Acquired 專訪台積電 93 歲創辦人張忠謀，關於黃仁勳、NVIDIA、蘋果、高盛、英特爾、三星、IBM 與高通的故事

Want us to be about technology companies again. Well, this is a technology company. It's a sign. All right, here we go. Who got the truth? Is it you? Is it you? Say it straight. Another story on the way. Got the truth. Welcome to the Spring 2025 season of Acquired, the podcast about great companies and the stories and playbooks behind them. I'm Ben Gilbert. I'm David Rosenthal, and we are your hosts. Today we have something very special to share with you. After becoming obsessed with semiconductors from our TSMC episode four years ago, David and I wound our way through the rest of the Hail Mary. We asked friend of the show Jensen Huang if he would ask Dr. Morris Chang, the 93-year-old founder of TSMC, if he would be open to an interview with us. Yeah, it is kind of insane and super cool that Jensen made time to help us with this. It's not like he doesn't have a lot of other things going on. Yes, well listeners, it happened. So today's episode is a conversation that we recorded in Taipei last week at Dr. Chang's office. We flew to Taiwan for a 48-hour whirlwind where we spent some time at TSMC's headquarters in Hsinchu Science Park, where many of TSMC's fabs are located. Super cool to see. Totally. So conveniently, Dr. Chang just published Volume 2 of his autobiography.

It focuses on a few crucial stories from TSMC's history that Dr. Chang shares in his memoir about Apple, Nvidia, and the birth of the fabless industry. Yes, and big thank you to Karina Baa, who we were lucky to connect with after we set this up and who has been translating Morris's memoirs with funding from Tyler Cowen and Emergent Ventures. Right now the memoirs are not published in English, and we will let you know if and when that happens. Yep. All right listeners, you can join our email list at acquire.fm/email. You'll get an email every time a new episode drops once a month, and this is also where we announce past episode corrections, plus a fun little game where we give hints at what the next episode will be. I always have fun writing those. You do. That's a clear David job. This episode is presented by our partners at JP Morgan Payments.

Companies that we discuss and this show is for informational and entertainment purposes only. Please enjoy this conversation with Dr. Morris Chang, with some of David and my reflections following its conclusion. We thought as a fun way to start things off would actually be to talk about the man who introduced us. Could you tell us a little bit in your words about your relationship with Jensen and TSMC's special relationship with Nvidia?

Was that chip the Riva 128?

