Posts
Chapter 1: The Godfather
Morris Chang and the Foundry Revolution That Made One Company Indispensable—And Put the World at Risk
```The Most Important Company You've Never Thought About
Morris Chang built the most important company in the world—and he did it in the most dangerous place possible.
Taiwan Semiconductor Manufacturing Company (TSMC) produces over 90% of the world's most advanced computer chips. If you own an iPhone, use AI, drive a modern car, or depend on cloud computing, you depend on TSMC. Apple, NVIDIA, AMD, Qualcomm—the titans of tech are all customers of this one Taiwanese company.
TSMC's market capitalization exceeds $800 billion. Its annual revenue approaches $70 billion. It employs over 73,000 people across the most sophisticated manufacturing facilities ever built.
But none of those numbers capture what TSMC actually is:
- The single point of failure for global technology
- The most vulnerable chokepoint in the world economy
- The company that could trigger World War III
Because TSMC's crown jewel facilities sit 100 miles from mainland China—a country that claims Taiwan as its territory and has explicitly refused to rule out military invasion to reclaim it.
This is the story of how we got here. How one man's vision created a technological monopoly. How efficiency trumped security. How the pursuit of the perfect chip made the entire world dependent on a vulnerable island.
And why, despite knowing all this, we can't escape. We're trapped by our own success.
Part I: The Man Who Changed Everything
Born Between Empires
Morris Chang's story begins in chaos—mainland China in 1931, as the Japanese Empire expanded and civil war loomed.
Born in Ningbo, Zhejiang Province, Chang's childhood was defined by displacement. His family fled the Japanese invasion, moving repeatedly as conflict consumed China. By age 18, he had lived through the Japanese occupation, World War II, and the Chinese Civil War.
In 1949, as the Communist revolution swept China, Chang made the choice that would shape his life: he left for America.
The Education of Morris Chang:
- 1952: Bachelor's degree in mechanical engineering, MIT
- 1953: Master's degree in mechanical engineering, MIT
- 1964: PhD in electrical engineering, Stanford
Chang didn't just get educated—he positioned himself at the cutting edge of the emerging semiconductor revolution. By the time he completed his PhD, integrated circuits were transforming from laboratory curiosity to industrial reality.
The Texas Instruments Years (1958-1983)
Chang joined Texas Instruments in 1958, just as the semiconductor industry was being born. Over 25 years, he rose to become the company's Group Vice President for semiconductors—the third-highest position at TI.
These weren't just corporate ladder-climbing years. Chang learned how semiconductors were made, how the industry worked, and critically—what was broken about the existing model.
What Chang Observed at Texas Instruments:
- Companies designed their own chips AND manufactured them (vertical integration)
- This required enormous capital investment in fabrication facilities ("fabs")
- Most companies weren't good at both design and manufacturing
- Fabless design companies were emerging but had nowhere to manufacture their chips
- The industry was inefficient, leaving opportunity for someone who could specialize
Chang saw the future: Design and manufacturing should separate. Specialization would create efficiency. Someone could build a company that just manufactured chips—for everyone.
But Texas Instruments wasn't interested in this model. They were committed to vertical integration.
The Detour Through General Instrument (1984-1985)
Chang briefly led General Instrument's semiconductor business, but his vision was bigger than running someone else's chip division. At age 53—when most executives are thinking about retirement—Chang was thinking about revolution.
Then Taiwan came calling.
The Taiwan Opportunity (1985)
Taiwan in the 1980s was an economic success story, but its prosperity was built on low-cost manufacturing and electronics assembly. The government wanted to move up the value chain into high-technology industries.
Taiwan's Minister of Economic Affairs recruited Morris Chang with an extraordinary offer: Come build a semiconductor industry for Taiwan. We'll provide government backing, initial capital, and complete authority to execute your vision.
Why Taiwan?
For Chang, Taiwan offered what nowhere else could:
- Government support: Initial funding and political backing
- Labor advantages: Highly educated workforce at lower cost than U.S.
- Manufacturing culture: Precision and quality control already established
- Geographic position: Between Japan and U.S., in growing Asian market
- Personal connection: Chinese language and culture, but democratic system
- Clean slate: No existing semiconductor industry meant no resistance to new model
Chang accepted. In 1985, he moved to Taiwan to build what would become TSMC.
