Chapter 092　What Did TSMC Redefine?

Prescription for Incumbents What did TSMC redefine? To answer “a contract manufacturer that won the race to smaller geometries” is to mistake the level of the question. What this enterprise rewrote is not its own business. It is the division of labor of the semiconductor industry itself. It broke the assumption that design and manufacturing must be one. In this chapter, working only from public information, we establish on which dimension that redefinition occurred and where it could break. Praise is not analysis. What we could confirm and what we could not are handled separately.

1 What makes this enterprise worth a question

Start with the numbers. The numbers are not the conclusion. Revenue for the second quarter of 2026, ended June 30, 2026, was NT$1,270.3 billion, or US$40.20 billion. That is up 36.0 percent year on year. Net income was NT$706.5 billion, up 77.4 percent year on year. Gross margin was 67.7 percent, operating margin 60.3 percent, and net margin 55.6 percent. Diluted earnings per share were NT$27.25 (results release of July 16, 2026). Of wafer revenue in the quarter, 2-nanometer accounted for 3 percent, 3-nanometer for 30 percent, 5-nanometer for 33 percent, and 7-nanometer for 11 percent. Advanced technologies at 7-nanometer and below came to 77 percent in total. Third-quarter guidance was given as $44.6 billion to $45.8 billion (same release). These margins are not the numbers of what is generally called a contract manufacturing business. But numbers are outcomes. First Principle 2 states, Future Value Precedes Enterprise Value. Before enterprise value, there is future value. So what we should ask is when, and on which dimension, the redefinition that produced this outcome took place. There are three reasons this enterprise is worth a question. First, the business category did not exist. The company’s official profile records that at its founding in 1987 it created the business model of the dedicated IC foundry. It did not win in an existing market. It redrew the market’s dividing lines. Second, that division of labor became the structure of the whole industry. The company states that in 2025 it manufactured 12,682 products using 305 technologies for 534 customers. Annual capacity exceeds 17 million wafers on a 12-inch equivalent basis (both from the official company profile, verified August 1, 2026). This is less the order book of one firm than the operating record of the industry’s common base. The figure of 534 customers captures the character of the case well. This is not a subcontractor dependent on a few giant buyers. A wide span of the industry sits on the same manufacturing base. A base is a structure on which many depend at once. Third, that success produced concentration. For one company to sit at the center of a division of labor also means that if that company stops, the whole structure stops. The strength and the fragility come out of the same place.

2 Conventional answers and their limits

Three explanations of this company are in circulation. Each is partly right. Each drops the part that matters. The first conventional answer: “It is a subcontractor that makes things cheaply, in volume” On this account, scale and yield brought costs down, and price competitiveness pulled in orders. That scale is working is doubtless a fact. But the account contradicts the margins. Gross margin in the second quarter of 2026 was 67.7 percent. The definition of a subcontractor is that the buyer holds the power to set the price. A company that has lost pricing power does not keep margins like that. The first conventional answer mistakes where in the industry this company stands. The mistake comes from drawing the industry as a vertical picture. Buyer on top, supplier below. But in a division of labor, what matters is not top and bottom. It is center and periphery. Whoever many parties depend on stands at the center. This company sits downstream in the process, and at the center of the structure. The second conventional answer: “It won on the technology of miniaturization” On this account, the company won the development race in advanced process nodes, and so customers gathered. That a level of technology is a necessary condition is beyond doubt. The ramp of 2nanometer is officially cited as a factor lifting 2026 performance. But there are facts that technology alone cannot explain. The most advanced manufacturing technology was once held by vertically integrated firms that did design and manufacturing together. Other companies had the technology. What needs explaining is why design companies came to entrust their most important asset — their design data — to another company’s factory. That is not a question about technology. It is a question about trust. The third conventional answer: “It is a product of Taiwan’s industrial policy and location” Policy support, a concentration of talent, and the depth of the industrial cluster. That these contributed can be read from public information. But the conditions of location were given equally to other companies on the same island. Conditions explain only the part common to everyone who shares them. Further, this account cannot explain the company’s overseas expansion. The moves into the United States, Japan, and Germany are proceeding without reproducing those conditions of location. If location were the essence, that expansion could not work. What all three answers miss All three ask how the company makes things. But the second dimension of Enterprise Redefinition does not ask what a firm sells. It asks what value the firm delivers. Change the question and what can be seen changes. What this company delivered is not manufacturing capacity. It is the condition under which a design company can exist without owning a factory. Whether a company without a factory is viable depends on whether there is someone to whom manufacturing can be entrusted. This company designed itself to be that someone.

