CHAPTERS
- 0:00 – 2:31
Why TSMC matters now: remastered episode and the chip-world stakes
Ben and David reintroduce their 2021 TSMC episode (now remastered) and explain why it’s even more relevant in the AI era. They position TSMC as the de facto manufacturer of leading-edge chips for Apple, Nvidia, and many other fabless giants—and a rare trillion-dollar outlier outside the US West Coast.
- •TSMC’s role as the leading-edge manufacturer behind iPhones, Macs, Nvidia, AMD, Qualcomm, hyperscalers
- •Semiconductors’ rising importance from smartphone era to AI era
- •TSMC market-cap growth and global uniqueness
- •Set-up for a story combining technology, business strategy, and geopolitics
- 2:31 – 4:51
TSMC episode framing: the unknown giant and Morris Chang’s improbable arc
The hosts frame TSMC as a company few consumers know despite it being among the world’s most valuable. They preview Morris Chang’s unusual founder journey—founding at 56, retiring and returning multiple times, and eventually landing Apple’s defining manufacturing partnership.
- •TSMC as a “never heard of it, top-10 company” case study
- •Morris Chang’s late-founder, multi-retirement narrative
- •TSMC’s ubiquity across consumer tech, autos, defense, and infrastructure
- •Geopolitical tension around Taiwan as a critical backdrop
- 4:51 – 8:54
Morris Chang’s early life: war, displacement, and the path to America
David traces Chang’s childhood through multiple wars and relocations across China and Hong Kong. At 18, Chang leaves for Harvard—an enormous shift into stability and opportunity that shapes his worldview and ambition.
- •Born 1931 in Ningbo; family moves for father’s work
- •Three major conflicts before age 18 (Sino-Japanese War, WWII, Chinese Civil War)
- •Accepted to Harvard in 1949 with help from an uncle
- •Chang’s later reflections on US stability and leadership
- 8:54 – 17:43
Harvard to MIT: engineering focus, failure, and a career detour into semiconductors
Chang transfers from Harvard to MIT to study mechanical engineering, then fails his PhD qualifying exams twice—forcing him into industry. A near-Ford job offer and a $1/month difference (apocryphally) pushes him to Sylvania’s semiconductor division, where he teaches himself EE.
- •Harvard lacked undergrad engineering; MIT becomes the path
- •PhD qualifying failures redirect him into industry
- •Fork-in-the-road job choice: Ford vs Sylvania
- •Self-education via Shockley’s semiconductor textbook
- •Creative mentorship: learning from a senior engineer at the hotel bar
- 17:43 – 24:27
Leaving Sylvania: recognizing mismanagement and discovering where the industry’s center really is
Chang realizes Sylvania isn’t positioned to win after hearing a manager admit the company can’t make what it can sell. Instead of Silicon Valley, he heads to the real semiconductor juggernaut of the era—Texas Instruments in Dallas.
- •Sylvania’s strategic dysfunction crystallizes Chang’s desire to leave
- •Myth-busting: TI (Dallas) as the dominant semiconductor force then
- •TI’s origins in oil exploration instruments
- •Context: integrated circuit invention era (Noyce/Kilby)
- 24:27 – 30:33
Texas Instruments breakthrough: yield engineering, credibility, and rapid ascent
At TI, Chang is assigned to a failing IBM transistor production effort with near-zero yields. Applying mechanical-process thinking, he doubles and then quadruples yields, earns executive visibility, and quickly moves into leadership—eventually being sponsored for a Stanford PhD.
- •IBM second-source transistor manufacturing partnership and yield crisis
- •Chang improves yields to ~20% in four months
- •Promoted into management with a new transistor development department
- •TI sponsors Stanford PhD while paying his salary
- •Return to TI as silicon becomes the dominant substrate
- 30:33 – 36:14
Learning-curve pricing: Chang’s business innovation that made TI #1 in ICs
As a GM at TI, Chang recognizes the economic logic of fabs: maximize utilization early to accelerate learning and yield improvements. With BCG’s help, TI pioneers learning-curve pricing—starting low and cutting prices proactively—driving volume, share, and profit leadership.
- •Fab economics: fixed-cost dominance makes utilization critical
- •Learning curve: volume accelerates yield and cost improvements
- •Controversial strategy: price cuts even without market pressure
- •Outcome: TI becomes the biggest and most profitable IC business globally
- 36:14 – 42:51
TI derailment: leadership track breaks, MOS transition misses, and the IBM PC CPU loss
Despite being a top CEO candidate, Chang is rotated into consumer products, struggles with the different skill set, and is later demoted. In parallel, TI misses the MOS wave and—most fatefully—loses the IBM PC CPU contract to Intel, cementing x86 dominance.
- •Possible discrimination + strategic mis-rotation to consumer products
- •Demotion to “Quality and People Effectiveness” as a career end
- •TI fails to lead MOS transition; talent drains to competitors
- •1980 IBM PC RFP goes to Intel (8088), not TI
- •A pivotal “off-ramp” moment in semiconductor history
- 42:51 – 50:43
General Instrument and the end of a US CEO dream—then an unexpected call from Taiwan
Chang leaves TI, tries General Instrument’s finance-driven, acquire-and-sell culture, and exits within a year. Concluding he won’t become CEO of a major US company, he’s recruited by Taiwanese minister K.T. Li to run ITRI—Taiwan’s attempt at a Bell Labs-like technology engine.
