holocron
中/en
← Business
Business

Morris Chang

In one sentence: A man who spent 25 years in American semiconductors, got "put out to pasture" twice, and at 56 flew to Taiwan with zero equity in the company he was about to found — and out of the pricing discipline he'd learned at Texas Instruments built the most durable moat in the history of the chip industry. TSMC：纯代工模式发明者 (the 2021 company history) and Morris Chang 访谈：张忠谋亲述 TSMC (the 2025 interview, recorded when he was 93) approach him from opposite directions — corporate strategy versus personal decision-making — but land on the same underlying habit of mind: design the company for the constraints, not for the ambition.

Life Timeline

YearEvent
July 1931Born in Ningbo, China; his father a county official who later became a bank manager. David opens the TSMC episode by placing him beside Warren Buffett（沃伦·巴菲特）, born one year earlier in Omaha
1937–1948Lives through three wars and flees three times before turning 18: to Hong Kong in 1937; back to Shanghai after Japan invades Hong Kong three hours after Pearl Harbor (Dec 8, 1941); back to Hong Kong again in 1948 as the Chinese Civil War breaks out
1949–1954At 18, with an uncle's help in Boston, gets into Harvard; transfers to MIT in fall 1951 for mechanical engineering (Harvard had no undergraduate engineering program), finishing undergrad and master's in three years
~1954–55Fails his MIT PhD qualifying exams twice and has to job-hunt; Ford offers $479/month (his dream job), Sylvania's new Semiconductor Division offers $480 — for the sake of $1, semiconductors got Morris Chang
1958Joins Texas Instruments (the same year Kilby and Noyce separately co-invent the integrated circuit); his first project, the IBM 7090 second-source line, goes from 0% to 20% yield in about four months (double IBM's own 10%)
1961–64TI sponsors his Stanford PhD on full salary; he finishes in two and a half years
1967Promoted to division general manager; with a then-tiny BCG (Bill Bain embedded on-site for nearly two years), co-invents learning-curve pricing
1972Promoted to VP over TI's entire worldwide semiconductor business, one level below CEO; R&D sits at 4.8% of revenue, and his push for 5.5% "was denied every time I raised it"
1978–83Moved to VP of Consumer Products, five years without a turnaround; demoted in 1983 to "head of quality and people effectiveness" and resigns at 52 ("My career at TI, 30 years, done" — by his 1958 start date it was actually 25)
1984–85COO of General Instrument, a culture he calls a "proto tech private equity firm," at odds with his own values; in his final months there he watches Gordie Campbell's funding ask collapse from $50M to $5M ("I'm not going to build a fab") and fabless Atmel show up asking GI to foundry its wafers — a live prototype of the model he'd found a couple of years later; gone within a year at 54
1985KT Li recruits him to run ITRI; he takes it as a near-retirement, considering himself "financially pretty secure... I could live according to the way I desire, which was actually pretty modest, for the rest of my life"
1987KT Li gives him three days to write a business plan; TSMC is founded (he's 56) at a $0 pre-money valuation with $220M raised (government 50%, Philips 28%, the Premier strong-arming the remaining ~22% out of local business leaders) — Morris gets no equity. The global semiconductor market that year is about $26B (his own figure, from the interview)
1987–early 1990sThe first 5–7 years run mostly on non-strategic overflow orders dumped by IDMs; fabless customers are not yet the majority
1990sChronic capacity shortage; institutes customer deposits and coins the threat of "confiscate" — though no deposit is ever actually seized
1994 / 1997Lists on the Taiwan Stock Exchange (~$4B market cap) / lists on the NYSE
1995Annual revenue crosses $1B
1996President Don Brooks proposes splitting the company by fab; McKinsey's two-month, multi-million-dollar study concludes functional is best, citing Boeing
1997Jensen Huang (Jensen Huang（黄仁勋）) mails a letter to Hsinchu after TSMC's San Jose office never calls back; Morris phones him cold the next week — "Quiet. Morris Chang is calling me." NVIDIA is a top-five TSMC customer within 2–3 years
1997–99Qualcomm abruptly shifts its orders to TSMC (its previous supplier, IBM Semiconductor, is in trouble); in 1999 IBM proposes co-developing the 130nm node, TSMC declines without a second thought, IBM goes to UMC instead — which regrets it within a few years
2001–03The internet recession (Q1 2001 to Q3 2003); deposits are rolled forward rather than confiscated
2005First retirement at 74, handing the CEO role to Rick Tsai while staying chairman; operations splits into advanced (Mark Liu) and mainstream (C.C. Wei) technology groups
2007–09The iPhone launches; the financial crisis hits; the sitting CEO cuts 600–700 people under the label of "worst performance review," triggering two rounds of protest outside Morris's home — the second defused by his wife Sophie delivering breakfast to protesters camped in a nearby park at dawn
2009Returns as CEO at 78; spends nearly half of his first 4–5 weeks settling the NVIDIA 40nm dispute (a home dinner followed by a study-room offer of over $100M with a 48-hour deadline); merges the two operations groups; fixes R&D at 8% of revenue
2010CapEx jumps from roughly $2–2.5B a year over the prior decade to nearly $6B, all on 28nm, over an independent-director revolt before the board meeting
Nov 2010Terry Gou brings Apple COO Jeff Williams to dinner at Morris's home: "We like you to foundry our wafers" — but Apple wants 20nm, not the soon-to-ramp 28nm
2011Financial planning with Goldman Sachs: no dividend cut, no new equity, bonds only, and TSMC takes only half of Apple's stated demand; in February, Intel (Paul Otellini) lobbies Tim Cook and talks pause two months; in March, Tim Cook settles it over a cafeteria lunch — "Intel just does not know how to be a foundry"; Jeff Williams later sets the price himself over dinner at a three-star Taipei restaurant
2011–12Taking 20nm delays 16nm development; Samsung, having lost the 20nm business, skips straight to 16 and gets there first
~2013–14Apple places its first 16nm orders with Samsung ("the real shock"); about half a year later TSMC's own 16nm is ready and most of Apple's 16nm demand returns
2013 / June 2018Steps down as CEO again after locking in Apple, stays chairman / retires completely at 86 (gross margin was still slightly under the 50% he'd chased for years — it clears that bar only after his tenure, via COVID and process leadership, reaching 55–57% today)
2021 (at TSMC：纯代工模式发明者 recording)Global chip shortage; $550B market cap, ninth in the world
Late 2024Memoir volume two published in traditional Chinese, 26 years after volume one
Jan 2025 (at Morris Chang 访谈：张忠谋亲述 TSMC recording)93 years old, interviewed in his own Taipei office; market cap has doubled since 2021, past $1T

How He Thinks — Key Decisions

1. Design the company for the constraints, not for the ambition (source: acquired-tsmc) — Given three days to write a business plan in 1987, he inventoried Taiwan's endowments: "We had no strength in research and development... The only possible strength in Taiwan... was semiconductor manufacturing... The answer was a pure-play foundry." Structurally identical to his 1950s choice to transfer to MIT after reading the ceiling on Chinese career paths in America.

2. The learning curve is an operating system, not a slogan (source: acquired-morris-chang, with the earlier TI pricing story from acquired-tsmc) — Around 1970, working with BCG's Bill Bain for nearly two years, he refined the experience curve into a tool "a semiconductor company can use effectively": the goal is to be the largest-volume player at the end of the game, so you cut prices toward the endgame level early — even at a loss — to squeeze out competitors and aggregate demand. The risk is just as extreme: a 5–10% demand forecast miss can tank an entire node generation's profitability.

3. Never neglect a small customer (source: acquired-morris-chang) — In 1997, near-bankrupt NVIDIA (50–60 people) had to mail a letter to Hsinchu because TSMC's San Jose office wouldn't call back; Morris phoned personally and showed up the next day. "I had always told our salespeople that we should never be negligent in talking to future customers, even if the customer seems to be a very small one."

4. Don't litigate with your best customer: a unilateral fair price plus a deadline beats haggling (source: acquired-morris-chang) — Retaking the CEO seat in 2009 with the NVIDIA 40nm dispute unresolved (his predecessor had offered zero), he spent weeks mastering the facts, then offered a number — over $100M, 48 hours, no negotiation — in the study after a family dinner, physically separating the relationship from the deal.

5. Big capital decisions run on responsibility, not persuasion (source: acquired-morris-chang) — He fixed R&D at 8% of revenue to end the annual budget fight; when independent directors revolted before the 2010 CapEx tripling, he admitted "you really can't convince anybody on something like this" and fell back on governance: "I am still the guy that's responsible for the operation of the company. You need to let me go ahead with this one."

6. Bet the company, but stay confident (source: acquired-morris-chang, with the Apple narrative from acquired-tsmc) — Taking the Apple deal was, in Ben's framing, a bet-the-company move. Morris: "Yeah, I know, bet the company. But I didn't think I would lose." The hosts note that Jensen has said the exact same words.

Quotes, Verbatim

"My reaction entering Harvard was sheer ecstasy, almost disbelief. What a country!" — acquired-tsmc

"In the early 50s in the United States, there were Chinese laundrymen, Chinese restaurateurs, Chinese engineers, and Chinese professors. Those were the only respectable professions for Chinese." — acquired-tsmc

"We had no strength in research and development... The only possible strength in Taiwan... was semiconductor manufacturing... The answer was a pure-play foundry." — acquired-tsmc

"It was like in the movie The Godfather. It was an offer I couldn't refuse." — acquired-tsmc

"Quiet. Morris Chang is calling me." — what Jensen shouted at his own staff when Morris called cold in 1997, as Morris recounts it — acquired-morris-chang

"Yeah, I know, bet the company. But I didn't think I would lose." — acquired-morris-chang

"Unlikely, in your opinion." — his reply when Ben calls TSMC an "unlikely success of grand scale" — acquired-morris-chang

"The semiconductor business is like a treadmill that speeds up all the time. If you can't keep up, you fall off." — acquired-tsmc

Two Episodes, Two Angles

acquired-tsmc (2021, company history) is an outsider's strategic retrospective: it runs TSMC through the 7 Powers checklist and reads the pure-play choice as a case of seeing more clearly from the outside ("What very few people saw, and I can't tell you that I saw was the rise of the fabless industry, I only hoped for it. But I had better reasons for hoping for it... because I was now standing outside") — then pushes the argument all the way out to Taiwan Strait geopolitics and "semiconductors are the new oil." The register is the hosts' own awe: "everything but bulletproof."

acquired-morris-chang (2025, the interview) is first-person decision archaeology: 93-year-old Morris reconstructs the NVIDIA 40nm negotiation, the three rounds of Apple pricing, and the origin of the 8% R&D rule down to the minute ("8:00 sharp I looked at my watch"). The register is restrained and de-mythologizing — asked if TSMC was an "unlikely success of grand scale," he answers only "Unlikely, in your opinion," then adds "it did exceed my expectations... but not by an order of magnitude." The two episodes genuinely disagree on one point: whether pure-play was foresight. The company-history episode frames it as an epistemic edge; in the interview's own post-game, the hosts concede it was more "what could we win at" than prophecy — a strategy with "an accident of history" component (Ben quotes Steve Jobs: "you can connect the dots looking backwards"). The two accounts complement rather than contradict: one records the retrospective strategic reading, the other records the honest attribution at the moment of decision.