3 What was redefined — an analysis across the five

dimensions We read along the five dimensions of Enterprise Redefinition. The order follows the canon: Purpose, Business, Organization, Capital, and Leadership. What follows stays within what can be read from public information. It is not an assertion about internal decisions. Purpose — a design that does not compete with the customer The company has no branded end products of its own. This is not an absence of capability. It is the design of the business. The dedicated foundry as a business category rests on a single point: not becoming the customer’s competitor. If the party to whom a design company entrusts its designs sells products in the same market, the designs cannot be entrusted. Put the other way, the moment you decide not to sell products, a reason to entrust appears. This is the core of the case. The redefinition of Purpose and the redefinition of Business are bound together inseparably. The answer to “what do we exist for” is “we exist for the success of our customers.” To make that answer believable, the company chose the constraint of holding no products of its own. The constraint has become the asset. Core Purpose can remain stable while its expression and its means of realization evolve. This company’s core reads as making other companies’ designs real. Its customers moved from design houses serving personal computers to communications and then to AI, and the core did not move. Business — creating a category that did not exist, and creating the other side’s category too The rewriting in the business dimension is doubled. First, the company created its own category. A firm that only manufactured did not exist in 1987. Semiconductors then were something one company carried from design through manufacturing. This company drew a business boundary partway along that process. Second, it created the customer’s category. A design-only company without a factory is not viable without somewhere to entrust its work. Because this company created its category, the fabless category became possible. Its own redefinition produced the redefinition of the other side of the industry. That is the decisive difference from a mere change of business. The doubling shows up in the unit of the business as well. What the company sells is not the object called a wafer. It is the probability that another company’s design becomes real on schedule. So customers choose on certainty rather than on price. The difference in unit changes the nature of the competition. A company that sells objects competes on unit price and volume. A company that sells certainty competes on the continuity of its record. In the second, a single failure cancels the record that came before. Conversely, the longer the run without failure, the fewer the comparable alternatives. This company sits inside that asymmetric competition. Organization — a value-creation system that runs past the boundary of the firm In the organizational dimension, the way the boundary is drawn is distinctive. The canon defines an organization not as a collection of people but as a value-creation system made of people, AI, partners, universities, and customers. In this company’s case, the suppliers of design tools, design IP, equipment, and materials, together with the customer’s design teams, all have to move on the same process design. Every time the process generation changes, that whole set of relationships is updated. In overseas expansion, too, the boundary is drawn across capital relationships. JASM, operating in Kumamoto, Japan, is the example. At the time of the February 6, 2024 announcement, the shareholdings were given as TSMC about 86.5 percent, Sony Semiconductor Solutions about 6.0 percent, Denso about 5.5 percent, and Toyota Motor about 2.0 percent. It is a structure in which customers take part in the factory as shareholders. Capital — a trust term that moves financial capital The capital dimension carries the most weight in this case. The third equation of the canon defines capital as follows. Future Capital = Financial × Human × Learning × Trust × AI × Knowledge × Ecosystem × Purpose This is multiplication. If any single term is zero, the whole product is zero. Semiconductor manufacturing is among the most capital-intensive industries in the world. The company has been reported to have raised its 2026 capital expenditure plan to $60 billion to $64 billion. That comes from reporting on the July 16, 2026 earnings call, and this chapter treats it as reporting. Investment on that scale cannot be justified by financial capital alone. There have to be customers who have already decided to use a fab before it is finished. What makes that advance commitment possible is the Trust term. First Principle 8 states, Trust Compounds Faster Than Capital. Trust grows faster than capital. In this company’s case, trust stands first and financial capital moves after. The order must not be read in reverse. Leadership — taking on the industry’s timetable In the leadership dimension, the distinctive feature is capability built ahead of demand. The fourth equation of the canon defines leadership as follows. Leadership = Purpose × Question Design × Capital Allocation × System Architecture × Trust What stands out in this company’s leadership are the Capital Allocation and Trust terms. Prepare production capacity before demand is fixed, and make people believe the preparation will hold. As First Principle 9 states, Leadership Means Designing the Future. But this style of management is vulnerable to misreading. Prepare too much capacity and fixed costs remain. Prepare too little and customers’ plans collapse. CEO C.C. Wei was reported to have said on the July 16, 2026 earnings call that demand will stay strong from 2029 through 2030. He added that he does not know whether there will be a dip along the way. It was a statement that avoided assertion.

