Public incentives and private commitments worldwide have mobilized more than $600 billion toward domestic chip factories since 2022, the largest industrial policy push since World War II[s]. The US CHIPS Act allocated about $52 billion in direct incentives and R&D. The EU Chips Act mobilizes roughly 43 billion euros in combined public and private investment. South Korea’s K-Semiconductor Belt aims to attract more than $450 billion in private investment by 2030, backed by government tax credits, loans, and infrastructure support. The goal: secure the semiconductor supply chain by bringing chip manufacturing home.
The results so far suggest that goal is failing.
The Semiconductor Supply Chain Remains Vulnerable
Despite new factories rising across the United States, the country still depends on Asia for a critical step: turning processed silicon wafers into packaged chips[s]. More than 90% of chip assembly and testing happens in East Asia[s]. More than 60% of chip packaging is done in China specifically. Even if American fabs produce the chips, they still travel overseas to become usable products.
This gap exists because policymakers focused on the most visible part of manufacturing: the fabrication plants, or fabs. The less glamorous back-end work received far less attention and investment.
Projects Running Years Behind Schedule
Intel’s Ohio fab, announced in 2022 with a target opening of 2025, will now begin production between 2030 and 2031[s]. That is a five to six year delay. The $28 billion project continues at a slower pace while Intel waits for customer demand to justify acceleration.
Across all CHIPS Act projects, companies had submitted milestone completion reports for only 24 of 161 milestones as of July 2025[s]. The Commerce Department had reviewed and verified 18 of 35 disbursement requests, releasing $6 billion of the $30.9 billion in direct funding awarded.
Meanwhile, TSMC’s Arizona fab is operating with about 3,000 employees[s], but its founder Morris Chang called the US expansion “a very expensive exercise”[s]. Operating costs run 30% higher than equivalent Taiwan facilities. Efficiency is being sacrificed for security that may not materialize.
The Labor Gap
Even if fabs open on schedule, who will run them? The US semiconductor workforce needs to grow by 115,000 jobs by 2030, but roughly 67,000 of those positions risk going unfilled[s]. Technician and engineering roles face the steepest shortages. You cannot secure a semiconductor supply chain without people trained to operate it.
Global Interdependence By Design
The semiconductor supply chain crosses more than 70 international borders before a finished chip reaches consumers[s]. Three critical bottlenecks make complete independence impossible[s]:
- TSMC produces 92% of the world’s most advanced chips, those below 7 nanometers
- ASML in the Netherlands builds the only machines capable of manufacturing advanced processors, each costing $200 million and taking 18 months to build
- China controls 60% of rare earth production essential for chip manufacturing
More than 90% of raw silicon wafers come from Japan, Taiwan, Singapore, and South Korea[s]. Concentration exists at every layer.
The Cost of Trying
McKinsey projects that supply chain regionalization will increase global semiconductor costs by 15 to 25% over the next decade[s]. That is the price of reducing single points of failure. Whether nations can sustain that premium while building supplier ecosystems that took Asia decades to develop remains an open question.
China’s own push for self-sufficiency illustrates the difficulty. Despite spending hundreds of billions, China’s overall semiconductor self-sufficiency remains around 30% in 2024 by most estimates, while Beijing has reportedly imposed an unofficial rule requiring chipmakers to use at least 50% domestically made equipment when adding capacity[s]. An 80% self-sufficiency goal by 2030, drawn up by Chinese semiconductor industry executives, looks increasingly unrealistic.
The semiconductor nationalism paradox is this: billions can buy fabs, but they cannot quickly replicate an interdependent global ecosystem. The supply chain vulnerabilities that sparked the spending spree persist.
The global semiconductor supply chain has attracted more than $600 billion in combined government incentives and private investment commitments since 2022[s]. This represents the largest coordinated industrial policy push since World War II, spanning the US CHIPS Act ($52 billion in direct incentives and R&D), the EU Chips Act (roughly 43 billion euros in mixed public-private mobilization), South Korea’s 2021 K-Semiconductor Belt initiative (more than $450 billion in targeted private investment by 2030, backed by government tax credits, loans, and infrastructure support), and parallel programs in Japan, India, and elsewhere. The strategic objective: reduce dependency on concentrated manufacturing nodes, particularly Taiwan.
The execution has exposed structural constraints that capital alone cannot overcome.
The Back-End Gap in the Semiconductor Supply Chain
CHIPS Act investments prioritized front-end fabrication, the wafer processing stage. The less capital-intensive but equally critical back-end processes, including wafer dicing, die attach, wire bonding, packaging, and testing, remain concentrated in Asia[s]. More than 90% of outsourced assembly and test (OSAT) occurs in East Asia, with more than 60% of packaging volume specifically in China[s].