He was 54 years old. Most people that age are winding down careers. Chang was about to change the world.
Part II: The Foundry Model—A Revolution in Business and Technology
What Chang Proposed Was Radical
The semiconductor industry in 1987 operated on a simple model: companies that designed chips also manufactured them. Intel, Motorola, Texas Instruments, National Semiconductor—all were "integrated device manufacturers" (IDMs).
Chang's proposal: Build a company that ONLY manufactures chips. Never design your own. Never compete with customers. Be a pure-play foundry for hire.
The Foundry Model:
- TSMC would manufacture chips for anyone who designed them
- Customers (fabless design companies) could focus on chip design without capital-intensive manufacturing
- TSMC would never design its own products—avoiding customer competition
- Economy of scale: One fab serving many customers more efficient than each company building its own
- Specialization: TSMC could become world-class at manufacturing while customers focused on design
This sounds obvious now. In 1987, it was revolutionary and risky.
Why Everyone Thought It Would Fail
Industry veterans were skeptical for real reasons:
The Objections to Pure-Play Foundry:
1. "Nobody will trust you with their designs"
- Chip designs are incredibly valuable intellectual property
- Handing designs to third party creates theft risk
- What stops foundry from stealing designs and selling to competitors?
2. "You'll never achieve competitive manufacturing quality"
- Semiconductor manufacturing requires decades of accumulated expertise
- Taiwan had no semiconductor industry to build on
- How could startup match TI, Intel, Motorola's manufacturing prowess?
3. "The economics don't work"
- Fabs cost billions to build
- Need to run at high capacity to be profitable
- How do you guarantee enough customer demand?
4. "Customers will always prefer vertical integration"
- Control over entire process valuable
- Why would companies give up manufacturing?
- Fabless model unproven
Chang had answers to each objection. But answers don't matter if you can't execute.
February 21, 1987: TSMC Is Born
Taiwan Semiconductor Manufacturing Company was founded with:
- $220 million initial capital (Taiwan government provided significant portion)
- Joint venture structure: Government, Philips Electronics, and private investors
- Morris Chang as President with nearly complete operational authority
- The mission: Prove the foundry model works
The skeptics gave TSMC 5 years before failure. They were spectacularly wrong.
Part III: Building Trust, Building Monopoly
The First Customer Problem
TSMC's initial challenge: Nobody trusted them.
Why would a chip design company hand over valuable intellectual property to an unproven Taiwanese foundry? The risk of IP theft, manufacturing defects, or business failure seemed too high.
Chang solved this through:
The Trust-Building Strategy:
1. Ironclad IP Protection:
- Contractual guarantees of design confidentiality
- Physical separation of customer designs within facilities
- Third-party audits of security practices
- The nuclear option: TSMC would never design its own chips, eliminating incentive to steal
2. Quality Obsession:
- Recruited top manufacturing talent from U.S. and Japan
- Invested heavily in process development
- Guaranteed yields matching or exceeding industry standards
- Customer defect rates had to be lower than their own fabs
3. Customer Service Excellence:
- Responsive to customer needs in ways big IDMs weren't
- Flexible on order sizes (would manufacture small runs)
- Transparent pricing and capacity allocation
- Made it easy to work with TSMC
The Breakthrough: Early Customers
TSMC's first major customers were small fabless design companies that didn't have alternatives:
- Smaller chip designers couldn't afford to build their own fabs
- Companies pivoting from IDM to fabless needed manufacturing partner
- Startups with good designs but no capital for fabrication
Chang deliberately targeted these customers. They were desperate enough to take a chance. If TSMC delivered for them, word would spread.
It worked.
The Flywheel Effect (1990s)
By the early 1990s, TSMC had proven the model. Success created a powerful flywheel:
The TSMC Flywheel:
- More customers → More revenue → More R&D investment
- More R&D → Better manufacturing processes → Higher quality/better yields
- Better processes → Attract even more customers (including larger ones)
- Higher volume → Economies of scale → Lower costs per chip
- Lower costs → More competitive pricing → Attracts more customers
- Cycle repeats, accelerating
By mid-1990s, even major chip companies were considering fabless models. Why invest billions in fabs when TSMC could manufacture chips cheaper and better?