So there was a great partnership forged there. TSMC would fab the chips, manufacture them. Nvidia would design them. That is true all the way to today at immense scale. But it hasn't always been easy and it hasn't always been perfect. I want to go to this moment in 2009 when you had stepped down as CEO to give the CEO job to a potential successor while you still retained the chairmanship. In Taiwan, usually the chairman is the top man anyway, even though the CEO is another person. The problem you just mentioned happened during the period when someone else was the CEO. Apparently it was a manufacturing problem. It was a 40 nanometer problem with Nvidia. A director of quality insisted that TSMC was not at fault. On that basis, the manager arguments he had not offered Nvidia anything. As far as the manufacturing problem was concerned, it was a yield problem and everybody was suffering from it. Nvidia at that time was perhaps the biggest customer of that node. The problem apparently continued. Even though I was not the CEO, I was getting a little impatient. Then some other problems cropped up other than this 40 nanometer Nvidia problem. So I decided to take the CEO position back. In 2009 I did that. There were several priority problems. I remember in the first few days after I took back the CEO ship, I called all the major customers including Jensen. Qualcomm was also in that call. The top customers didn't change much except for maybe one. Apple came later. In my call with Jensen, I said I knew about the manufacturing problem on the 40 nanometer. I said it's one of my priority problems, but give me a couple weeks and I will get back with you. I did have several problems aside from the 40 nanometer manufacturing problem and the argument we were having with Nvidia. One problem was that we had committed to a schedule of price drops with customers, but we weren't able to drive down our manufacturing costs at the same rate. Another problem was the immediate one that triggered me to retake the CEO ship. The previous CEO had laid off people, except he didn't use the term layoff. He used the term performance review. The worst performance review people, there were about six or seven hundred of them, and he laid them off on the basis of their poor performance. Under your watch, you never did a layoff and you never looked at performance reviews which are meant to be a tool for development. In 2008, of course there was a financial crisis and the semiconductor business got affected. Our revenue dropped, our business dropped pretty seriously. I was not the CEO, I was the chairman, but I just knew that any general manager or CEO without much experience would lay off people in a situation like that. But I knew that in the semiconductor industry and Moore's Law means no matter what happens, you will always need people. Semiconductor people actually think the same way as I do. They all lay off. They are all people too. I had a lot of experience at Texas Instruments, but at Texas Instruments I was not the CEO, I was just one of the top managers under the CEO level. When the company decided to have a layoff, the CEO conferred with the top managers. The method was exactly the same as what our TSMC CEO did in late 2008, 2009, which was go by performance. I was the only one at Texas Instruments in the early 70s that said that would not be a credible way of doing it. People would not respect us if we lay off by performance. The 700 supervisors who gave the bad ratings, it's very subjective. It's not something that people will respect. If in a year you have to hire people back, you have to hire the laid off people back, then you shouldn't lay off because the separation expense is usually about half a year, and it takes at least half a year to train a person. So if you need the people back within a year, you shouldn't lay off. So what did you do? The laid off employees, as I said there were 700 of them, six or seven hundred, they came to my home to demonstrate and protest. The company TSMC was pre-warned that hundreds of people would appear in front of my home, so they notified the police department in my district. The police department sent 50 or 60 police officers to try to maintain the order. The problem was not solved. I was still not the CEO. So they appeared again. Some protesters, about 25 of them, decided to spend the night, sleep over in the little park that's about a block away from my home. My wife literally didn't sleep that night. She woke up and went over to the window to take a look to see what was going on. Then very early the next morning, my wife at 6:00 AM, she went to the market and bought breakfast. She bought bread, fried bread, buns, soy milk, and took enough breakfast for 25 or 30 people. She went back to the park and distributed it to the protesters. They were thankful. They actually decided to leave. This precipitated my taking back the CEO job. There's another thing. I told the previous CEO before he laid off the 600 or 700 people, I said if you want to lay off, you need my permission. But he decided to circumvent that. The COO, what he did, he did not consider it to be a layoff. It was just punishment for the poor performers. As far as the CEO is concerned, I didn't keep him. I had more than one nice talk with him. I intended to and I told him that he was still a potential successor to me. But I put him in charge of new businesses. Back then we had high hopes for the so-called new businesses, which was solar cells and LEDs.

It's the great irony that your core business of manufacturing integrated circuits ended up becoming the largest market opportunity of all. You didn't need any new businesses. The biggest market opportunity. Why is it so ironic?

In 2009, you rehired anyone who was laid off who was interested in coming back, and you set the new vision and strategy as CEO, in many ways returning to the old one. How did you resolve the Nvidia dispute?

It's an amazing example of a situation where you had a strong partnership together for many years, you built this close personal relationship such that you could have an hour and a half family dinner and not talk business. You were able to then come up with a large sum of money, over a hundred million dollar settlement, and since then there have been many, many billions of dollars of business done together. It's a great story.

This episode is presented by our partner JP Morgan Payments. You don't often hear stories of how important payments are to supply chain success, but the last few years have really shown how critical it is to the global economy. In our remastered TSMC episode, you heard how JP Morgan's trade and working capital group helps companies manage and access liquidity for their global trade. We just got a new stat from them: 18 of the top 20 largest corporates are payments clients, so they have an unparalleled view into global trade flows. Zooming in on supply chain, think about a typical manufacturer. They're trying to balance paying their suppliers quickly to keep good relationships while also optimizing their own working capital. Meanwhile, their suppliers, especially the smaller ones, often struggle with cash flow while they wait 30, 60, 90 days to get paid. With early payments at favorable rates based on the buyer credit rating, buyers can easily switch between supply chain finance and dynamic discounting as business needs change. It's also real time. You get complete visibility into supply chain finances data so you can track invoices and optimize payment timing all in one place. To simplify it, here's an example. Imagine you're a lumber mill working with dozens of furniture manufacturers. The terms you've negotiated with a big furniture manufacturer mean you have to wait 60 days for payment on multi-million dollar monthly invoices. Of course this limits your ability to maintain optimal timber inventory and process orders quickly. With JP Morgan supply chain finance solutions, you could receive payment in just days while the furniture manufacturer gets to keep their original payment terms. So now you can stock up on timber and fulfill orders faster. JP Morgan is at the forefront of both the financial and operational aspects of supply chains with trusted, innovative infrastructure that you can rely on. So whether you're looking to strengthen your working capital or just make sure your supply chain is more efficient, check out jpmorgan.com/acquired and see how they can help your business grow.