4 Structure — maturity and the value chain

4.1 Where it sits on the Enterprise Redefinition Maturity Model

Within what can be read from public information, this company’s maturity varies by dimension. Nothing can be asserted. In the business dimension, evolution of the business model can be observed. Creating the dedicated foundry category, moving the center of gravity to advanced processes, and extending into advanced packaging. This reads as behavior close to a Continuous Redefinition Enterprise, because capability is placed ahead of the point at which external disruption demands it. The capital dimension reads the same way. Against the assessment question of whether resources are allocated toward Future Value rather than historical success, allocating tens of billions of dollars to demand that is not yet fixed is one answer. The Purpose dimension, by contrast, is hard to assess. The source paper’s question is whether the organization periodically re-examines its Core Purpose and adapts its expression without unnecessarily weakening organizational identity. This company’s Core Purpose looks to have barely moved since its founding. Whether to read that as consistency or as insufficient re-examination cannot be determined from outside. The organizational and leadership dimensions cannot be observed sufficiently from outside either. Here we recall the canon’s notes. The Enterprise Redefinition Maturity Model (ERMM) evaluates organizational coherence rather than isolated excellence. An organization can be at Level 4 in business while the other dimensions do not reach it. Placing this company at a single level is itself a misuse of the model. Maturity is also assessed across all five dimensions in balance: exceptional technological capability with weak leadership redesign does not produce higher maturity, and strong purpose without adaptive organizational systems remains insufficient. And Level 5 must not be treated as a target to be reached as fast as possible. The appropriate level differs by industry and environment.

4.2 Where in the Future Value Chain the value was created

The causal order fixed by the canon is as follows. Purpose → Learning → Redefinition → Creation → Enterprise Value Lay this company’s path over that order. Value did not arise at Creation. It arose between Purpose and Redefinition. The Purpose of not competing with the customer was set first. Because it was set, the redefinition of the industry’s division of labor became possible. Because that redefinition happened, other companies’ designs gathered, scale emerged, learning accelerated, and Creation became possible. Enterprise Value came last. The margins of 2026 are the result of an order held for close to forty years. The second equation confirms the same structure. FVCC = Purpose × Learning × Redefinition × AI Integration × Ecosystem × Capital Allocation × Trust Seven terms, multiplied. In this company’s case, Purpose, Trust, and Ecosystem held high values over a long period. Had any one of them been zero, the whole would have been zero however large the other six. In particular, had Purpose been zero — had the company reached for products under its own brand — Trust would have been lost at the same moment. The two terms are structurally linked.

4.3 Time and trust

The first equation defines value as follows. Value = Purpose × Trust × Capability × Time Trust and Time stand together in the same expression. That pairing is the point of this case. Trust is not produced by declaration. It can only be proved by the length of time over which a promise is not broken. The fifth equation says the same thing. Future Value = Future Time × Future Capability Capability can be bought. Time cannot. A record of never once becoming a customer’s competitor across close to forty years cannot be replicated with money. That is the substance of the barrier to entry.

