"My reaction entering Harvard was sheer ecstasy, almost disbelief. What a country! The United States was at its peak in its moral leadership, its political leadership in terms of democracy, and it was the richest country in the world." — Morris Chang [00:08:07]
"Real men have fabs." — Jerry Sanders, co-founder & CEO of AMD(said in the mid-1980s, ironically just before AMD went fabless in the 2000s) [00:56:59]
Disclaimer: Orignal content owned by or sourced from third parties. It does not represent the views of 'Nuggets' platform or it's team. AI is used extensively across this platform including for summaries. Accuracy is not guaranteed, there can be mistakes. Any info or content on this platform is not a financial, legal, or investment advice. Do your own research. Refer for complete disclosures:- Terms of Use · Full Disclaimer
"The semiconductor business is like a treadmill that speeds up all the time. If you can't keep up, you fall off." — Morris Chang [01:21:17]
"We had no strength in research and development, or very little anyway. We had no strength in circuit design, IC product design. We had little strength in sales and marketing, and we had almost no strength in intellectual property. The only possible strength in Taiwan that we had, and even that was just a potential one, not an obvious one, was semiconductor manufacturing, wafer manufacturing. And so what kind of company would you create to fit that strength and avoid all the other weaknesses? The answer was a pure-play foundry." — Morris Chang [00:46:24]
"When I was at TI and General Instruments, I saw a lot of integrated circuit designers wanting to leave and set up their own business, but the one thing or the biggest thing that stopped them from leaving those companies was they couldn't raise enough money to form their own company." — Morris Chang [00:59:15]
"It was like in the movie The Godfather. It was an offer I couldn't refuse." — Morris Chang(on being directed by the Taiwanese government to start TSMC) [00:54:57]
"What we didn't realize then was that the integrated circuit would reduce the cost of electronic functions by a factor of a million to one, nothing had ever done that for anything before." — Jack Kilby, co-inventor of the Integrated Circuit [00:20:43]
Speakers & Credentials
Ben Gilbert — Co-host of Acquired; Managing Director of Pioneer Square Labs (PSL), a Seattle-based startup studio. Investor and operator with deep expertise in B2B software, hardware ecosystems, and platform businesses.
David Rosenthal — Co-host of Acquired; General Partner at Wave Capital, a crypto-native investment firm. Former VC at Social Capital. Expert in semiconductor history, platform dynamics, and long-cycle technology investing.
No external guests — this is a two-host deep-dive episode.
Executive Summary
TSMC (Taiwan Semiconductor Manufacturing Company) is the world's dominant contract chip manufacturer and, as of the remaster in 2025, a $1+ trillion market cap company — one of only two trillion-dollar companies globally not headquartered on the US West Coast (the other being Saudi Aramco).
Founded in 1987 in Taiwan by Morris Chang at age 56 with a $0 pre-money valuation, TSMC pioneered the "pure-play foundry" business model — manufacturing chips exclusively for third parties, never competing with its own customers — a concept that was widely ridiculed at inception ("Real men have fabs").
TSMC manufactures virtually every advanced chip in the modern world: every chip in every iPhone and MacBook, all of NVIDIA's AI/data center GPUs, chips for AMD, Qualcomm, Broadcom, Apple, AWS, Google, and Microsoft's custom silicon.
The company has achieved a 19.9% IRR over 27 years from its $4 billion market cap at its 1994 Taiwan IPO to $550 billion at time of original recording (2021), and has since doubled to $1 trillion+.
TSMC's moat is arguably the deepest of any company on earth, combining Scale Economies + Process Power so profound that no amount of government spending can replicate it in less than a decade — a point Ben and David argue constitutes a near-monopoly on human technological progress.
The company's singular risk is geopolitical: TSMC's primary operations sit on Taiwan, a strategically disputed island claimed by the People's Republic of China. A Chinese annexation of Taiwan would, in the hosts' assessment, effectively halt the world's technological advancement.
Morris Chang's story mirrors a three-act founder arc unlike any other: refugee child who survived three wars → 25-year TI executive passed over for CEO → founder of a trillion-dollar company at 56 → retired at 74 → returned as CEO at 78 to land the Apple deal → fully retired at 86.
Semiconductors are described as "the new oil": the substrate upon which all of modern civilization — defense, automotive, AI, consumer electronics, cloud — runs, making TSMC's position as the monopoly manufacturer of leading-edge chips the most strategically important corporate position in the world.
1. Chronological Table of Contents
[00:00:00] Remaster Context: TSMC's Growth from 2021 to 2025 [00:00:00]
00:02:32 Introduction to Morris Chang and TSMC’s Global Reach [00:02:32]
00:04:54 Early Life in China and Survival Through Three Wars [00:04:54]
00:08:01 Harvard, MIT, and the Limitations of Being a Chinese Engineer [00:08:01]
00:11:59 Anecdote: Morris, the Pipe, and Longevity Data [00:11:59]
00:13:35 The $1 Salary Gap: Choosing Sylvania over Ford Motor Co. [00:13:35]
00:15:52 Self-Education: Shockley’s Textbook and the Hotel Bar Teacher [00:15:52]
00:19:12 Texas Instruments (TI): The Origins of the Dallas Juggernaut [00:19:12]
00:23:59 Fixing the IBM Line: Morris’s First Major Manufacturing Breakthrough [00:23:59]
02:16:45 Grading: Where TSMC Sits in the Pantheon of Greatness [02:16:45]
02:24:30 Carve Outs: Ted Lasso, Greek, and Michael Ovitz [02:24:30]
2. Chronological History & Inflection Points
Act I — The Refugee Who Found His Calling (1931–1964)
Born July 1931 in Ningbo, China (a city of ~8M people today, though far smaller then). His father was a county official turned bank manager. The family was comfortably middle class. [00:04:54]
Before turning 18, Morris survived three wars: the Second Sino-Japanese War (family fled to Hong Kong), WWII (Japanese invaded Hong Kong December 8, 1941 — three hours after Pearl Harbor — forcing a return to Shanghai), and the Chinese Civil War (fled back to Hong Kong in 1948). [00:07:30]
1949: With help from an uncle in Boston, accepted to Harvard the same year the PRC was established. Described his arrival as "sheer ecstasy, almost disbelief." [00:08:01]
Transferred to MIT after freshman year to study Mechanical Engineering — the only path he saw respectable for a Chinese man in early 1950s America. Completed both BS and MS in three years (1951–1954). Failed his PhD qualifying exams twice. [00:10:08]
The $1 Decision: Offered $479/month by Ford Motor Company; offered $480/month by Sylvania's new Semiconductor Division. Asked Ford to match — they didn't. Took the Sylvania job and began his semiconductor career in Ipswich, Massachusetts. [00:13:35]
At Sylvania, taught himself electrical engineering from Shockley's 1950 textbook, studying in his hotel room and buying drinks for a senior alcoholic engineer who answered his technical questions every night at the bar. [00:15:52]
1958: Joined Texas Instruments (TI) in Dallas — the same year Jack Kilby invented the integrated circuit at TI. Morris was assigned to fix a failing transistor manufacturing line for IBM's 7090 mainframe. IBM's line was at 10% yield; TI's was at ~0% yield. Within four months, Morris brought TI's yield to 20% — double IBM's — using first-principles mechanical engineering intuition. This made him famous internally overnight. [00:23:59]
TI sponsored Morris to get his PhD from Stanford, paying his full salary, which he completed in 2.5 years. By 1964, he was back at TI as silicon replaced germanium industry-wide. [00:27:08]
Act II — The Rise and Fall at Texas Instruments (1964–1983)
1967: Promoted to General Manager of a semiconductor division. Made his signature intellectual contribution to the industry: learning curve pricing — deliberately pricing chips below cost at launch to maximize volume, accelerate yield improvements via learning, and capture market share, then systematically lowering prices every quarter even when the market didn't require it. [00:30:15]
Hired BCG (then a tiny firm) to validate and develop the framework. Result: TI's IC business became the largest and most profitable in the world — surpassing Fairchild, Intel, and National Semiconductor simultaneously. [00:32:34]
1972: Promoted to VP of the Semiconductor Division at TI — one level below the CEO — and the presumed heir apparent to lead the company. [00:36:14]
Three strikes ended his TI career: (1) Possible ethnic discrimination in passing him over for CEO; (2) Moved in 1978 to run the Consumer Products Division (calculators, etc.), a fundamentally different business requiring marketing intuition he didn't have; (3) TI missed the transition to MOS semiconductors and critically lost the IBM PC processor contract to Intel in 1980, cementing the x86 era. [00:36:51]
1983: Demoted to "head of quality and people effectiveness." Resigned. His 25-year run at TI was over. [00:38:39]
Briefly became COO of General Instrument (GI) in Manhattan — a proto-semiconductor private equity firm that acquired and flipped chip businesses. Left within a year. His stated dream — to be CEO of a major US company — was officially over. [00:43:04]
Act III — The Godfather's Offer: Founding TSMC (1985–1987)
1985: Recruited by KT Li, Taiwan's Minister of Economic Affairs (later known as the "Father of Taiwan's economic miracle"), to run ITRI (Industrial Technology Research Institute) — essentially Taiwan's attempt to build a Bell Labs of its own. [00:48:44]
Morris viewed it as semi-retirement. Taiwan's manufacturing economy at the time had average gross margins of 4%–5% — commodity physical goods (toys, Barbie dolls, clothing). ITRI had previously executed a tech transfer from RCA (already 2.5 generations behind the US leading edge at the time of transfer) that had produced UMC (United Microelectronics Corporation). [00:49:10]
Three days after being told to develop a business plan, Morris conceived the pure-play foundry model: a semiconductor company that manufactures chips exclusively for fabless third parties, competing with no one, enabling everyone. Quote: "It was an offer I couldn't refuse." [00:54:57]
Founding cap structure (1987): $220 million total capital. Government of Taiwan: 50%. Philips (the Dutch electronics giant, which had co-founded ASML three years earlier in 1984): 28%. Other Taiwanese business leaders (strong-armed by the Premier): 22%. Morris Chang's equity: $0. He received no founder's equity — he was a government employee. He would later accumulate his ~$3 billion net worth by buying TSMC shares on the open market after the IPO. [01:01:02]