The Defection of the Giants
The real validation came when large companies started using TSMC:
- Qualcomm became fabless, manufacturing at TSMC
- NVIDIA founded in 1993 as fabless company, relied on TSMC from beginning
- AMD spun off manufacturing (GlobalFoundries), became TSMC customer
- Even Intel's competitors increasingly manufactured at TSMC
The foundry model wasn't just viable—it was superior. And TSMC was becoming the only game in town for advanced chips.
Part IV: The Race to the Cutting Edge
Technology Leadership Wasn't Guaranteed
Manufacturing chips for others was one thing. Being the BEST at manufacturing—staying ahead technologically—required relentless innovation.
Through the 1990s and 2000s, TSMC faced serious competition:
The Competitors:
- Intel: World's best chip manufacturer, but vertically integrated (didn't do foundry work)
- Samsung: Massive investment, aggressive technology development, willing to do foundry work
- GlobalFoundries: Spun out from AMD, positioned as TSMC competitor
- UMC (Taiwan): Smaller Taiwanese foundry, struggled to keep pace
- SMIC (China): State-backed Chinese foundry, years behind but trying
The Technology Race: Node by Node
Semiconductor technology advances through "process nodes"—the size of the smallest features that can be manufactured on a chip. Smaller nodes mean:
- More transistors per chip (more powerful processors)
- Better energy efficiency (longer battery life)
- Higher performance (faster computing)
The progression: 180nm → 130nm → 90nm → 65nm → 45nm → 32nm → 22nm → 16nm → 10nm → 7nm → 5nm → 3nm...
TSMC's Technology Leadership Timeline:
- 1990s: Playing catch-up to Intel, Samsung
- Early 2000s: Achieving parity at larger nodes
- 2010-2015: Pulling ahead at 28nm and 20nm
- 2016-2018: Clear leadership at 10nm and 7nm
- 2020: First to high-volume 5nm production (Apple A14 chip)
- 2022: First to 3nm production
- 2024: Working on 2nm, planning 1.4nm
The result by 2024: TSMC is 2-3 years ahead of Samsung and 5+ years ahead of everyone else at the cutting edge.
How TSMC Won the Technology Race
Why did TSMC pull ahead of competitors with more resources (Intel, Samsung)?
TSMC's Competitive Advantages:
1. Focus:
- ONLY does manufacturing (no distraction from product design)
- 100% of R&D budget goes to manufacturing processes
- Samsung splits focus between memory, displays, phones, and foundry
- Intel distracted by design challenges and strategic missteps
2. Customer Feedback Loop:
- Working with hundreds of customers reveals problems faster
- Learn from diverse chip designs what works and what doesn't
- Customers like Apple push TSMC to cutting edge
3. Manufacturing Culture:
- Taiwanese work ethic and precision
- Engineers sleeping at fabs during critical production runs
- Obsessive attention to yield rates and defect reduction
- Culture of continuous improvement
4. Massive Investment:
- $30-40 billion annual capex (capital expenditures)
- Reinvesting profits into next-generation technology
- Each new node requires billions in R&D
- TSMC spends more than competitors because larger customer base justifies it
5. The Morris Chang Factor:
- Long-term strategic vision
- Willingness to make massive bets on next-generation technology
- Relentless push for excellence
- Chang remained CEO until 2018 (retired at age 87!)
The Intel Collapse
Perhaps most shocking: Intel fell behind.
Intel—the company that defined semiconductor manufacturing excellence for decades—stumbled badly in the 2010s:
- 10nm delays: Took years longer than planned
- 7nm abandoned: Technical challenges forced rebranding (now "Intel 4")
- Lost Apple as customer: Apple switched from Intel to ARM chips manufactured by TSMC
- Lost technology leadership: TSMC now definitively ahead
By 2020, the unthinkable had happened: TSMC was better at manufacturing chips than Intel.
Intel is now trying to become a foundry (offering manufacturing services to others)—copying the model TSMC pioneered 35 years ago. Whether Intel can catch up remains uncertain.
