On the morning of May 28, 2026, Shen Yi published a systematic analysis of TSMC's gross margins. His core judgment was unambiguous: TSMC's 60% gross margin is not a number to envy — it is a narrative that needs to be taken apart. Behind that figure lies a deep structural inequality between the manufacturing and the design stages of the chip value chain.
Whether TSMC's gross margin is high depends entirely on which reference frame you use to measure it.
Within contract manufacturing, a 60% gross margin is indeed enviable — most manufacturers are content to sustain 20%. But the point is this: TSMC's competitors were never other foundries. Apple's design margin is the real high-profit benchmark — iPhone hardware gross margin has held above 60% for years, and Apple is essentially a "software company" (as Luo Yonghao once put it), with a composite margin far above TSMC's.
The United States carved up the chip industrial chain and left a small slice for TSMC. TSMC turned that small slice into a 60% gross margin, and many people look at the number and think "getting to where TSMC is would be quite good." But that mindset, in the words of the Koreans, is "being content to be America's dog."
The metaphor exposes a structural fact: TSMC's share of global chip profits is not "60% of the big slice," but "60% of the small slice." Apple, NVIDIA, Qualcomm and other upstream design firms take the lion's share of the profit; what TSMC receives is the payment for "dirty work and hard labor" — though it does that work exceedingly well.
The "Technological Republic" Blueprint — America's Upstream-Control Strategy
Shen Yi drew on The Technological Republic: Hard Power, Soft Belief, and the Future of the West to trace the historical origin of this structure: since the 1990s the U.S. technology-development strategy has been to keep the upstream (software, design, IP) — the "high value-added" core competencies — at home, while sloughing off manufacturing — "the dirty, exhausting work that demands endless tens-of-billions in equipment investment and depreciation risk" — onto Asia. The model nearly held for half a century.
Fifteen years ago, Shen Yi met a Taiwanese businessman in Fujian who said bluntly: "TSMC gets its rice bowl from the Americans." Five years later, running into one of them again, the man added: when America is rich it is willing to hand the bowl over; when its own pockets tighten, it wants to take it back. But America is not what it was — it cannot reintegrate the asset-heavy chip-manufacturing base back onto its own soil, and so it has no choice but to turn a blind eye.
The result of this division of labor is twofold: for the United States, controlling chip design realizes a value capture over the entire semiconductor industry; for TSMC, even though it grew into the world's most advanced foundry, it has remained in the position of "the one being fed" within the value chain.
What China Needs Is Not TSMC, But a Complete Ecosystem
Shen Yi's core conclusion: what China needs is not a "TSMC-style success" in manufacturing, but a complete industrial chain running from design to fabrication to packaging and testing. "What China wants is not just TSMC, but the whole set of circuits that includes America's Apple, Intel and NVIDIA."
That judgment is, in logic, upstream–downstream complementary to the Huawei "Tau's Law" phenomenon that this archive has recorded earlier: Tau's Law resolves the technical-route question of "manufacturing without dependence on advanced process nodes" (shifting from geometric to temporal scaling), while Shen Yi's analysis reveals the industrial political-economy question of "even with manufacturing capability, how to keep the profit at home." Both lines of evidence point to the same strategic target — China's semiconductor industry must close the full "manufacturing + design" loop independently; otherwise it will fall into the same TSMC-style structural bind of "strong manufacturing but weak position in the value chain."
Shen Yi also offers a politically significant footnote: TSMC can never become a "hostage card" in the Taiwan question. Whether TSMC exists or not will not block the cause of reunification. "We will not only have it, we will do it better — and across the complete industrial chain, top to bottom."
The Paradox of Sanctions — Accelerating End-to-End Self-Reliance
The actual effect of U.S. chip sanctions on China — as Shen Yi has analyzed repeatedly elsewhere — has been precisely to play the unintended role of "accelerating China's end-to-end self-reliance across the chip chain." When NVIDIA CEO Jensen Huang admitted in May 2026 that China's "competitive advantage" in the AI-chip market had been "eroded," and ASML's CEO used a "vegetable garden in the desert" metaphor to describe China's autonomous acceleration, both lines pointed to the same conclusion: the institutional premise of the "upstream control + Asian manufacturing" model the United States built from the 1990s is being broken by America's own sanctions.
The strategic target of China's semiconductor industry is not to become "the next TSMC," but to become "a complete ecosystem answerable to no one." That is both the necessary extension of industrial logic and the bedrock requirement of geopolitical competition.
For the upstream-downstream complement to this analysis — Huawei's "Tau's Law" and the technical-route question of manufacturing without dependence on advanced process nodes — see the related-reading list below.