This creates a chokepoint that domestic fab construction does not address. Chips fabricated in Arizona still route through Asian OSAT facilities before reaching end customers. The tools used in OSAT, manufactured by firms like Kulicke and Soffa, BESI, and ASM Pacific, are themselves produced in China, Singapore, Malaysia, and Vietnam[s].
Project Execution Failures
Intel’s Ohio One campus, announced in 2022 with a $28 billion budget and 2025 production target, has slipped to 2030 or 2031[s]. The company is pacing construction to match foundry customer demand rather than government timelines. Naga Chandrasekaran, Intel’s chief global operations officer, cited the need to “manage our capital responsibly and adapt to the needs of our customers.”
The CHIPS Act portfolio shows systemic execution lag. Of 161 milestones across 40 funded projects, companies had submitted milestone completion reports for only 24 as of July 2025[s]. The Commerce Department had reviewed and verified 18 of 35 disbursement requests, releasing $6 billion against $30.9 billion in awarded direct funding; one project, a leading-edge logic facility in Arizona, was certified complete in June 2025. Project completion timelines extend through October 2033, meaning program success cannot be measured for nearly a decade.
Europe’s approach has fared worse. Intel’s planned Magdeburg megafab collapsed entirely, exposing what one policy paper called “a fundamental flaw in Europe’s approach to chip sovereignty: betting on a handful of giant, foreign-owned factories is not a strategy; it is a vulnerability”[s].
Semiconductor Supply Chain Workforce Constraints
The US semiconductor workforce must expand by 115,000 positions by 2030 to staff planned fabs. Approximately 67,000 of those roles risk going unfilled, concentrated in technician positions (39% of the gap) and engineering roles (41%)[s]. Fabs cannot operate without trained personnel regardless of capital availability.
TSMC Arizona, the most advanced domestic fab currently operational, employs approximately 3,000 workers[s] and projects needing 6,000 across its three planned facilities. The company cited skilled-worker shortages as a factor in previous construction delays.
Structural Interdependencies
The semiconductor supply chain operates through three bottlenecks that resist diversification[s]:
- Advanced manufacturing: TSMC handles 92% of chips below 7nm, including processors for iPhones, data centers, and military systems. TSMC’s foundry market share exceeds 60% overall.
- EUV lithography: ASML maintains a global monopoly on extreme ultraviolet lithography equipment. Each tool costs $200 million, takes 18 months to build, and requires components from over 5,000 suppliers. Only 200 exist worldwide.
- Raw materials: China controls 60% of rare earth production. More than 90% of silicon wafers come from Japan, Taiwan, Singapore, and South Korea[s].
A typical integrated circuit crosses more than 70 international borders before reaching consumers[s]. US firms control more than 40% of IC design market share and more than 50% of core IP[s], but manufacturing remains concentrated in Asia.
Economic Costs of Regionalization
McKinsey estimates that semiconductor supply chain regionalization will increase global costs by 15 to 25% over the next decade[s]. This reflects the efficiency losses from operating smaller, geographically dispersed fabs rather than consolidated giga-fabs optimized for throughput.
TSMC’s Arizona facilities demonstrate the premium. TSMC founder Morris Chang characterized the US expansion as “a very expensive exercise,” with operating costs running 30% higher than equivalent Taiwan capacity[s]. The company committed $165 billion to US manufacturing after renegotiation with the Trump administration[s], but geographic diversification does not eliminate dependency on Taiwan-based supply chain links.
China’s Parallel Struggle
China’s overall semiconductor self-sufficiency remains around 30%–33% in 2024, depending on methodology; late-2025 reporting describes a separate unofficial rule requiring chipmakers to use at least 50% domestically made equipment when adding capacity[s]. A reported 80% self-sufficiency target by 2030 was drawn up by Chinese semiconductor industry executives rather than confirmed as an official Five-Year Plan target. SMIC produces 7nm chips for Huawei and has entered 5nm pilot production[s], but yield improvement remains the binding constraint. Analysis suggests China remains a decade behind frontier manufacturing despite hundreds of billions in subsidies.
China’s Ministry of Commerce warned that US chip export legislation would “severely disrupt the international economic and trade order”[s]. The warning highlights that fragmentation imposes costs on all parties, not just targeted nations.
The Paradox
Semiconductor nationalism assumes that capital expenditure can replicate ecosystems that developed over decades through market-driven optimization. The evidence suggests otherwise. Building leading-edge fabs requires three to five years minimum. Training a semiconductor workforce takes longer. Developing supplier networks and achieving competitive yields takes longer still.
The supply chain chokepoints that prompted this industrial policy remain largely intact. Governments have demonstrated they can write large checks. They have not yet demonstrated they can build resilient, cost-competitive domestic semiconductor supply chains within policy-relevant timeframes.