After the 40 nanometer node, after you fixed these problems, as you said the next node was 28 nanometers. As we understand your story and the company's story, 28 nanometers is when TSMC really started to take the leadership role at the leading edge in the industry. How did you decide to commit so hard to 28 nanometers after having had all the problems?

The smartphone era coincided with 28 nanometer. When the business development group was looking at this and you were looking at this, did you see how big smartphones were going to become and the immense opportunity that that would unlock for you?

Before we do that, let me offer how we made CC, actually the business development director, the current CEO, the current chairman and CEO. When Rick was the CEO between 2005 and 2009, he had split operations into two groups: the advanced group and the mainstream group. Actually, I should say the lesser one. Mark Liu was the head of the advanced group. Each group had a small business development section, maybe 30 or 40 people each. So I came back to be the CEO and I never thought the split up of two groups was a good idea. In 1996, the president, who was American, Don Brooks, wanted to split. I think he got a little tired of running this company. He was going to be here for only a year at first, but he ended up staying six or seven years in Taiwan. Towards the end, he was getting a little tired of running this thing. He thought he would do it like TI, for instance, when TI had a germanium transition department, silicon department, etc. But TSMC, as a pure-play foundry, was not suitable for the divisional structure because we have almost the same group of customers. How do you divide up that group? Don Brooks was going to divide it by fab. But customers move from one fab to another, the same customers. Not to mention TSMC had a consultant, McKinsey. We got McKinsey in. McKinsey, after a month or two, actually a couple million dollars, told us the same answer: functional is best. Then Don Brooks said, "Tell me one company, one big company that's functionalized." McKinsey immediately answered Boeing. Boeing is a good answer. They make 747, 757, 737. They don't divide by fab. If we divided by fab, it would be like dividing up 707 from 757, 737. Don Brooks' attempt was in 1996. By 2005, Rick decided to tread the same ground. He did it. This time I didn't stop him. Of course, not if the whole company is going down the drain. You have to interfere then, but only then. So that was the background. Two groups when I came back to be the CEO: the advanced group and the mainstream group. Each group had a small business development section of 30 or 40 people. I think the advanced group had more. I decided to combine the two groups. I didn't call it marketing because I decided to use "business development" in English because it has a good translation in Chinese. In 2009, when I decided to combine the two groups, the advanced group had something like 10,000 employees and the mainstream group had something like 10,000 employees.

In technology leadership, you're saying that around this time, the 28 nanometer node, you were talking about 2010. You were still among a select few at the leading edge.

Well, take us there now. So after November of 2010, you had this dinner with Jeff Williams. He said he would let you have 40% gross margin. You thought to yourself that he was being generous when he said that. You also thought that this dinner was not the time to go into a pricing discussion. You had a lot of other things to discuss. You said you were about to go into production with 28 nanometer at that time. You thought it was going to be 28 nanometer. But frankly, it was also a disappointment because the Moore's Law progression after 28 was going to be 16. Apple, Jeff Williams, wanted a 20, a half step. A half step is a detour. We would have to spend effort on the 20, which would help us on the natural progression to 16. But the point is that back then, our R&D did not have enough resources to do two nodes at the same time. Later we did. So you have this conundrum. This is right after you had just spent $6 billion in capex the previous year going all in on 28 nanometer. You're asking Apple, which could be your biggest customer ever, this is for 28, and you hear back, "No, we want you to go do something that you're not planning on."

It seems like it would be in a customer's interest to come to you and say, "I need to buy zillions of chips from you. I need all your wafers," because they have no skin in the game of you spending all the money.

Listeners, this is a great time to reintroduce a good friend of the show, ServiceNow. Last year we shared their incredible story from founding through becoming one of the best performing public software companies in history. Well, today we want to share a more recent story. In November of 2022, ChatGPT launched. Everyone started using this one cool function thing with AI to get more done faster. But then again, predictably, all of this just created what Bill McDermott, the CEO of ServiceNow, calls the hornet's nest of complexity. Businesses already had too much software in different departments doing different things, and now all of a sudden they had too many different AI tools which just compounded the problem. Ultimately, AI is powerful, but it's only as powerful as the platform and data that it's built on. Fortunately, not only is ServiceNow a powerful platform, Bill had foreseen the coming importance of AI. As some of his very first moves as CEO in 2020, he took all the R&D work that ServiceNow had been doing for years in AI and elevated it within the company. They also bought a company called Element AI founded by a Turing Award winner. Over the next few years, they kept building. Today, ServiceNow has AI agents that you can deploy across every corner of your business. They all work with each other, architected on one enterprise-grade platform and built on the same data.

Workflows so whether it's your IT or HR or Finance CRM supply chain etc you can deploy AI agents in every part of your company. David it's funny you hear all these Fortune 500 CEOs talking on earnings calls about the AI agents that they're deploying to increase productivity behind the scenes a huge number of those are actually doing this with ServiceNow.

Yeah it's incredible so if you want to bypass the hornet's nest of complexity caused by disparate software vendors and put AI to work on one platform and in every corner of your business go to servicenow.com.

I mean this is why you started TSMC as a pure-play foundry business because you didn't want to compete head-to-head you said we should not be an integrated design manufacturer of the design of the chips and the manufacturing we have to compete on a different vector we'll never catch Intel.

So today Intel doesn't make the chips in the iPhone what happened and in fact TSMC makes all of Apple's chips.

Your interpretation of the meaning behind that statement?

So to finish the Apple story the short answer is it worked on 20 nanometer were there any tradeoffs where pursuing 20 nanometer?

Could you tell us a little more about what goes into considerations around pricing I imagine things like the yields you think you'll be able to get hugely impact that.

Now I think you were asking whether there was any tradeoffs.

Yeah so I would feel away and I said you know we invested in all these equipment and we were counting on you to take the 16 from us but now you know we found out you were buying 16 at the first 16s anyway from Samsung so Jeff replied immediately don't worry I'll be here I'll be there I'll be in San Jose next week and explain to you.

And the pure-play foundry business model Samsung is Apple's cheap competitor.

Yeah it seems so fundamental to the industry that you want to get through the low volume period as fast as you can ideally you spend no time in the low volume period.

All right listeners now was a great time to thank friend of the show Fundrise. We've gotten to know Fundrise CEO Ben Miller and the folks there quite well over the last several years and they're huge Acquired listeners just like all of you and since we first worked together three years ago Fundrise itself has gone through quite a transformation. Longtime listeners may remember that they have a growth stage venture that they actually first launched here via an Acquired sponsorship back in 2022 at the time Fundrise was primarily known as the US's largest real estate investment platform they could break into the venture industry in a big way. Ben Miller and Fundrise have invested in Databricks Anthropic Canva Anduril Ramp and fellow friends of the show Vanta and also ServiceTitan which just went public in December in a successful IPO.

Yep it's genuinely awesome what Fundrise has done here which is something that many have tried over the years but no one else has actually been able to accomplish in venture. They've taken a retail platform that any American can invest in and gotten pre-IPO access to some of the best private companies in the world. It's democratized access to all the value creation that otherwise has been locked in these private companies over the last decade plus as these growth companies are delaying IPOs and staying private longer.

Yep so when the ServiceTitan IPO happened thanks to Fundrise tens of growth stage founder looking for a great Series C or later investor just get in touch and tell them that Ben and David sent you. As our time comes toward a close one question David and I wanted to ask you is TSMC is essentially the only trillion dollar company in the world not on the west coast of the United States it is this incredibly important thing in the world it's this unlikely success of grand scale unlikely in your opinion?

Here's why I say unlikely world and you were taking their worst you know excess you were their second source supplier for manufacturing on the stuff that they didn't want to make on their own did you see fabless coming or was that a very lucky thing?

Well at least I had a glimpse of it you know.

Well we can't thank you enough Dr. Chang.