Part V: The Monopoly That Ate the World
The Numbers Tell the Story
TSMC's Market Dominance (2024):
Overall Foundry Market Share:
- TSMC: 62% of global foundry revenue
- Samsung: 13%
- GlobalFoundries: 6%
- SMIC: 5%
- Everyone else: 14%
Advanced Node Dominance (7nm and below):
- TSMC: Over 90% market share
- Samsung: Most of remaining share
- Nobody else can manufacture at these nodes at scale
The Customer List:
- Apple: A-series chips for iPhones, M-series for Macs (TSMC exclusive)
- NVIDIA: GPU chips for graphics cards and AI (TSMC manufactured)
- AMD: CPUs and GPUs (TSMC manufactured)
- Qualcomm: Smartphone processors (TSMC manufactured)
- MediaTek: Mobile and IoT chips (TSMC)
- Broadcom, Marvell, countless others
Financial Scale:
- Revenue: ~$70 billion (2023)
- Market cap: $800+ billion
- Employees: 73,000+
- Annual capex: $30-40 billion
What This Monopoly Actually Means
Every advanced technology product on Earth depends on TSMC:
- Your smartphone (iPhone, high-end Android)
- Your laptop (Apple Silicon, AMD Ryzen)
- Your graphics card (NVIDIA GeForce, AMD Radeon)
- Data center servers (AMD EPYC, NVIDIA AI chips)
- AI chips powering ChatGPT and every other AI system
- Advanced military electronics
- Autonomous vehicle processors
- 5G infrastructure
If TSMC stopped production tomorrow, the global technology industry would experience instant, catastrophic collapse. There are no alternatives for advanced chips. None.
How Did We Let This Happen?
The concentration of semiconductor manufacturing in one company wasn't a conspiracy. It was the natural result of economic forces:
Why Monopoly Emerged:
1. Economics of Scale
- Each new fab costs $15-20 billion
- Must run at high capacity to be profitable
- Larger players spread costs across more customers
- Smaller foundries couldn't compete on price
2. Technology Barriers
- Leading-edge manufacturing incredibly difficult
- Requires decades of accumulated expertise
- Each new node exponentially harder than previous
- Most competitors fell behind and gave up
3. Customer Stickiness
- Switching foundries expensive and risky
- Chip designs optimized for specific manufacturing processes
- Once TSMC proved reliable, customers stayed
4. Network Effects
- More customers → Better processes → Attracts more customers
- Ecosystem of suppliers, tools, expertise built around TSMC
- Self-reinforcing dominance
Everyone chose TSMC because TSMC was best. Now we're trapped by that choice.
Conclusion: The Godfather's Legacy
What Morris Chang Created
Morris Chang is now 93 years old. He retired from TSMC in 2018 after 31 years as CEO—one of the longest and most successful corporate leadership tenures in history.
What he built is extraordinary by any measure:
- The most important manufacturing company in the world
- A Taiwanese national champion worth more than $800 billion
- The foundation of global technology infrastructure
- Living proof that the foundry model works
But Chang also created something else: the most dangerous single point of failure in the global economy.
The Uncomfortable Truth:
TSMC's success made the world more efficient and technologically advanced. It also made the world vastly more vulnerable.
We optimized for performance and cost. We didn't consider what happens when:
- Geopolitical tensions between China and Taiwan escalate
- The most important company in the world sits 100 miles from a major power that claims sovereignty over its location
- There are no backup plans because nobody else can manufacture advanced chips at scale
Morris Chang himself has acknowledged this dilemma. In a 2021 interview, he called geopolitical tensions around Taiwan "very troublesome" and admitted that TSMC's concentration in Taiwan creates risks.
But by then, it was too late to change course. The entire technology industry had been built around TSMC's dominance.
The Silicon Shield Strategy
Taiwan's government has a name for TSMC's strategic importance: "The Silicon Shield."
The theory: TSMC makes Taiwan indispensable to the global economy. China wouldn't invade because:
- Destroying TSMC would crater the global economy (including China's)
- The world would defend Taiwan to protect chip supplies
- Taiwan's technological crown jewel provides security through dependence
Does the Silicon Shield Work?
