> 00:36:19.960 is that the growth is tied to two powerful but 00:36:19.960 --> 00:36:23.199 very concentrated forces. The revenue stability 00:36:23.199 --> 00:36:26.000 from legacy software assets acquired for cash 00:36:26.000 --> 00:36:29.440 flow and the future growth tied to highly customized 00:36:29.440 --> 00:36:33.420 XPU chips designed for a handful of massive hyperscale 00:36:33.420 --> 00:36:37.050 clients. Google. Meta, and OpenAI. Just a few. 00:36:37.389 --> 00:36:39.409 This raises an important question for you to 00:36:39.409 --> 00:36:42.170 mull over and explore further. Given that Broadcom's 00:36:42.170 --> 00:36:44.449 model relies so heavily on acquiring established 00:36:44.449 --> 00:36:46.849 giants and immediately prioritizing margin and 00:36:46.849 --> 00:36:49.010 scale over nurturing fundamental R &D in those 00:36:49.010 --> 00:36:52.030 acquired companies, what potential long -term 00:36:52.030 --> 00:36:54.449 risks does reliance on such an aggressive acquisition 00:36:54.449 --> 00:36:56.730 strategy pose? Right. I mean, if the acquired 00:36:56.730 --> 00:36:58.829 software assets eventually become technologically 00:36:58.829 --> 00:37:01.469 obsolete, or if one of those massive XPU customers 00:37:01.469 --> 00:37:03.710 decides to move fabrication or design entirely 00:37:03.710 --> 00:37:05.940 in -house, What does that mean for a company 00:37:05.940 --> 00:37:07.920 built primarily on integration and consolidation 00:37:07.920 --> 00:37:10.739 rather than singular internal innovation?