Well David and I are coming at you now from our home studios back in Seattle and San Francisco and we wanted to do a little post game on that interview a little bit of analysis kind of our conclusions the things that are still sitting with us a few days later after we've crossed the ocean and David this felt essential to me because it felt like we were just recording history there with Morris I didn't want to interrupt him to try to make a business model point or it just kind of felt like we should let him talk and then we could do our part after.

Yeah totally and fortunately we have a model for doing analysis at the end of story which is play our Playbook so let's do it.

Okay so the first thing in the TSMC museum of innovation they have Morris's original pitch like his little original slide original business plan that he pitched to the Taiwan government and then to investors there's like two different versions of this extremely simple pitch deck and one of the bullet points it's right in there of be a dedicated pure-play foundry at the time I get the sense it was actually much more about what can we win at versus what will be the most important and valuable semiconductor company in the world in the future right at the time they didn't have the capabilities certainly not TSMC and didn't exist in Taiwan to be able to design chips and products so like it was impossible for them to compete with like an accident of history the pure-play foundry ended up being the best way to do this I guess best as evaluated on market cap versus other foundries and integrated device manufacturers such as Intel.

Well and best that like this is the path that has led them to being essentially alone operating at the leading edge like they have surpassed technology-wise all of the other integrated and quasi integrated chip foundries out there.

Yeah I guess that's my first thing is this connect you can connect the dots looking backwards as Steve Jobs said and that famous quote but forwards is difficult this primarily I think was the main reason why TSMC has worked so well that they do not compete at any other part of the value chain with them right.

Okay so if you're asking yourself how did the world arrange itself in this way such that you could have a trillion dollar company that doesn't do any design that doesn't do any architecture that doesn't do any EDA tools like Cadence or Synopsys so they're not Nvidia they're not Arm they're not Cadence Synopsys they're not ASML like they're not their own equipment vendor so what enabled this one of the things that I think is underappreciated and we didn't talk that much about with Morris but the rise of Arm if you try to play forward a world where Intel and with Renee from Arm this architecture that became dominant in phones and then computers and then servers and now is coupled on with all these AI chips you open the door to have a dedicated foundry for Arm chips in a way where if it had stayed x86 it's not like you could start a new foundry for all the fabless x86 companies for the longest time Intel was the only x86 company and then AMD of course is the second source and AMD is a TSMC customer so that's sort of the one edge case is like well there is AMD that designs x86 chips that TSMC manufactures but that's not like the common case of the way it would have gone for an in an x86 dominated world it would have been fully platform and Apple hadn't standardized their Apple Silicon on Arm probably Intel would be making all of the chips that go into your iPhone all the leading edge chips that go into your iPhone like they already had the Intel relationship Mac were running on x86 Intel chips.

Yeah you have to keep peeling the onion because this of course supposes that Intel actually could have gotten their act together and made a chip for mobile phones that was performant but maybe all the baggage from x86 actually prevented them from structurally doing that it wasn't like a competency thing it was like that it never could have happened that x86 could run on phones.

Yeah I think all this is true but if Arm hadn't existed like are standalone designers you know Apple Nvidia in a large part that's due to Arm yes and Arm and TSMC are sort of coupled at the hip of history of when this how this came to be in fact didn't you find that a bunch of these were started within 12 months of each other?

Yes totally the mid to late 80s were like an absolute golden period for all these companies getting started not only TSMC Arm Synopsys Cadence and ASML all founded right within a couple years of each other which brings us to Hsinchu Science Park going there in person we talked about this on our original TSMC episode that you know even if you wanted to you couldn't airlift TSMC and this you know kind of government sponsored Industrial Park which it sort of was it was Silicon Valley you know as we talked about in our Lockheed Martin episode oh the early Lockheed yeah the early Lockheed years but that's what it's like today it's all right there it's not just TSMC that's there it's all of their partners it's all of their customers you know we're driving by and this is a Cadence building there and that's a Synopsys building there and that's an Arm building there there's Qualcomm there's MediaTek right there headquartered right there and right across the street the craziest thing to me we saw there are two universities that are just like there in the Science Park yes that are cranking out PhDs every year that are just getting absorbed right there in the ecosystem I other that you know Synopsys has to be closely tied with TSMC to understand what the next node will look like so that they can make it easy for people who are using Synopsys tools to design chips to actually manufacture using TSMC's process you kind of get the sense of oh I see because they all are walking across the street to each other and having this extremely close communication not to mention David both of our flight experiences kind of felt like oh there are a bunch of chip design fabless companies that are making the pilgrimage over to Taiwan to meet with people in this ecosystem my plane felt like the semiconductor version of the tech buses and the sort of outside of Taiwan Fabs you know why is TSMC doing it because it's not their leading edge it's not volumes it's not leveraging this really close geographic ecosystem that they have in I believe there's three science parks in Taiwan yes we saw the original but there's one that's even bigger I think it's the Tainan one in the South but it just kind of becomes clear that there are customers and government reasons to build Fabs in other countries but you're not going to be able to recreate the magic of that ecosystem like physically instantiated right there.

