Morris Chang: The Man Who Redesigned an Industry

One idea, held by a man with the experience to see it, changed how the world makes chips. This story covers the early life, the Texas Instruments years, the yield obsession, the invitation to Taiwan, the founding of TSMC, the return from retirement and the management philosophy — part of the Taiwan Company Stories hub.

How did Morris Chang learn semiconductors?

By running manufacturing at Texas Instruments during the industry’s formative decades. Chang joined the American semiconductor industry in the 1950s, studied engineering at MIT and later completed a doctorate at Stanford while working, and rose to lead TI’s global semiconductor business.

His formative insight came from yield. Early in his career he was assigned to a struggling production line and improved its yield dramatically through systematic experimentation, learning that manufacturing outcomes in semiconductors are determined by disciplined process control rather than by equipment alone — and that yield improvement is a learning curve that compounds.

He also learned pricing strategy, most notably the practice of pricing ahead of the cost curve to win volume that would then drive costs down. That understanding of the relationship between scale, learning and cost underpins everything he later built.

Why did he leave the United States?

Because his path at Texas Instruments had reached its limit and an unusual opportunity arrived. After being passed over for the chief executive role, he moved to General Instrument briefly, then accepted an invitation from the Taiwanese government to lead the Industrial Technology Research Institute in 1985.

The invitation reflected Taiwan’s strategy of recruiting experienced overseas Chinese technologists to build industrial capability. Chang arrived at fifty-four with a mandate to develop the island’s technology industries and considerable freedom to decide how.

What he found was a country with engineering talent, government willingness to invest and no obvious path into an industry dominated by American, Japanese and European integrated manufacturers. Competing directly was hopeless; a different business model was the only viable entry.

What made the foundry idea contrarian?

That it appeared to occupy the least attractive position in the industry. Conventional wisdom held that value lay in designing and selling chips, with manufacturing as a necessary cost; a company doing only manufacturing would be a subcontractor with no pricing power.

Chang saw the opposite. Fab costs were rising faster than most companies could fund, brilliant designers were being excluded by capital requirements, and a manufacturer serving many customers would accumulate yield learning faster than any integrated firm with a narrower product mix.

The neutrality condition was the essential detail. A foundry that also sold its own chips would be a competitor holding customers’ secrets; one that sold nothing of its own could be trusted by everyone, and that trust was the actual product.

How was TSMC funded and launched?

With Taiwanese government capital through the National Development Fund, an investment and technology partnership with Philips, and private Taiwanese capital, established in 1987 with Chang as chairman and a workforce drawn substantially from ITRI.

The early years were difficult. Customers were small, the fabless industry barely existed, and the company was manufacturing at trailing process nodes for clients that larger foundry-capable firms did not want. Building credibility required consistent delivery over years.

The fabless industry then grew up around it — Nvidia, Broadcom, Qualcomm, Marvell and eventually Apple’s silicon team — all premised on the availability of manufacturing they did not have to own. Chang had created the supply that made an entire industry structure possible, as the TSMC story describes.

Why did he return from retirement?

Because the company needed direction during a difficult period. Chang stepped back from the chief executive role in 2005, returned in 2009 after the financial crisis and a period of strategic drift, and drove an aggressive capital expenditure programme precisely when peers were retrenching.

That decision proved decisive. Investing through the downturn positioned TSMC with capacity and process leadership when demand recovered, and it established the pattern of counter-cyclical investment that has defined the company since.

He finally retired in 2018, leaving a dual-leadership structure and a company whose strategic direction was thoroughly institutionalized. Unlike many founder-dependent firms, TSMC continued its trajectory without disruption — itself a measure of how well the succession was prepared.

What is his management philosophy?

Discipline, long horizons and honesty about competitive reality. Chang has spoken consistently about the importance of a defensible business model over operational cleverness, about pricing to reflect value rather than cost, and about the necessity of investing through cycles rather than with them.

He is also known for demanding intellectual rigour: strategy discussions grounded in specific analysis rather than in ambition, and a willingness to state uncomfortable conclusions plainly. His public commentary on the difficulty and cost of replicating Taiwanese semiconductor manufacturing elsewhere has been notably blunt.

The culture that resulted is demanding, engineering-centred and unusually stable, with long tenures and internal promotion. It is not a culture of individual heroics but of systematic execution — which is exactly what a yield-driven business requires.