- •GI’s proto-private-equity model clashes with Chang’s technology orientation
- •Chang accepts the US CEO dream is over
- •K.T. Li recruits Chang to lead ITRI as Taiwan’s technology catalyst
- •Taiwan’s mid-80s reality: low margins, manufacturing-focused economy
- 50:43 – 58:13
From ITRI to TSMC: the ‘Godfather offer’ and inventing the pure-play foundry
At ITRI, Chang faces culture shock but inherits early semiconductor groundwork, including an RCA tech transfer and UMC’s formation. K.T. Li then directs Chang to start a world-class semiconductor company—leading Chang to choose a pure-play foundry model that fits Taiwan’s strengths and avoids its weaknesses.
- •ITRI culture clash: government-job mentality vs Chang’s intensity
- •RCA tech transfer is outdated but foundational; UMC and later MediaTek emerge
- •Taiwan’s strategic insight: manufacturing is the only scalable advantage
- •Chang’s defining leap: pure-play foundry as the right company type
- •“Offer I couldn’t refuse” urgency forces fast invention
- 58:13 – 1:13:51
Funding the ‘bad idea’: Philips backing, forced local capital, and the shocking $0 pre-money
Chang struggles to find customers and investors because “real men have fabs.” After Intel, TI, and others decline, Philips invests (28%), Taiwan’s government funds 50%, and local investors are strong-armed into the remainder—on a $0 pre-money where Chang receives no founder equity.
- •Chicken-and-egg: no obvious market for a pure-play foundry in 1987
- •Industry skepticism embodied by Jerry Sanders’ ‘real men have fabs’
- •Philips becomes the crucial foreign backer; total raise ~$220M
- •Government plus local capital completes the financing
- •Chang starts with zero equity and later buys shares personally
- 1:13:51 – 1:18:01
Creating the fabless era: surviving on IDM ‘dregs’ until startups arrive
TSMC’s early business is unstable: IDMs outsource only when capacity is tight or products are unprofitable. Chang’s deeper bet is that designers will finally start fabless companies once fabs are no longer required—turning TSMC into the enabling platform for semiconductor entrepreneurship.
- •Early revenue is volatile and low-quality: excess-capacity work and money-losing legacy chips
- •Chang’s insight from TI/GI: designers want to leave but can’t fund fabs
- •TSMC as the ‘YC/AWS for chips’: lowering the capital barrier to entry
- •First wave of fabless winners: Qualcomm, Broadcom, Nvidia, etc.
- •Platform logic: TSMC captures the upside of the whole innovation ecosystem
- 1:18:01 – 1:30:06
The flywheel and the hard-tech moat: CapEx, ASML/EUV, and leading-edge consolidation
As fabless demand grows, TSMC reinvests high margins into ever-more advanced fabs and tooling, reinforcing its technology lead and attracting more customers. The hosts unpack the modern semiconductor value chain and the pivotal role of ASML’s EUV lithography—an almost science-fiction manufacturing capability—driving the industry from 22 leading-edge players down to (nearly) one.
- •TSMC reinvestment flywheel: profits → CapEx → better nodes → stronger customers → more demand
- •Value chain unbundling: EDA, IP (Arm), fabless designers, equipment makers, foundries
- •ASML EUV: molten tin + laser-generated plasma; machines ~$200M+, tiny annual output
- •Export controls and geopolitics restrict EUV access (notably for China)
- •Leading-edge market shrinks from 22 → 14 → 6 → 2 (TSMC/Samsung), with TSMC trending to sole leader
- 1:30:06 – 1:41:38
Third act: Morris returns, seizes mobile (and later cloud/AI), and lands Apple
After retiring in 2005, Chang returns as CEO in 2009 amid the financial crisis, citing ‘golden opportunities’—the smartphone and the changing cloud workload landscape. The pivotal win is Apple: a high-risk, capacity-intensive partnership that helps cement TSMC’s dominance and shifts Apple away from Samsung dependence.
- •2009 CEO return at age 78; leadership reset after the crisis
- •Mobile + cloud as the new compute paradigms; Arm-based ecosystem accelerates
- •Apple-TSMC deal dynamics: massive investment, dedicated resources, no backup plan
- •Strategic incentive: Apple wants independence from Samsung amid fierce rivalry
- •Chang’s final retirement in 2018 after TSMC reaches clear top position
- 1:41:38 – 1:53:00
TSMC today: financial compounding, pricing power, and a ‘winner-take-all’ cost curve
The hosts quantify TSMC’s long-run growth, margins, and reinvestment intensity—highlighting extraordinary CapEx plans and a rare shift to price increases. They connect Moore’s Law with Rock’s Law (fab cost inflation) to explain why the leading firm can become structurally monopolistic.
- •27-year compounding: ~17% revenue CAGR; ~20% market-cap IRR from IPO base
- •Operating leverage: massive profits plowed back into CapEx ($100B over 3 years discussed)
- •Pricing power: moving from decades of price cuts to notable price increases
- •Leading-edge dominance: majority of profit + ~90% share at critical nodes
- •Moore’s Law + Rock’s Law logic implies winner-take-all dynamics
- 1:53:00 – 2:27:49
Moats, geopolitics, and the playbook: Seven Powers, Taiwan risk, and Intel ‘what-ifs’
Ben and David apply the Seven Powers framework, emphasizing scale economies, switching costs, and especially process power as TSMC’s defining moat. They then confront the existential geopolitical risk around Taiwan, explore onshoring diversification, and close with playbook lessons—especially Intel’s history of costly strategic misreads.
- •Seven Powers: strongest fit is process power; also scale economies and switching costs
- •Deep ecosystem integration with EDA/IP/tooling makes switching slow and costly
- •Geopolitical risk: Taiwan’s strategic vulnerability as the core bear case
- •Onshoring efforts (e.g., Arizona/Japan) vs the strength of Taiwan’s ecosystem
- •Intel alternate histories: iPhone chip bid, EUV skepticism, strategic decisions vs pricing