1987 was the Annus Mirabilis of semiconductors: TSMC founded. ARM founded (as a JV between Apple, Acorn, and VLSI Logic). The entire ecosystem of fabless semiconductors was born in a single year — unknown to the participants. [01:03:36]
Act IV — Building the Flywheel (1987–2005)
Early years were brutal: Morris "begged" IDMs (Integrated Device Manufacturers — Intel, TI, Motorola) to give TSMC their overflow capacity and their money-losing product lines. IDMs would give work during shortages, yank it back when their own capacity came online, and offload chips that were losing money — transferring losses to TSMC. [01:14:03]
The core strategic bet: fabless chip companies will emerge once they realize they don't need to fund their own fabs. Morris had seen dozens of chip designers at TI and GI who wanted to leave but couldn't raise the capital. TSMC removed that constraint. [00:59:15]
Morris positioned TSMC as the "Y Combinator of semiconductors": by becoming the manufacturing platform, he effectively took an index bet on all future semiconductor entrepreneurship rather than picking individual winners. [01:15:35]
Early fabless customers: Qualcomm, Broadcom, Marvell, NVIDIA (founded 1993, only raised $20 million, never opened a fab, grew to a $350–400B company). Jensen Huang has credited TSMC for making NVIDIA possible. [01:17:16]
The Flywheel: Fabless companies outsource to TSMC → TSMC revenue grows → 50% gross margins, 40% operating margins generate enormous cash → cash reinvested in CapEx to advance process technology → better processes enable better chips → fabless customers grow and multiply → repeat. From 22 companies at leading-edge process in the early 2000s (when TSMC first joined the pack at ~150nm), the field narrowed to 14 → 6 → 2 (TSMC + Samsung) by the late 2010s, converging to essentially 1 (TSMC) at 3nm. [01:18:02]
Morris's second great innovation: applying his own learning curve pricing to TSMC — he continued to systematically lower prices to customers to drive volume. This was completely counter-cultural for the industry. [00:32:54]
1994: IPO on Taiwan Stock Exchange at a $4 billion market cap. 1997: Listed on NYSE. [01:45:06]
2005: Morris retires at age 74. Hands TSMC to longtime lieutenant Rick Tsai. [01:30:58]
Act V — The Comeback & the Apple Deal (2008–2018)
Summer 2009: Surprise press conference. Morris returns as CEO at age 78, citing "golden opportunities ahead" — the mobile smartphone wave and the emerging cloud computing era. Rick Tsai departs; later becomes CEO of MediaTek. [01:31:50]
At this moment, Apple had the iPhone but was still using off-the-shelf Samsung processors. Morris understood that all the leading mobile chip companies (Qualcomm, Broadcom, MediaTek, and eventually Apple's custom silicon) were ARM-based and fabless — his customers. [01:33:40]
The Apple Deal (~2010–2012): Morris personally closes the deal with Apple (Apple side represented by Jeff Williams, "Tim Cook's Tim Cook"). TSMC committed an initial $9 billion investment in a dedicated fab and 6,000 employees to build it in 11 months — with no backup plan. Apple committed to TSMC exclusively as Samsung was simultaneously a chip fab competitor and a smartphone competitor. [01:40:05]
The A4 chip (the first Apple-designed chip, built on the P.A. Semi acquisition) was fabbed by TSMC. By the iPhone 6 (2014), all Apple chips were solely manufactured by TSMC — a mega-hit product that cemented the relationship. [01:38:37]
Lineage of Apple Silicon: ARM (1987, JV with Apple) → DEC (StrongARM) → Intel acquisition (XScale) → engineers defect → P.A. Semi (founded) → Apple acquires P.A. Semi for ~$278M (2008) → all Apple chips in perpetuity. [01:13:21]
Intel's Strategic Error: Steve Jobs approached Intel CEO Paul Otellini about manufacturing iPhone chips. Intel considered the price offer too low and declined. Apple went to Samsung, then TSMC. Intel's x86 obsession and pricing sensitivity cost them the entire mobile chip era. [02:06:26]
2013: Morris steps down as CEO again but remains Chairman. [01:41:17]
June 2018: Morris fully retires from Chairman role at age 86. [01:41:24]
Act VI — Monopoly of the Leading Edge (2019–Present)
2020 Financials: $48 billion revenue, $20 billion operating profit, $17 billion adjusted net income. Reinvested $17 billion back into CapEx that same year. [01:44:49]
Revenue growth 2019→2020: 31% at an already massive scale. [01:46:15]
January 2021: CapEx guidance raised from $17B (2020) to $25–28B. By April 2021: raised further to $30B for 2021 and $100B over the next three years. [01:42:04]
3nm process: TSMC launching imminently; Samsung slipped to 2024; Intel has no viable 3nm roadmap at time of recording. [01:20:03]
Pricing power unlocked: First time since Morris's original learning curve strategy, TSMC announced a 20% price increase — the clearest signal of irreplaceable monopoly power. [01:46:58]
Balance sheet: $28 billion in cash and equivalents. [01:47:28]
Market share: 50%+ of all contract chip manufacturing; 90%+ of the profit pool in the foundry industry. [01:47:47]
Geopolitical diversification: $12 billion fab announced in Arizona (not leading-edge; partially US-subsidized under Biden semiconductor legislation); additional fab announced in Japan; existing operations in China. Morris publicly skeptical of the business rationale, calling it subsidized diversification rather than strategic necessity. [02:01:33]
2025 (remaster context): Market cap has doubled from $550B to $1 trillion+, making it one of only two trillion-dollar companies outside the US West Coast. [00:01:35]
3. The Acquired Playbook & Themes
Power Context (Hamilton Helmer's 7 Powers)
Counter-Positioning(strong, historical): When TSMC launched, no IDM would dare adopt the pure-play foundry model — doing so would mean abandoning their vertically integrated profit centers and explicitly ceding the design-to-manufacture premium. IDMs also previously strong-armed fabless companies into allowing co-branding of products as a condition of excess capacity — something TSMC explicitly rejected. [01:54:08] The incumbent response was dismissal ("Real men have fabs"), [00:57:02] not imitation, which is the textbook definition of Counter-Positioning.