Yeah it would take decades initiative of all time like anywhere in the world at least to spur innovation with this particular of a mandate totally the land grant universities here in America but like this was like a rifle shot like you know we are going to spur semiconductor industry innovation in this industrial park in this location and it worked and there you have one of the 10 most valuable companies in the world and the only I guess one of two trillion dollar companies that are not on the west coast of the United States I would say it worked.

Yeah it worked it worked and the scale too we drove by a construction site where it looked like a quarter of the building was done this is where process I wonder what they're going to make on that lots of Nvidia GPUs and lots of iPhone chips massive building phase one was open which I think is a quarter of the building but then there's three other phases for this 2 nanometer facility that are not even ready for prime time yet but I think they're actually doing the small production runs getting ready to ramp in the second half of this year on the 2 nanometer process like you said the scale of the physical buildings of these Fabs smacked me in the face I felt like I was looking at a sphinx in Egypt I mean like it's huge it's like many football fields of size just per phase of the Fab these are taking the excess capacity from Intel and other IDMs and giving them a place to manufacture their least critical least leading edge least interesting chips but that is true I mean he believed that fabless was going to be a thing but for the first I don't know at least five years the only real business that they had was IDMs who were willing to say how cheap can you give me some of your manufacturing capacity and it's not strategic at all but here you go here's some revenue this is a major difference in Intel's Fab strategy versus TSMC Intel is constantly taking their existing Fab footprint and repurposing it and processed node generations and it's not like demand goes away for those chips and those products it does it just does slowly.

It does slowly yeah and I mean like replacement parts is a great example like you know there are technology systems and products manufacturing things even automobiles built 10 20 30 years ago that have specific chips that were made with old process technology that when they break and they need replacing like you need those exact same chips so this is the business that TSMC started in right so that is the fundamental philosophical difference is I think Fab so Fab 1 belonged to ITRI the government where Morris was president applications for older nodes if you're in this world of 40 nanometer and up and you know one micron and I don't know all the names of the previous generations but the less high resolution etching on silicon CMOS sensors are a great example of that the cameras that we're talking into right now that have these great Sony sensors those don't require a 2 nanometer process but they do require etching the same way that you would etch a chip and so that's a specialty use case of TSMC's older Fabs which by the way on an accounting basis are fully depreciated so they're almost like free to run right right all the capital expenditure now there's maintenance capex that needs to go into different one Intel is going to keep closing the old stuff so they can own a smaller footprint and keep all the equipment and everything focused on making the latest greatest just not what TSMC does.

Totally totally but that point of I'm obsessed with this idea that it was funny that Morris went on the record and said no I knew I knew fabless was coming and he had a couple great anecdotes about that which is funny because in older interviews sometimes he goes well the timing was a little lucky on when fabless happened but I think he even said to Jensen in the first few years of TSMC growth wasn't very high because we were waiting for the customers to emerge but it really is this idea that he saw the future he made a happened and I don't know you know I think he was within 12 months of when he thought it would happen but it is crazy that when especially in his memoir you're reading the story about the early customers year five year six year seven the majority of the business is still not fabless it's someone else's worst orders which that actually gets to the heart of learning curve pricing that we spoke about with Morris we brought it up sort of tangentially with him but it's probably worth dwelling on what is the learning curve.