What has he said about the global chip race?

That subsidized fabs elsewhere will be far more expensive to operate than their sponsors expect, and that the accumulated ecosystem, workforce discipline and supplier density of Taiwan cannot be reproduced with capital alone.

He has been direct about the cost differential of American manufacturing, about the difficulty of the Arizona project, and about the strategic reality that concentrating advanced capability in Taiwan creates both vulnerability and deterrence — the argument examined in the chip geopolitics story.

His position is not that diversification is impossible but that it is expensive, slow and unlikely to shift the leading edge away from Taiwan within any policy timeframe. Subsequent experience has largely supported this view.

What is his legacy?

An industry structure. The separation of design from manufacturing enabled thousands of chip companies that could never have afforded fabs, accelerated innovation across computing, and concentrated manufacturing in a company whose importance now shapes international relations.

For Taiwan, he provided the model that turned a manufacturing economy into an indispensable one. For the semiconductor industry, he provided the infrastructure on which its most valuable companies were built. Few individual business decisions have had comparable consequences.

The personal dimension is notable too: he began the work at fifty-four, in a country he had not lived in as an adult, building a business model that did not exist. It is the most consequential late-career founding in modern business history.

What did the yield lesson actually teach him?

That semiconductor manufacturing is a learning problem rather than an equipment problem. Early in his Texas Instruments career he was assigned a production line performing poorly and improved it dramatically through systematic experimentation with process variables, discovering that the same machines could produce very different results depending on how they were operated.

That insight scales into a competitive theory. If yield improves through accumulated learning, then the manufacturer processing the most diverse volume learns fastest, and the learning compounds into a cost advantage competitors cannot close by buying identical equipment. The entire foundry thesis rests on this observation.

It also explains the emphasis on customer diversity. A foundry serving dozens of designers across many product types encounters more failure modes and solves more problems than an integrated manufacturer producing a narrower range, which is why breadth of customer base is a technical advantage rather than merely a commercial one.

How did he think about pricing?

As a tool for shaping the cost curve rather than as a reflection of current costs. At Texas Instruments he was associated with pricing ahead of cost to win volume, on the understanding that volume drives learning, learning drives cost down, and the initially aggressive price becomes profitable as the curve descends.

Applied at TSMC the logic inverted as the company gained leadership. Once its process capability became scarce, pricing reflected value to the customer rather than cost to produce, and the company demonstrated willingness to raise prices through periods when manufacturers conventionally discount.

The consistent principle across both phases is that price should reflect strategic position rather than accounting cost, which is a more sophisticated view than most manufacturers hold and a substantial part of why the company’s margins are unusual for heavy industry.

What does his succession planning show?

Unusual deliberateness. He stepped back once, returned when the company needed direction, then retired completely in 2018 with a leadership structure already in place and a strategic direction thoroughly embedded in the organization’s planning processes.

The company’s continued execution after his departure is the strongest evidence that the strategy was institutional rather than personal. Capital expenditure discipline, customer neutrality, process roadmap cadence and counter-cyclical investment all continued without the founder present.

This contrasts sharply with founder-dependent organizations where strategic clarity departs with the founder. The difference lies in whether the founder built systems and taught reasoning, or made decisions personally and expected compliance.

What is his view on talent and work culture?

That semiconductor manufacturing demands a workforce willing to accept demanding shift patterns and a culture of exacting discipline, and that this cultural factor is a genuine competitive variable rather than an incidental detail. He has said so publicly and directly, including in the context of overseas fab difficulties.

The observation is uncomfortable and empirically supported. Fabs run continuously, problems occur at any hour, and process engineers are expected to respond regardless of schedule — expectations that vary considerably across labour markets and cannot simply be imposed by management preference.

How does he describe competitive strategy?

As choosing a position where the competition cannot follow rather than as outperforming rivals at the same activity. His account of founding TSMC emphasizes that direct competition with integrated device manufacturers was hopeless, and that the only viable entry was a business model those competitors could not adopt without abandoning their own products.

That framing is more disciplined than most strategic reasoning, which typically assumes the goal is to be better at what everyone does. Structural analysis asks instead what the incumbent is prevented from doing by its own commitments, and builds there.

Frequently Asked Questions