Scale Economies(the dominant, structural power): Rock's Law (Moore's Second Law) states fab costs double every four years. [01:50:53] This means the industry is a winner-take-all capital arms race: whoever can invest the most in the next-generation fab captures the leading edge, captures the highest-margin customers, generates the most profit, and reinvests the most in the next next-generation fab. TSMC has out-invested everyone for 40 years. From 22 leading-edge competitors → 2 → effectively 1. [01:19:25] CapEx trajectory: $17B (2020) → $30B (2021) → $100B over three years — [01:42:04] no competitor can match this pace.
Switching Costs(massive and growing): Customers cannot simply "port" their chip design to a competing foundry. The manufacturing process is so deeply integrated — yield recipes, design rules, EDA software certification, ASML equipment calibration — that switching takes years even if a competitor foundry exists. [01:56:00] At the 5nm node, there is no competitor to switch to. Switching costs are now existential constraints.
Network Economies(limited but present via ecosystem): Not network effects in the Facebook sense (no direct customer-to-customer value). However, TSMC has built what it calls an "Open Innovation Platform" — deep integrations with EDA vendors (Synopsys, Cadence), IP providers (ARM), and equipment manufacturers (ASML, Lam Research, Applied Materials). [01:05:31] Designing for TSMC's process using ARM IP and Synopsys EDA tools is a deeply optimized stack. Switching foundries means leaving this entire ecosystem.
Process Power(the clearest and most defensible): Ben and David argue this is the purest example of Process Power they have ever encountered on Acquired. [01:57:15] TSMC's manufacturing capability is the accumulated product of 40 years of continuous iteration — a living institutional knowledge base of yield optimization, equipment operation, process recipes, and engineering judgment that cannot be purchased or replicated regardless of capital available. The EUV laser/molten tin plasma process (50,000 pulses/second, more precise than Apollo mission calculations) [01:26:38] is operated by a small number of humans on earth who learned from TSMC. Even 10 senior TSMC engineers defecting to start a competitor could not replicate what the institution has accumulated. The Toyota Production System comparison: even Toyota couldn't fully replicate its process power at NUMMI/GM — and that transfer wasn't under threat of war. [02:04:04]
Branding(deliberately absent — anti-brand): TSMC explicitly has no consumer brand. The brand is the customer's. "The brand is the customers." [01:48:24] This is a feature, not a bug — it ensures customers have no competitive anxiety about engaging TSMC as their manufacturer.
Cornered Resource(partial, structural): TSMC has historically had a privileged position due to Taiwan's government co-investment and political will, [01:00:43] and has a preferential allocation of ASML's EUV machines (TSMC reportedly has ASML's entire production backlogged for years). [01:29:15] However, the Taiwan location is also an anti-power / bear case risk factor (see below).
Value Creation vs. Value Capture
Value Created: Immeasurable. Marc Andreessen's thesis — "Software is eating the world" — is only true because TSMC has made it economically viable to manufacture billions of transistors for near-zero marginal cost. [02:17:58] Every smartphone, AI model, autonomous vehicle, cloud server, and IoT device runs on TSMC-manufactured silicon. Ben's framing: "The cost of compute asymptotically approaches zero because TSMC, TSMC, TSMC." [02:18:04] The integrated circuit reduced the cost of electronic functions by a factor of a million to one (Kilby). [00:20:50] TSMC enabled that trajectory to continue.
Value Captured: Extraordinary relative to a contract manufacturer, but modest relative to the value created. 50% gross margins, 40% operating margins, $20B operating profit on $48B revenue (2020). [01:41:49] 17.4% compound annual revenue growth for 27 years. 19.9% IRR from $4B to $550B market cap. [01:45:06] Yet the companies built on TSMC's foundation — Apple ($2T+), NVIDIA ($300–400B at time of recording), Qualcomm, Broadcom — have captured enormous incremental value on top of TSMC's manufacturing foundation.
The Paradox: TSMC's lack of consumer brand, deliberate customer deference, and "customer-first" ethos have likely constrained value capture — but maximized value creation and competitive insulation. [02:22:14] Their 40% operating margins in a "commodity" manufacturing business are the proof that abstaining from branding and customer competition is the right long-term capture strategy.
The Bull and Bear Case
Bull Case:
Compounding structural monopoly: At 3nm, TSMC is on track to be the only manufacturer in the world. [01:20:09] Every successive process node further entrenches the lead because capital requirements, process knowledge, and ecosystem integration make catching up logarithmically harder. [01:22:55]
AI tailwind is generational: Every AI chip — NVIDIA H100/H200, Google TPU, Amazon Trainium, Meta MTIA, Microsoft Maia — is fabbed by TSMC. [01:36:56] The AI wave makes TSMC the picks-and-shovels play on the entire AI economy.