The Optimistic View:
- China hasn't invaded despite claiming Taiwan for 75 years
- U.S. commitment to defend Taiwan strengthened by chip dependence
- Economic cost of invasion (TSMC destruction) astronomical
- TSMC's importance gives Taiwan leverage in international relations
The Pessimistic View:
- Making yourself indispensable also makes you a target
- China might calculate that controlling/capturing TSMC worth the risk
- Silicon Shield only works until it doesn't—one miscalculation and it fails catastrophically
- Dependence creates vulnerability, not security
Morris Chang created a company so important that it might prevent a war. Or it might cause one. Possibly both.
Chang's Own Assessment
In retirement, Morris Chang has become more candid about the geopolitical implications of what he built.
Key quotes reveal his thinking:
On TSMC's Strategic Importance (2021):
"Geopolitically, TSMC is a very important company. It's an important company for the free world."
On U.S. Attempts to Replicate TSMC (2022):
"The U.S. is not able to have a complete semiconductor supply chain domestically. That will be very high cost. It will be economically unfeasible."
On Taiwan's Vulnerability (2021):
"Taiwan is a very dangerous place to invest at the moment." (He later clarified but the initial assessment was stark)
On the Future:
"I hope that will not be the case, but right now it's a troubled time."
Chang understands what he created better than anyone: A technological marvel that's also a geopolitical time bomb.
The Questions Chang's Legacy Raises
As we examine TSMC's story through this series, Morris Chang's creation forces us to confront uncomfortable questions:
- Was efficiency worth the vulnerability? We got better chips and lower costs by concentrating manufacturing. But we created systemic risk.
- Can we escape the dependency? Or is the technology too hard, the investment too large, the expertise too concentrated?
- Who really benefits from TSMC's existence? Taiwan gains security (maybe). The world gains technology. But at what cost?
- What happens if China invades? This isn't hypothetical—it's the question that keeps Pentagon planners awake at night.
The Central Dilemma of Our Era:
We built a global technology system that works brilliantly—as long as nothing goes wrong in a 100-mile stretch of water between Taiwan and mainland China.
Morris Chang gave us the future. But he also gave us the world's most dangerous chokepoint.
What Comes Next
Understanding TSMC's dominance is just the beginning. The rest of this series explores:
- How the technology actually works (Chapter 2) - Why advanced chips are so impossibly hard to make
- The geography problem (Chapter 4) - Taiwan, China, and 100 miles of the most dangerous water on Earth
- The invasion scenarios (Chapter 5) - What actually happens if China attacks
- The escape attempts (Chapters 7-9) - Can America replicate TSMC? Can Europe? Can China build alternatives?
- The future (Chapter 12) - Three scenarios for how this ends
But first, we need to understand why TSMC's monopoly is so hard to break.
Why can't competitors just build better fabs? Why can't countries simply invest their way to semiconductor independence?
The answer lies in the technology itself—and in the next chapter, we'll explore why making advanced semiconductors is the hardest manufacturing challenge humanity has ever attempted.
Because if you understand why it's so hard, you'll understand why we're trapped. And why Morris Chang's creation is both miracle and curse.
Sources & References
Primary Sources on Morris Chang and TSMC History:
- TSMC corporate history and annual reports (1987-2024)
- Morris Chang interviews and public statements (various sources, 1990s-2024)
- Taiwan government economic development archives
- Semiconductor Industry Association historical data
Biographical Sources:
- Morris Chang autobiography and authorized biographical materials
- Interviews in Commonwealth Magazine (Taiwan), Financial Times, Wall Street Journal
- MIT and Stanford alumni records and publications
Industry Analysis:
- Gartner, IC Insights - Foundry market share data and analysis
- TechInsights, Semiconductor Intelligence - Technology node tracking
- Industry publications: EE Times, Semiconductor Engineering
Strategic Analysis:
- Center for Strategic and International Studies - Semiconductor geopolitics
- Peterson Institute for International Economics - Taiwan economic analysis
- Council on Foreign Relations - Taiwan strait security assessments
Books:
- Chip War by Chris Miller - Comprehensive semiconductor history
- The Chip by T.R. Reid - Earlier semiconductor industry history
Methodology Note: This chapter synthesizes TSMC's history from corporate disclosures, biographical sources, industry analysis, and strategic assessments. Market share data from multiple industry sources. Morris Chang quotes from various interviews across three decades, with sources cited where specific quotes used. Technology timeline verified against multiple semiconductor industry tracking sources.
No comments:
Post a Comment