Yeah the core insight of the learning curve from BCG and Bill Bain and Bain and and Morris that they all developed together which by the way how crazy is it the founders of BCG and Bain are the ones who sort of that as a given of like if we get to be the largest volume player this is a fixed cost business this is a scale economies business we can spread that fixed cost over the maximum number of customers how do we get to the maximum number of customers in the early stages of the game where it's more competitive we accelerate the pricing to where we think it will get to at the end of the game so that's why doing these price cuts and also starting low with your prices like you can even start unprofitable with your prices in the early days in a given node generation because the goal is to outcompete the competition become the industry dominant number one player get all the customers once you node what do we think demand will be for you know call it 40 nanometer how many orders of individual chips will there be in 40 nanometer okay well to have the cheapest price for customers we need to do the biggest ordering and so then it's just a matter of like how fast can we get into volume production everyone sort of intuitively grasps this oh economies of scale but the implications across your whole business your pricing strategy the way like strategic finance how do you when do you decide to take on debt when do you not when do you decide to take on more shareholders it's this incredible orchestration to make it happen you know it's almost Costco like which the learning curve would tell you you want to get the deepest down the learning curve possible we should go take all their order but you all that kind of exposes you to existential risk in your business when you're not within spitting distance of doing that volume on your own so is it really worth betting the entire company you got to be so precise and accurate in your forecasting of the ultimate market demand which means the ultimate demand for your customers products which in the Apple case means ultimately forecasting accurately how many customers are going to buy the next generation iPhone in order to run your business right or in Nvidia's case how big is AI going to be you know these are this is kind of a crazy thing for a manufacturer to have going to tank your entire profitability for that node generation which is going to tank your free cash flow which is going to mean you can't play the game in the turn to this point though if you actually are good at all of this and you are good at forecasting and the execution is flawless once you internalize the learning curve the story of TSMC goes from one where it's surprising and unlikely and it becomes an inevitability of course the company that is taking on all the orders to have the lowest prices right of course this will be the end state of this industry is to have a dominant player like right now it costs market has natural monopoly characteristics.

Yep yep and that's just the capex side of the equation as we talked about with Dr. Chang there's also the R&D side of the equation that needs to go into creating the next process node that can be built on that capex yeah it is crazy that if you just look at every year the capex versus the net income of this company they basically spend all the money not all the money but their capex grows in a very similar way if you look at the bar graph to their net income from the year and so that is even before R&D David to your point if they're looking around at competitors at for their manufacturing processes and that's how you get CoWoS which is the technology that they use for packaging for AI chips that's their proprietary thing which by the way once you have proprietary packaging then it's even harder for customers to go and double source manufacture elsewhere they have a similar technology for packaging of mobile chips that doesn't use CoWoS but it seems like this is a market where those in the lead are only going to get further in the lead over time absent some big strategic mishaps or some big execution mistakes.

Yep totally and then I think the last Playbook theme here for me and for us is just that circuit was invented the compounding growth of that industry is all that mattered everything else is just downstream of the fact that the world is going to demand more computing at this monotonic exponentially increasing pace every 18 to 24 months you know and of course the technical definition of Moore's law expired a long time ago but like spiritually the world demands roughly 2x the computing power that it had two years ago every two years and that has continued for 50 60 years at this point and shows no signs of slowing down and as a result well no the innovation side of getting better at design and manufacturing and that is getting harder than ever because we're having to like call more things Moore's law you know packaging was never a part of the original Moore's law and software improvements and proprietary interconnects my point is that it's a self-reinforcing system as long as the demand is there that the world wants twice as much compute as it had yesterday there are going to be market incentives to drive the supply side and that is why people work so hard to make it happen.

All right here's the stat since TSMC was founded in 1987 the world's semiconductor market has grown you could build a trillion dollar value foundry yep the scale of the numbers are so staggering I keep thinking about the fact that they can go spend $20 billion to build a building and the stuff that they spit out is so valuable that that $20 billion was a profitable investment in a matter of I don't know how many years if it's three five seven whatever the payback period is like they know for sure that it's a worthwhile investment to do that.

Yeah the whole thing comes down to oh my God silicon has become really valuable like integrated circuits are the fabric of our world today.