Price increases signal irreversibility: For the first time since Morris's learning curve strategy 50+ years ago, TSMC is raising prices by 20%. [01:46:58] This is the clearest possible signal that customers have no alternative. Pricing power in a previously commodity-priced business is the hallmark of a true monopoly.
The flywheel cannot be reversed: 40%+ operating margins → $100B CapEx over 3 years → next-gen process → more customers → higher margins. [01:18:02] The compounding advantage grows with every cycle.
Fabless model is the future of chip design: Every hyperscaler (Apple, Google, Amazon, Microsoft, Meta) is now designing custom silicon. [01:37:03] All of them are fabless. All roads lead to TSMC.
Bear Case:
Geopolitical catastrophe: China annexes Taiwan. [02:03:04] This is the existential scenario — not a business disruption but a civilizational one. The hosts argue this would effectively halt global technological progress. [02:03:22] The Arizona and Japan fabs are partial hedges but not leading-edge capable. [02:01:43] The question of whether TSMC's process power could be "airlifted" — all people, all ASML machines, all IP relocated to a new geography under duress — is genuinely unanswerable. [02:03:43]
Compute paradigm shift: If quantum computing, photonic computing, or a new substrate (non-silicon) makes EUV lithography obsolete, the playing field resets. [01:57:59] TSMC's entire process power moat is built on EUV/silicon. A paradigm-level change is the only technological path to disruption.
The "good enough" argument: As transistors approach physical limits (~5 atoms wide at 3nm), [01:50:00] future performance gains will come from architecture and packaging, not process shrinks. If "good enough" silicon at 7nm or 5nm handles the vast majority of workloads, the leading-edge premium erodes. Ben and David largely dismiss this — software always expands to consume available hardware. [01:48:52]
Intel's foundry ambitions: Pat Gelsinger's Intel is attempting to rebuild a leading-edge foundry business (Intel Foundry Services). [02:11:12] The hosts are deeply skeptical — Intel invested $4 billion in EUV in 2012, was the earliest major investor, [02:13:17] yet as of recording has not shipped a single chip manufactured using EUV. The Innovator's Dilemma in action. [02:14:04]
ASML concentration: The entire leading-edge semiconductor supply chain depends on one Dutch company making ~50 EUV machines per year. [01:27:56] Any disruption to ASML — trade restrictions, geopolitical events, technical failures — disrupts TSMC. [01:24:06] Though TSMC has reportedly pre-ordered ASML's entire production queue for years, this is a single-point-of-failure in the global supply chain. [01:29:15]
The Reference Vault
4. Data & Figures
Data Point
Value
Context
Timestamp
TSMC Market Cap (2021 recording)
$550 billion
9th most valuable company in the world at time of recording
1. The Pure-Play Foundry Model (Morris Chang's Strategic Insight)Application: Rather than try to compete across the entire semiconductor value chain (design + manufacture + sell), TSMC became the world's first dedicated contract chip manufacturer — serving only companies that design chips, competing with none of its customers. This was a deliberate counter-positioning strategy that the incumbents (Intel, TI, Motorola) could not replicate without destroying their own integrated profit centers. The model unlocked the entire fabless semiconductor industry and, by extension, the smartphone, cloud, and AI eras. [00:46:24]
2. Learning Curve Pricing (The BCG/Morris Framework)Application: Morris's signature insight from his TI days: price below cost at launch to maximize volume, accelerate yield improvements via production learning, and capture market share. Then systematically reduce prices every quarter even absent market pressure. At TI this made the IC division the world's most profitable. At TSMC, he applied the inverse: priced for volume to seed the fabless ecosystem, subsidizing early customer growth. Only in 2021 — 34 years later — did TSMC announce its first price increase. The reversal signals the transition from ecosystem-building to monopoly rent extraction. [00:32:54]
3. Moore's Law + Rock's Law = Winner-Take-AllApplication: Moore's Law (transistor count doubles every ~2 years, reducing cost of compute) describes the operating side of the chip equation. Rock's Law (Moore's Second Law) states that fab construction costs double every 4 years — describing the capital side. Together, they imply a brutal structural dynamic: to stay at the leading edge you must invest exponentially more CapEx every cycle, but if you stay at the leading edge your operating costs per chip fall exponentially, giving you the cash flow to fund the next CapEx cycle. This is an inexorable winner-take-all compounding engine. TSMC's CapEx trajectory ($17B → $30B → $100B over 3 years) is the machine executing this logic. [01:50:53]
4. The Y Combinator / Platform BetApplication: Morris (who had considered becoming a VC before going to Taiwan) realized that by becoming the manufacturing platform for chip designers, he was effectively taking an index position on all future semiconductor entrepreneurship. Instead of picking winners, he enabled all winners. NVIDIA, Qualcomm, Broadcom, Marvell, Apple — TSMC didn't need to pick them. They all came to TSMC. This is directly analogous to AWS (vs. picking internet startups), Stripe (vs. picking payments startups), or WeChat (platform enabling the entire Chinese mobile economy). Ben calls it "the Y Combinator of semiconductors." [01:15:35]
