Pacific Money

From Foundry to Forum: Taiwan’s Repositioning in the New Semiconductor Order

Recent Features

Pacific Money | Economy | East Asia

From Foundry to Forum: Taiwan’s Repositioning in the New Semiconductor Order

The lesson of SEMICON 2025 is clear: no single country can control every layer of the semiconductor value chain.

From Foundry to Forum: Taiwan’s Repositioning in the New Semiconductor Order

The opening ceremony of SEMICON Taiwan 2025, Sep. 10, 2025.

Credit: Facebook/ SEMICON

The theme of this year’s annual semiconductor trade exhibition and conference, SEMICON Taiwan 2025 – “Lead with Collaboration, Innovating with the World” – reflected a subtle yet significant repositioning for Taiwan. No longer content to be regarded solely as the world’s contract fabrication powerhouse, Taiwan is recasting itself as an active architect of innovation platforms and collaborative forums. This shift acknowledges that semiconductors are more than an industrial output; they are the foundation of global technological competition and cooperation, shaping the balance of national security, economic resilience, and technological progress.

Continuity and Change in the Global Chip Industry

The semiconductor sector epitomizes globalization. Its supply chain is broadly distributed, yet heavily reliant on chokepoints controlled by a handful of states: the United States (EDA tools and intellectual property or IP), the Netherlands (lithography), Japan (materials and equipment), South Korea (memory), and Taiwan (foundry and packaging). This interdependence has long been the basis of efficiency and innovation, but it has also created vulnerabilities. 

Geopolitical tensions, pandemic disruptions, and the recognition of semiconductors as “industrial rice” – essential for commercial electronics, defense systems, and daily life – have pushed governments worldwide toward greater technological self-reliance. This new logic mirrors the idea of “sovereign AI”: each administration seeks to establish its own semiconductor ecosystem, building what might be called “mini Hsinchu Parks” in Arizona, Kumamoto, or Dresden.

However ambitious these projects may be, the reality is that no single country can control every layer of the semiconductor value chain. The true objective, therefore, is not complete decoupling from global supply networks, but the pursuit of minimum viable autonomy – the ability to secure access to critical components under conditions of geopolitical stress.

SEMICON 2025 reaffirmed an enduring truth: progress in semiconductors still depends on cross-border cooperation, even as competition intensifies. The emerging landscape is thus defined by a paradox. On one hand, there is broad structural continuity, as the globalized and interdependent architecture of the industry persists. On the other hand, there are dynamic new forces, as the actors (firms and states), the surges (new frameworks of collaboration), and the currents (policies, technological advances, and geopolitical pressures) become more fluid, contested, and complex.

Balancing Efficiency and Security

Reconciling the pursuit of efficiency with the imperative of security has become a defining dilemma for the semiconductor industry – one with no perfect answer. For decades, the sector thrived under a just-in-time model: lean supply chains, highly specialized global production, and minimal redundancy. Today, however, both firms and governments are moving toward just-in-case strategies, embedding redundancy into capacity, materials, and logistics to prepare for systemic shocks.

This shift exposes three tensions. First, overcapacity versus secure stocks. The buildout of new fabs in the United States, Japan, and Europe risks oversupply if the current AI-driven demand proves short-lived, forcing policymakers to balance the value of strategic buffer capacity against the costs of inefficient overinvestment. 

Second, outsourcing versus IP protection. Firms are under pressure to expand abroad to meet localization and security requirements but remain concerned about intellectual property leakage. This tension is leading most companies to keep advanced R&D and leading-edge processes at home while relocating mature-node production overseas. 

Third, security versus innovation. While full-scale localization may seem to reduce vulnerabilities, it often compromises efficiency, raises costs, and slows innovation, highlighting the trade-off between national security objectives and global competitiveness.

Shifting Focus: From Advanced Manufacturing to Packaging and New Materials

For decades, the industry’s competitive edge lay in transistor scaling – pushing nanometers smaller to sustain Moore’s Law. With physical and economic limits now constraining further shrinkage, the focus has shifted to advanced packaging and new materials as the next frontier of performance gains. 

3DIC and heterogeneous integration enable vertically stacked chips and multi-die systems, extending improvements beyond simple feature-size reduction. Silicon photonics integrates optical interconnects onto chips, easing data-transfer bottlenecks in high-performance computing (HPC) and AI servers. Compound semiconductors such as SiC and GaN offer superior efficiency and lower power consumption, making them vital for electric vehicles, power electronics, and next-generation communications.

This shift has also provided an industrial survival strategy for Taiwanese firms. Companies once dominant in LEDs, panels, or touch technologies – industries devastated by Chinese oversupply, the so-called “red waves” – are reinventing themselves by moving into semiconductor packaging and materials. By leveraging strengths in precision manufacturing and materials science, they are upgrading into higher-value segments of the ecosystem and reestablishing their relevance.

A new paradigm – Moore’s Law 2.0 – is taking shape, defined by system-level optimization rather than transistor scaling alone. As Hou Yung-ching, chairman of the Taiwan Semiconductor Industry Association (TSIA) and former senior deputy general manager at TSMC, emphasized, the issue is not the end of Moore’s Law but the escalating cost and slower pace of traditional scaling. The path forward lies in shifting from chip-level miniaturization to hardware-software co-optimization and integrated system design. 

Future performance gains will depend less on shrinking transistors and more on co-design across processors, memory, accelerators, and software. In this phase, competitiveness will be determined not by feature-size reduction, but by the ability to integrate, optimize, and orchestrate entire hardware–software ecosystems.

New Collaboration Models: Competition and Cooperation Coexisting

Four decades ago, TSMC transformed the industry by committing to the pure-play foundry model, separating design from manufacturing and enabling a new ecosystem. Today, however, as demand for computing power accelerates, new patterns of collaboration and rivalry are emerging.

At one end of the spectrum, industry giants are becoming even larger, pursuing both vertical and horizontal integration. Cloud service providers, for example, are designing their own ASICs (application-specific integrated circuits), while TSMC is expanding into advanced packaging to capture more system-level value. At the other end, small and medium-sized firms are carving out opportunities by specializing in niches such as materials, equipment, or design services – deepening expertise that sustains the wider ecosystem.

The result is a vibrant industry where competition and cooperation coexist more tightly and dynamically than ever before. Firms must compete fiercely for market share and technological leadership, while simultaneously collaborating across standards, supply chains, and innovation networks to safeguard interoperability and resilience. This dual dynamic will shape not only the industry’s structure, but also the pace, direction, and sustainability of future technological breakthroughs.

Invest, Invest, and Invest

Geopolitical pressures make semiconductor sovereignty an unavoidable policy priority. Yet the lesson of SEMICON 2025 is clear: the blind pursuit of full localization is a trap. The industry advances not through isolation, but through a careful balance of resilience and openness, competition and collaboration. 

In an uncertain environment, the most effective strategy is also the clearest: invest relentlessly in technology R&D. Firms cannot escape geopolitics, but they can hedge against it – and secure their relevance – by remaining indispensable at the technological frontier.