5 What it looks like in practice — the pivots, and the

conditions under which they break Redefinition is not the act of adding. It is the act of deciding what to protect and what to release. We set out the pivots together with what was released. Pivot 1 — 1987, creating the category The company drew a business boundary that separated design from manufacturing. What it released was the option of having branded products of its own. It gave up the margin of the end product and chose the position of not being the customer’s competitor. That judgment closes off a short-term revenue opportunity by choice. Decisions that create Future Value usually do. Pivot 2 — moving the center of gravity to advanced processes The more advanced designs gather, the more the development cost of the next generation can be recovered. The more it can be recovered, the sooner the next generation can be brought out. That 7nanometer and below made up 77 percent of revenue in the second quarter of 2026 is where the loop has arrived. What was released was diversification across technology generations. The heavier the bet on the leading edge, the larger the impact of missing. Pivot 3 — the turn to geographic dispersion On July 16, 2026, the company announced an additional $100 billion investment in Arizona in the United States. Cumulative investment in the United States comes to $265 billion. The additional tranche plans four manufacturing sites, two advanced packaging facilities, and one research and development center (announcements by the Arizona Commerce Authority and the U.S. Department of Commerce). In Japan, JASM in Kumamoto is operating. The February 6, 2024 announcement set a plan to bring a second fab into operation by the end of 2027. Combined investment across both fabs was given as more than $20 billion, and capacity as more than 100,000 wafers per month on a 12-inch equivalent basis. The processes are 40, 22/28, 12/16, and 6/7 nanometer. In Europe, the official company profile records that a specialty technology fab has been under construction in Dresden, Germany, since 2024. We could not confirm primary information on the shareholding structure or the investment amount as of the time of writing. So we state no figures. What was released was the efficiency of concentrating in one place. Dispersion raises costs. That it is proceeding anyway reads as a response to concentration itself having become a risk for customers. We keep this to a statement of fact. We do not enter into judgments about policy or international relations. Pivot 4 — July 2026, the Kumamoto earthquake On July 29, 2026, a magnitude 7.1 earthquake occurred in Kumamoto Prefecture. The company stated that its employees’ safety had been confirmed and that there was no damage to the buildings of the Kumamoto fab. Operations were reported to be resuming in sequence after inspection, and construction of the second fab to have been temporarily suspended as a precaution against aftershocks. The event shows concretely what dispersion means. Geographic dispersion is also preparation against natural disaster. At the same time, it shows that the places you disperse to carry natural conditions of their own. The conditions under which this advantage breaks On that basis, we set out four fragilities. Praise is the enemy of analysis. First, geographic concentration. In the official list of manufacturing sites, most of the 12-inch fabs are located in Taiwan. Overseas sites are spreading across the United States, Japan, and China, but the center of the leading edge has not moved. Natural disaster and international conditions both bear on the same single point. Second, skew in demand. High-performance computing was reported to account for about two-thirds of revenue in the second quarter of 2026. If capital spending in that area changes, the effect is direct. A structure leaning to the leading edge moves with leading-edge demand. A large number of customers does not cancel the skew. Even with many customers, there is no dispersion if those customers are all looking at the same end demand. Third, the cost of dispersion. Holding equivalent quality and yield across several countries is not simple replication. For the effect on the cost structure, public information leaves nothing to do but wait for the record. We assert nothing here. Fourth, the operation of neutrality. The promise of not competing with the customer takes more than declining to sell products. When capacity falls short of demand, who gets how much becomes, in effect, a selection. A promise is kept not by its wording but by the operation of allocation. This point is hard to verify from outside. For a company whose greatest asset is trust, the greatest risk is also trust.