5. The Rosenthal Postulate: Never Make Strategic Decisions Based on PricingApplication: Illustrated by two catastrophic examples in this episode: (1) Ford Motor Company refused to pay Morris Chang $1 more than Sylvania, losing the employee who would go on to build the 9th most valuable company in history. (2) Intel's Paul Otellini turned down Steve Jobs' offer to manufacture iPhone chips because the price was "too low" — handing the entire mobile chip era to Samsung and then TSMC. The meta-lesson: when a strategic relationship appears economically suboptimal but structurally transformative, the strategic consideration must dominate. [02:08:49]
6. The Innovator's Dilemma — Intel and EUVApplication: Intel invested $4 billion in EUV technology in 2012 — the earliest major investor in the only technology that could enable sub-10nm chip manufacturing. Yet Intel's legacy business (x86, IDM model) made it impossible to commit fully to EUV, as doing so would cannibalize existing process investments. As of recording, Intel had not shipped a single EUV-manufactured chip. Meanwhile TSMC and Samsung raced ahead. Ben: "This is the most perfect, pure example of the innovator's dilemma in action." David: "That's why you need startups." [02:13:17]
7. Process Power as the Ultimate MoatApplication: Hamilton Helmer defines Process Power as embedded institutional knowledge in processes that produces superior outcomes and cannot be replicated — regardless of capital, talent, or intent. Ben and David argue TSMC is the clearest example of Process Power they have ever encountered. 40 years of yield optimization, EUV operation know-how, ASML integration, supplier relationships, and engineering judgment are embodied in the organization. Toyota's attempt to export the Toyota Production System to NUMMI (now Tesla's Fremont plant) is the comparative: even with full cooperation and resources, replicating Process Power is extremely difficult. At TSMC, the entire global supply chain of advanced chip manufacturing runs through this single institution. [01:57:15]
8. Horizontal Specialization as Industry ArchitectureApplication: The semiconductor industry evolved from vertical integration (one company designs, manufactures, and sells chips — TI/Intel/IDM model) to a deeply horizontal architecture with five distinct layers: (1) IP/Architecture (ARM, MIPS), (2) EDA Software (Synopsys, Cadence), (3) Fabless Design (Apple, NVIDIA, Qualcomm, AMD), (4) Equipment Manufacturing (ASML, Lam Research, Applied Materials, Tokyo Electron), (5) Foundry/Manufacturing (TSMC, Samsung). Moore's Law's pace is so fast that no single company can maintain excellence across all five layers simultaneously. Ben: "You can't be good at everything." The horizontal model maximizes specialization and best-of-class at every layer — and TSMC wins Layer 5 absolutely. [01:05:46]
6. Anecdotes & Lore
1. The $1 That Changed History
Ford offered Morris Chang $479/month. Sylvania offered $480/month. Morris asked Ford to match Sylvania. Ford refused. Morris took the Sylvania job. Morris went on to found TSMC — the 9th most valuable company in the world. David's framing: the same pattern repeated with Intel turning down Apple's chip contract over price. "Humans are so prone to cutting off their noses to spite their faces." [00:13:35]
2. The Alcoholic Teacher
During his early years at Sylvania's Ipswich, Massachusetts plant, Morris had no formal electrical engineering training. He discovered that a senior engineer at the plant was an alcoholic who frequented the hotel bar nightly. Morris would go down to the bar — not to drink — but to buy the engineer drinks and pepper him with questions from Shockley's 1950 textbook. "He didn't solve all my problems, but he solved enough so that I could move ahead. He was my main teacher about electrical engineering." This went on for three years. Morris was teaching himself a field at night while working in it by day. [00:17:01]
3. The $0 Pre-Money Valuation
Ben and David save the most jaw-dropping detail for a key reveal: TSMC was founded at a $0 pre-money valuation. Morris Chang received no equity. The government owned 50%, Philips owned 28%, and other Taiwanese investors owned 22%. He was a government employee who was ordered to start a company. His entire ~$3 billion net worth was accumulated by purchasing TSMC shares on the open market after the company's IPO. David: "This is so the opposite of Silicon Valley." [01:02:22]
4. The Godfather Moment
Morris was at ITRI (Taiwan's Bell Labs equivalent) thinking of it as a cushy semi-retirement when Minister KT Li came to him and essentially said: go start a semiconductor company and make it a world leader. Morris had been through TI, GI, and was by his own admission done dreaming of being a US CEO. As he put it: "It was like in the movie The Godfather. It was an offer I couldn't refuse." With three days to develop a business plan (not a week — Li moved up the deadline by a day), Morris invented the pure-play foundry model. [00:54:57]
5. Morris and the Pipe
David and Ben note that Morris Chang is famously funny and dry-witted. In a Stanford audience Q&A, an attendee asked how Morris finally quit smoking. His answer: he never quit. He still smokes a pipe at age ~94. He has "delved into the data" and concluded that pipe smokers live longer than non-smokers — and is fairly certain his pipe has prolonged his life. Ben: "Which I'm sure you can find data to support, and also sure you can find plenty of data to refute." [00:11:59]