Ah well Ben what an amazing kind of a hilarious I can't believe it's 2025 and this is my recommendation for any who's not a AAA member I highly recommend it oh I had a spectacular AAA experience where I went to fill up the air in my tires before a road trip and I went to the gas station and there was something wrong at my local gas station with their pump and I ended up draining the air in my tires to an unsafe level and so the car was actually not drivable away from this gas station and I was like crap I can't even go get the other car and I had my baby in the back seat and my wife and I are trying to figure out what to do and on a long weekend like a holiday weekend when other people aren't working drive out and fill up the air in my tires so we could be quickly on our way not ruin the weekend amazing and it was like a hundred bucks or something it's really not a bad price and you get this was see $100 to become a member whatever is $150 and then the service is actually free for something as trivial as this and you get three of them a year wow so I'll take it it was a phenomenal experience all right Triple A there we go.

My second one is a YouTube channel called Defunctland you and I were talking about this oh yes this is so good you turned me on to this yeah it is an entire YouTube channel that I actually haven't watched in a while but I only remembered it from our conversation and now I need to go back crazy stuff some of the most crazy entrepreneurs and executives within companies decided to build theme parks and it is very fun to see the weird old Nickelodeon hotels or Action Park in I think it's New Jersey the like wildly unsafe park from the 60s 70s and 80s ah man those were the days yes you could get lost for hours and hours and hours watching Defunctland so I highly recommend the YouTube channel I'm really glad that you and I grew up as kids in the era where we could still take unreasonable amounts of risk and nobody thought that there was anything wrong with that.

Yeah oh my carve out speaking of it being 2025 how are we talking for movies it's so good it's so good I think this was your carve out when it came out a couple years ago just so so so good truly enjoyed it lived up to the hype deserves every award that it won all right well we've got some thank you to folks who helped us prepare for this episode so first to our sponsors JP Morgan Payments our presenting partner ServiceNow and Fundrise you can click the links in the show to learn more and then some special shoutouts to Aart de Geus the co-founder and executive chair of Synopsys had a great conversation with us well first publicly with Sassine Ghazi the current CEO of Synopsys on an Acquired episode a little while back and then we chatted to prep for this episode and basically asked the question what should we be David I know you've got a few also also to Wally Rhines the former CEO of Mentor Graphics Wally is a legend in the semiconductor industry almost on par with Dr. Chang they were contemporaries at TI back in the day and to John Bathgate and Britton Johns from NZS Capital our go-to folks on anything semiconductors they were I think they were more excited even more excited than we were that we were doing this and that we got to talk to them about it yes also past Acquired guest I think that episode holds up really well where we did semiconductor and complexity theory with them totally and actually John is the one originally who explained to me how EUV lasers work which is still one of the most impressive accomplishments in human the actual silicon wafers themselves that's a sophisticated process before the etching even starts he's just got some awesome awesome videos on the Asianometry YouTube channel and very very kindly bought David and I dinner and hung out with us the night before the interview which was very fun to do in Taipei very fun also to Tim Culpan a former Bloomberg journalist who now has a Substack called ChipM also gave us some great topics to chat about and lastly as always to Arvind Navaratnam at Worldly Partners he did a great great write up on TSMC that he'll be posting publicly right before we post this episode so you all can see it it was great last minute prep for me after reading the memoir to get someone else's take on study on that that we'll link to in the show notes so if you like this episode go check out other semiconductor episodes Nvidia we've got four of them at this point one of them is an interview with Jensen and then we've got the whole history of the company across three different episodes we did a great live episode several years ago on Qualcomm which I think is a sleeper pick that's right that's right total sleeper pick amazing story Irwin Jacobs one of the greatest entrepreneurs in American history yes and our diving into how CDMA works was one of the most fun technical explanations I've ever done on an Acquired episode so if you want to understand how all of our cell phones work go check out the Qualcomm episode or of course if you did not last week listen to the TSMC remastered episode from Arm Holdings that we did it's our most recent episode so it's spectacular and if you're interested in semis go check that out come talk about this episode with us in the Slack acquired.fm/slack and if you want to know when future episode drops you can find out sign up at acquired.fm/mail and you'll also get episode corrections and hints at what the next episode will be so with that listeners we'll see you next time.

We'll see you next time who got the truth is it you is it you is it you got the truth now huh.