The semiconductor stack — what Chris Miller's *Chip War* taught policy makers
Chris Miller's 2022 history argued that semiconductor supply chains are not commodities but a layered industrial system — design, EDA tools, fabs, lithography, materials. Each layer has chokepoints, and the policy response since 2022 has been shaped by that map.
Key fact
ASML EUV machines shipped to date: ~250 units worldwide; none to mainland China since 2019 (SEMI / ASML annual reports).
Chris Miller's *Chip War: The Fight for the World's Most Critical Technology* (Scribner, 2022) reframed how policymakers in Washington, Brussels, Tokyo, and Seoul think about semiconductors. The book's core analytical move is to treat the industry not as a single supply chain but as a stack of distinct sub-industries — chip design, electronic design automation software, lithography equipment, specialty chemicals and gases, wafer fabrication, advanced packaging — each with its own competitive structure.
Miller's mapping showed something policy had not internalised. At the leading edge, every layer of the stack has at most three credible firms, and several layers have exactly one. ASML in EUV lithography. Cadence and Synopsys in EDA software. TSMC at the advanced-node foundry. Applied Materials, Lam Research, and Tokyo Electron in deposition and etch equipment. The geographic concentration is striking: most of these firms are in the United States, the Netherlands, Japan, Taiwan, and South Korea — a tight overlap with the post-1945 US security alliance system.
The 2022 US export controls drew directly on this map. Rather than ban all chip exports, the October 2022 rules cut specifically at the advanced-node layer — sub-14nm logic, certain memory thresholds — and at the equipment needed to produce those chips. The architecture of the controls follows the architecture of the stack.
Miller's book has been criticised for a US-centric reading. Dan Wang, writing in *Foreign Affairs* in 2023, argued the book underweights the speed at which Chinese firms have moved up the stack at trailing nodes — 28nm, 14nm — and the long-run risk that leading-edge concentration in Taiwan becomes a strategic vulnerability rather than a strategic asset for the US-led coalition.
What Miller and his critics agree on is that the stack is not a commodity market. Treating it as one — the analytical default in Western trade policy from 1995 to 2018 — was the mistake the post-2022 controls are trying to undo.
The book's analytical move is to treat the semiconductor industry as a vertically segmented stack — design, electronic design automation tools, intellectual-property cores, foundry process, equipment, materials — rather than as a single manufacturing industry. The strategic implication is that leverage points exist at every layer of the stack, that the leverage is asymmetrically distributed across countries, and that policy interventions targeting any single layer reverberate through the entire production graph in ways that are difficult to model without industry-specific knowledge.
The October 2022 US export control package, expanded in October 2023 and again in 2024, was the first major policy intervention to read the stack the way Miller's book frames it. The restrictions targeted advanced computing chips above defined performance thresholds, the lithography and deposition equipment required to make them at scale, and the US-person and US-technology employment at Chinese fabs that operationalises the transfer of process knowledge. The packaging of these three tracks together — chips, equipment, people — reflected the stack logic. The earlier export-control regime had targeted specific end products; the post-2022 regime targets the production capability itself.
The China-side response has been to invest aggressively in the domestic substitution of every layer of the stack that the restrictions affect. The Big Fund and its successor vehicles have channelled over $50 billion of public capital into domestic foundry, equipment, and materials capacity through 2024. SMIC's progress to 7-nm production using DUV lithography (without access to EUV machines) is the most-watched indicator of how rapidly the substitution is proceeding. The yields on the SMIC 7-nm process are estimated by industry observers to be substantially below the leading-edge norm, making the production economically marginal even as it is technically feasible. The yield-economics gap is what determines whether the substitution scales beyond demonstration runs.
The third-country dimension is the policy front that has developed since the initial 2022 package. The Netherlands and Japan, which together hold the upstream-equipment chokepoints for the leading-edge process, have aligned their export-control regimes with the US framework through 2023 and 2024 amendments. South Korea, which hosts both Samsung's leading-edge foundry operations and substantial existing equipment-supply relationships with Chinese fabs, has aligned more selectively. The aggregate result is that the equipment-supply pipeline into Chinese leading-edge fabs is narrower than at any time since the 1990s, and the narrowing is durable across multiple electoral cycles in the supplier countries.
The downstream demand side is the part most often missed in headline analysis. The 2024 demand surge for AI accelerators, concentrated in NVIDIA's H100 and successor product lines, produced a market in which TSMC's CoWoS advanced packaging capacity (rather than the wafer-fabrication capacity per se) became the binding constraint for the global AI build-out. CoWoS capacity expansion is on a 24-36 month build cycle, with TSMC adding Phase-Five and Phase-Six packaging fabs through 2026 and 2027. The leading-edge bottleneck has therefore moved down the stack from front-end wafer fabrication to back-end advanced packaging — exactly the kind of layer shift Miller's framework predicts when one segment becomes the binding constraint and capital flows to expand it. The next-binding constraint is the question that determines where the next round of strategic competition will concentrate.
The forward-looking implication of this analysis is that the structural drivers identified above will continue to shape policy trajectories across the second half of the 2020s. The doctrinal frameworks, institutional arrangements, and bilateral relationships described in the preceding sections are durable across multiple electoral cycles in the participating capitals, and any disruption of them would require shifts in underlying interests rather than rhetorical adjustment. The analytical reading developed here is not a prediction of a specific outcome at a specific date. It is a framework for reading the next round of developments — the summits, the policy announcements, the data releases, the bilateral and multilateral diplomatic moves — against the structural constraints the framework identifies. Each subsequent development can be read as confirming or refining the framework's predictions, and the cumulative pattern across multiple developments is what produces the analytical clarity that policy work most often needs. The headline-driven coverage of any specific event will continue to misread the broader trajectory; the data-driven, frame-anchored reading developed here is the antidote to that misreading and is the analytical discipline the policy community most needs across the remainder of the decade. The arithmetic of the underlying interests does not change quickly. The political and rhetorical surface above the arithmetic does change, sometimes quickly, and reading the two together is what produces analytical durability and policy-relevant insight that survives the news cycle.
The institutional research that underwrites this reading — the policy papers, the journal articles, the open-source datasets, and the running track records of the named scholars — represents a body of work substantially larger than any single explainer can summarise. Readers seeking deeper engagement should consult the primary sources cited in the preceding sections directly. The reading developed here aims to be a useful entry point rather than a substitute for that primary literature, and the framing has been chosen to surface the analytical moves that carry the most explanatory weight across the largest set of subsequent developments. A reader returning to this material in a year, in three years, or in five years should still find the framework usable, because the structural relationships it describes change more slowly than the headline developments they organise. The decade ahead will produce many specific events that this analysis cannot anticipate. The framework, if it is the right one, will help organise those events as they arrive.