6. The Apple Dinner
The deal that transformed TSMC from a great company into the only manufacturer of the world's most important chips — Apple's exclusive foundry relationship — was reportedly inked in a living room dinner. Jeff Williams (Apple's COO, "Tim Cook's Tim Cook") went to Morris's home. Apple was betting its entire custom silicon future on a company the market then viewed as an also-ran. TSMC committed $9 billion and 6,000 employees to build a dedicated Apple fab in 11 months. Bloomberg's quote: "If we were to bet heavily on TSMC, there would be no backup plan." A bet-the-farm moment for both companies that paid off beyond any forecast. [01:38:26]
7. The ASML Tin-Plasma Machine
Ben's description of what ASML's EUV machines actually do is the episode's most mind-bending technical passage. To etch features at 5nm or below: on one side of the machine, molten tin drops fall. On the other side, a specialized laser (made by the only company in the world capable of building it, based in Germany) fires and hits the tin droplet. The impact creates a plasma that emits extreme ultraviolet light (wavelength below 11nm). This must happen 50,000 times per second. The accuracy required is more precise than the calculations to send the Apollo missions to the moon. The EUV light then reflects off specially invented mirrors (existing mirrors absorb light at this wavelength) approximately 20 times before etching the silicon. Cost per machine: $200M (rising to $300M). Shipping requires 4 Boeing 747s. Annual production: ~50 machines worldwide. All allocated to TSMC for years in advance. [01:26:38]
8. The Three Retirements of Morris Chang
Morris Chang's career arc has no parallel in business history: he spent 25 years at TI, was passed over for CEO, demoted, and resigned at 52 thinking his career was over. He went to General Instrument for a year, confirmed it wasn't right, resigned again. Went to Taiwan at 54 thinking it was a soft landing. Founded TSMC at 56 with no equity. Retired at 74. Was called back as CEO at 78 during the 2008 financial crisis to capitalize on the smartphone wave and close the Apple deal. Stepped down as CEO in 2013, remained Chairman. Fully retired at 86 in June 2018. David compares him to Michael Jordan — except "he's beyond Jordan." [01:31:04]
7. References & Recommendations
Books & Publications
"Electrons and Holes in Semiconductors with Applications to Transistor Electronics" by William Shockley (1950) — The textbook Morris self-studied to teach himself electrical engineering while working at Sylvania; foundational text of the transistor era. [00:16:05]
IEEE Spectrum — Industry publication; primary sourced material for Morris's IBM mainframe yield breakthrough at TI. [00:26:32]
Bloomberg — Source for the Intel/Apple iPhone chip story (Jobs approached Otellini; Intel declined due to price) and the Apple-TSMC initial deal details ($9B investment, 11-month fab build, "no backup plan"). [02:06:26]
Wall Street Journal — Source for Intel's $4B+ EUV investment in 2012 and subsequent failure to commercialize it. [02:13:17]
Bloomberg Odd Lots Podcast — Discussed bear cases for TSMC and China's attempt to build domestic fabs (SMIC); featured a TSMC-focused reporter whose analysis the hosts reference extensively. [01:23:07]
The Knowledge Project Podcast (Shane Parrish) — Featured John Bathgate and Britton Johns discussing ASML and EUV; referenced by Ben as a key source for the EUV section. [01:24:55]
"7 Powers: The Foundations of Business Strategy" by Hamilton Helmer — The core analytical framework applied throughout the episode. Ben calls it "the best business theory book." David and Ben have featured Hamilton as a podcast guest. [01:53:07]
"Who Is Michael Ovitz?" — Ben's carve-out recommendation; a business memoir about Creative Artists Agency (CAA) from 1975–2000. Recommended for fans of the Andreessen-Horowitz Acquired series, the Disney episodes, and business strategy storytelling. [02:24:30]
Competing Companies & Key Industry Players
Intel — The great foil of the episode. TI's biggest competitor at the IC level; won the IBM PC contract (1980) that cemented x86. Later turned down Apple's iPhone chip contract. Invested $4B in EUV but failed to commercialize it. Now attempting to re-enter the foundry business under CEO Pat Gelsinger. [02:11:12]
Texas Instruments (TI) — Morris's 25-year employer; the dominant semiconductor company of the 1960s–70s (beating Fairchild, Intel, National). Lost the IBM PC CPU contract. The "secret" history of the semiconductor industry. [00:19:12]
Samsung — TSMC's only remaining leading-edge competitor at 5nm (5–10% share). Also a consumer electronics company (Galaxy phones), which creates strategic conflict with foundry customers. Apple initially used Samsung chips before switching to TSMC. [01:48:03]
ASML — Dutch company; the world's sole manufacturer of EUV lithography machines. Founded in 1984 as a Philips/ASM joint venture. Now Europe's most valuable company by market cap (larger than Intel). Makes ~50 EUV machines per year; TSMC is its largest customer. [01:21:18]
SMIC — China's Semiconductor Manufacturing International Corporation; the Chinese government's attempt to build a domestic leading-edge foundry. Blocked from accessing ASML EUV equipment. [01:23:22]
AMD — Jerry Sanders co-founded AMD and famously said "real men have fabs." AMD later spun out its fabs into GlobalFoundries in the 2000s and became entirely fabless — a validation of Morris's original thesis. [00:57:02]
GlobalFoundries — AMD's spun-out fab arm; attempting to IPO at time of recording. Not a leading-edge competitor to TSMC; serves legacy process nodes. [00:57:11]
NVIDIA — Founded 1993; only raised $20 million total; never built a fab; built entirely on TSMC manufacturing. Jensen Huang credits TSMC for making NVIDIA possible. [01:17:16]
Qualcomm, Broadcom, Marvell — Early fabless chip companies seeded by TSMC's platform; all ARM-based mobile chip leaders. [01:17:16]
ARM — Founded 1987 (same year as TSMC) as a JV between Apple, Acorn, and VLSI Logic. All ARM chips are manufactured by TSMC. [01:03:44]
UMC (United Microelectronics Corporation) — Taiwan's pre-TSMC foundry attempt, using RCA-licensed technology (2.5 generations behind leading edge). Spun out its chip design business as MediaTek. [00:53:16]
MediaTek — Spun out of UMC; one of the world's largest fabless chip companies. Rick Tsai (TSMC's interim CEO) became MediaTek CEO. [01:32:19]
Synopsys, Cadence — The two dominant EDA (Electronic Design Automation) software companies. "Figma or Excel for chip designers." [01:05:46]
Lam Research, Applied Materials, Tokyo Electron — Leading semiconductor equipment manufacturers. Produce the non-EUV equipment in the fab ecosystem. [01:08:02]
Motorola — One of the few IDMs that agreed to give TSMC early foundry business; also poached MOS engineers from TI in the 1970s. [00:40:26]
General Instrument (GI) — Morris's one-year stopover after TI; described as a "proto tech private equity firm." [00:43:04]
Sylvania — Morris's first job ($480/month); Don Valentine also worked at Sylvania's semiconductor division contemporaneously. [00:13:35]
Key Operators & Historical Figures
Morris Chang — Founder, TSMC. Order of Propitious Clouds with Special Grand Cordon. TI (25 years) → GI (1 year) → ITRI (Taiwan) → TSMC founder at 56. [00:03:00]
KT Li — Taiwan's Minister of Economic Affairs; recruited Morris to ITRI and then directed him to start TSMC. [00:48:44]
Rick Tsai — Morris's longtime lieutenant at TSMC; replaced by Morris in 2009; later became CEO of MediaTek. [01:31:10]
Jeff Williams — Apple's COO; negotiated the Apple-TSMC chip manufacturing deal with Morris Chang. [01:38:26]
Jack Kilby — Co-invented the integrated circuit at Texas Instruments in 1958. [00:20:24]
Bob Noyce — Co-invented the integrated circuit at Fairchild; later co-founded Intel. [00:20:15]
Gordon Moore — Co-founder of Intel; coined Moore's Law while at Fairchild. [00:20:36]
Jerry Sanders — Co-founder and CEO of AMD; coined "real men have fabs" in the mid-1980s. [00:57:02]
Paul Otellini — Intel CEO who turned down Jobs' offer to manufacture iPhone chips. [02:06:26]
Don Valentine — Sequoia Capital founder; also started at Sylvania's semiconductor division. [00:14:27]
Pat Gelsinger — VMware CEO → Intel CEO; attempting to rebuild Intel's foundry capability. [02:11:12]
Arthur Rock — OG Silicon Valley VC; "Rock's Law" (Moore's Second Law) named after him. [01:50:53]
LJ Sevin — TI engineer who left to found Mostek and later Sevin Rosen Ventures. [00:39:52]
Jensen Huang — NVIDIA co-founder and CEO; credited TSMC as foundational to NVIDIA's existence. [01:17:40]
Geopolitical & Macro Events
Second Sino-Japanese War (1937–1945) — Morris fled Ningbo to Hong Kong. [00:06:31]
World War II (Pacific Theater, 1941–1945) — Japan invaded Hong Kong December 8, 1941; Morris fled back to mainland China. [00:06:45]
Chinese Civil War & Communist Revolution (1945–1949) — Morris fled to Hong Kong; uncle facilitated Harvard acceptance. [00:07:23]
The Global Chip Shortage (2021) — Proximate trigger for the episode; Ford paused F-150 production. [00:04:01]
Taiwan's Geopolitical Status — The "disputed island" where virtually all leading-edge chip manufacturing occurs. Annexation described as existential risk. [02:03:04]
US CHIPS Act / Biden Semiconductor Legislation — Subsidizing TSMC's $12B Arizona fab re-shoring efforts. [02:01:43]
Netherlands-China Trade License Non-Renewal — Refused to renew ASML's trade agreement with China, blocking EUV machine sales to SMIC. [01:24:06]
Taiwan Martial Law — Context for why Morris's "Godfather offer" was not truly optional. [00:51:13]
8. The Bottomline (by AI)
TSMC's story is the ultimate proof that the most important company in any technological era is rarely the most visible one. Morris Chang — a refugee, a failed PhD candidate, a corporate castaway — built the world's most critical monopoly not by being the smartest chip designer, but by creating the platform that made all other chip designers possible. The lasting lesson is structural: identify the one thing every future innovator in a field will need but cannot afford to build themselves, and become the world's best provider of exactly that thing at a price that seeds the ecosystem rather than extracting from it. For founders and investors, the corollary is equally sharp: any time a massive industry is filled with companies doing something that isn't their core product — funding their own fabs, running their own servers, managing their own payroll infrastructure — the platform opportunity for someone to do only that thing, but do it for everyone and do it better than any of them ever could alone, is potentially worth a trillion dollars.
"Brookfield's the largest infrastructure owner in the world... We drew a pipeline and we showed all the different components of the payments ecosystem on a pipeline and said it's like a pipe that moves any commodity except what it's moving…
$220 million total
Gov of Taiwan: 50% ($110M); Philips: 28%; Taiwanese business leaders: 22%