6 What transfers, and questions for the executive

What transfers from this case, and under what conditions? Not a discussion of attracting factories. There are four things to take. First, the design of what you will not do. This company’s strength derives not from a list of what it can do but from what it decided not to do. The constraint of holding no products of its own produced its customers’ trust. Most companies write Purpose as what they aim at. But Purpose only functions once it states what will be given up. Second, the idea of creating the other side’s category. This company did not only redefine itself. It made a category — the company without a factory — viable on the customer’s side. That idea does not arise from thinking only about your own growth. The question to ask is whose business becomes newly possible because you change. Third, accounting for trust as capital. In the capital allocation meetings of most incumbents, plant and acquisitions reach the agenda. Trust does not. Yet in this case what moved financial capital was trust. First Principle 3 states, Capital Exists to Create Possibility. Capital exists to create possibility. The second point has a concrete form for any incumbent. Many hold high technology in the layers of materials, equipment, and components. Will that technology be used closed inside their own products, or opened as the condition on which other companies can start new businesses? The first is a transaction. The second is the design of a structure. The take is larger in the first. The life is longer in the second. Fourth, estimating the cost of concentration in advance. This company’s concentration arose as a result of success. Many companies call concentration a strategy and learn its cost afterward. The further success proceeds, the more the structural risk that success creates must be inspected. Sony Semiconductor Solutions, Denso, and Toyota Motor hold stakes in the Kumamoto joint venture. Customers of this company are therefore not bystanders to this case. But a factory arriving is not redefinition. Redefinition is redrawing the boundaries of your own business. Finally, three questions. Each can be answered at your next executive meeting. Question 1 — Have you become your customer’s competitor? If you have, your customers will never entrust you with the work that really matters. Your take rises and what is entrusted to you falls. Are you making that trade knowingly? Question 2 — What promise has your company kept for forty years? If there is nothing you have kept, the capital called trust has not accumulated. Trust is a function of time. Without a mechanism that enforces it as an institution, it disappears when the executive changes. Question 3 — On which single point is your greatest strength concentrated? What happens when that point stops? If you cannot answer, it is not a strength. It is a fragility that has not yet surfaced. This is the question of Recognize, the first stage of Enterprise Redefinition. What assumptions about our enterprise are becoming obsolete? AI can test assumptions. Which assumptions should be doubted is decided by people. What TSMC redefined was neither semiconductors nor manufacturing technology. It was the division of labor of the industry itself. It broke the assumption that design and manufacturing must be one. The reason it could break it was not technology. It was the decision not to compete with its customers. Because it placed the constraint first, it became the party that gets entrusted. What we are looking at is not the shape of success but evidence of order. There was Purpose, then Learning, then Redefinition, then Creation, and Enterprise Value came last. And what carried that order was time. A promise is worth only the number of years over which it was not broken. Securing those years as an institution is the part hardest to imitate.

In brief

What TSMC redefined is not semiconductors. It is the industry’s division of labor, and the assumption that design and manufacturing are one.
What can be read from public information is behavior close to a Continuous Redefinition Enterprise in the business and capital dimensions.
Value was not created at Creation. It was created between Purpose and Redefinition.
When the promise of not competing with customers is broken by the operation of capacity allocation, the advantage breaks.

Key concepts

Enterprise Redefinition / the Enterprise Redefinition Maturity Model / Future Value Chain / Future Capital / the Layer Shift Pattern (→ Vol. VI, Ch. 059)

The chain of ideas

The constraint on Core Purpose → Trust → Future Capital → Creation → Enterprise Value

Related first principles

Principle 1 — Purpose Precedes Profit. Principle 2 — Future Value Precedes Enterprise Value. Principle 6 — Enterprise Exists to Redefine Itself. Principle 8 — Trust Compounds Faster Than Capital.

Related chapters

Vol. V, Ch. 046 “What Does It Mean to Redefine a Business Model?” — the account of the pattern that creates a business category itself
Vol. VI, Ch. 051 “What Does It Mean to Redefine Competitive Advantage?” — why an advantage that cannot be imitated arises from time
Vol. VII, Ch. 068 “Does Trust Become Enterprise Value?” — the mechanism by which trust moves financial capital
Vol. IX, Ch. 081 “What Did NVIDIA Redefine?” — a redefinition that happened in a different layer of the same industry

Papers and companion volumes

Kadowaki, N. (2026). Future Value Theory: A Management Framework for Enterprise, Capital, and Society in the Age of AI. VURA Working stract=7120980 / Paper. Zenodo: SSRN: https://ssrn.com/abhttps://doi.org/10.5281/zenodo. 21255662
Kadowaki, N. (2026). Enterprise Redefinition: Toward an Enterprise Evolution Theory for the Age of AI. VURA Working Paper. (Published on Zenodo; under review at SSRN)
100 Questions on Management in the Age of AI, #034 “What Kind of Company Cannot Be Copied in the Age of AI?”

Chapter 092 What Did TSMC Redefine?