CHAPTERS
Why Acquired went back to Taiwan: setting up the Morris Chang interview
Ben and David explain how their semiconductor obsession led them from the original TSMC episode to a Hail Mary ask: getting Jensen Huang to help secure an interview with 93-year-old TSMC founder Morris Chang. They also note the importance of Chang’s newly released (Chinese-only) autobiography volume and the translated excerpts that informed the conversation.
The Jensen Huang letter: how TSMC and NVIDIA became partners (1997)
Chang recounts receiving a mailed letter from a small, struggling NVIDIA after TSMC’s San Jose office didn’t respond. He personally called Jensen without warning, met him, and was impressed by Jensen’s optimism and clarity—kicking off a relationship that turned NVIDIA into a top TSMC customer within a few years.
The 40nm crisis (2009): yield, quality disputes, and leadership transition
The hosts press on the infamous 40nm period where slow ramp, yield issues, and alleged quality problems hurt customers—especially NVIDIA. Chang explains this occurred while a successor CEO was in place, and how internal insistence that TSMC wasn’t at fault prevented a timely customer resolution.
Layoff backlash and the protest at Chang’s home: a cultural turning point
Chang describes how performance-review-based terminations during the financial crisis triggered employee protests at his home, including an overnight encampment. The episode highlights Chang’s belief that layoffs based on subjective performance ratings damage credibility—and how these events helped precipitate his return as CEO.
Re-taking the CEO role: fixing customers, margins, and reorganizing priorities
After returning as CEO, Chang outlines a stack of urgent problems: customer trust (NVIDIA), operational issues (40nm), and deteriorating economics where pricing fell faster than costs. He also kept the prior CEO but reassigned him to “new businesses,” reflecting TSMC’s temporary interest in solar and LEDs.
Settling the NVIDIA dispute: the pizza-and-salad negotiation and $100M+ resolution
Chang details his hands-on approach to resolving NVIDIA’s grievance: weeks of internal investigation, then a personal dinner at Jensen’s home followed by a firm, time-limited settlement offer. NVIDIA accepted within 48 hours, preserving a pivotal long-term partnership.
Betting on 28nm: locking in an 8% R&D rule and a major CapEx ramp
Chang explains how TI’s underinvestment scarred him and led to a simple policy: fix R&D at 8% of revenue to avoid annual fights. With R&D and strategic marketing forecasting 28nm as a “sweet spot,” he pushed a major capital spending increase despite board skepticism, capturing the smartphone wave.
Building strategic marketing leadership: CC Wei’s pivot into business development
Chang recounts dismantling a divisional split (advanced vs mainstream) in favor of a functional structure, echoing earlier debates (including a McKinsey study). He then places CC Wei into business development/marketing—an unusual move designed to create a well-rounded leader who could find and shape the next growth engines.
Apple arrives unexpectedly: Terry Gou dinner and Jeff Williams’ direct ask
A surprise dinner invitation from Foxconn founder Terry Gou brings Apple COO Jeff Williams to Chang’s home. Williams quickly gets to the point—Apple wants TSMC to manufacture wafers—but the request targets 20nm, not the 28nm TSMC had just invested heavily in.
Financing Apple’s 20nm ‘detour’: Goldman Sachs, bonds, and taking only half the demand
Chang explains that serving Apple required additional massive capital, so TSMC consulted Goldman Sachs (built via ADR history and Chang’s GS board role). Rather than cut dividends or issue stock, TSMC chose borrowing and limited Apple’s requested volume to reduce risk—anticipating how customers can overstate needs.
Intel enters the picture—and Tim Cook’s verdict: ‘Intel doesn’t know how to be a foundry’
Apple pauses discussions after Intel CEO Paul Otellini approaches Tim Cook, prompting concern about losing the account. Chang visits Cupertino and hears directly from Cook that Intel lacks the foundry mindset—especially in customer service and responsiveness—reinforcing why TSMC’s pure-play model wins trust.
Pricing and the real trade-off: 20nm delays 16nm and briefly gives Samsung an opening
After pricing negotiations, Chang highlights the hidden cost of Apple’s 20nm request: TSMC couldn’t develop 20nm and 16nm in parallel at the time, delaying 16nm. Samsung’s faster 16nm progress leads Apple to place initial 16nm orders there, shocking Chang—until Williams promises Apple will shift back once TSMC is ready.
Seeing the industry early: the IBM–Qualcomm signal and refusing co-development
Chang uses an earlier story to illustrate foundry pattern recognition: when Qualcomm shifts foundry volume away from IBM, he infers IBM Semiconductor is in trouble. IBM then proposes co-developing 130nm with TSMC—an offer Chang rejects to avoid dependency and cultural/operational friction—while UMC accepts and later regrets it.
Learning curve, scale flywheel, and why TSMC’s dominance became inevitable
Chang traces his learning-curve thinking back to TI and BCG founder Bruce Henderson, with Bill Bain working closely alongside him. The discussion connects learning curves to the foundry flywheel: volume drives cost down, cost wins more volume, and the resulting cash funds the next node—setting up market structure that favors a dominant leader.
Postgame reflections: pure-play trust, Arm’s role, and the Hsinchu ecosystem moat
Back in their home studios, Ben and David interpret the interview’s implications: pure-play foundry trust is TSMC’s core advantage; industry modularization (including Arm’s rise) enabled fabless + foundry dominance; and Hsinchu Science Park’s co-located ecosystem makes replication elsewhere extraordinarily hard. They connect it to learning-curve inevitability and Moore’s Law demand as the tailwind.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome