As the global artificial intelligence computing revolution accelerates, the semiconductor industry has become central to national resilience and technological competitiveness. Taiwan’s leadership is not solely the result of manufacturing strength, but also of a highly integrated value chain that depends on stable energy supply, access to top-tier talent, and forward-looking policy support. As the industry advances toward leading-edge nodes and large-scale AI deployment, structural constraints are becoming more pronounced and require coordinated public-private action.

The Committee identifies four priorities to sustain Taiwan’s competitive position. First, energy security going forward has become a critical factor. Rapid growth in AI-driven electricity demand requires diversification of energy sources and accelerated development of supporting infrastructure, with carbon-free energy (CFE) treated as a strategic resource to enhance resilience and supply stability. Second, intensifying global competition for talent requires a more competitive policy response. Taiwan should strengthen tax incentives and align with international standards to attract and retain highly-skilled professionals. Third, the Industrial Innovation Act should be refined to ensure that R&D tax incentives remain accessible to high-growth companies, given that current eligibility thresholds under implementation regulations may limit access as firms scale. Fourth, as AI applications expand beyond centralized data centers, Taiwan should promote a distributed cloud-to-edge AI development model, integrating edge computing into national programs to reduce pressure on power infrastructure while supporting data security and system resilience.

Addressing these structural constraints will enable Taiwan to reinforce its role not only as a manufacturing leader but also as a critical platform in the global digital economy.

Suggestion 1: Ensure resilient and predictable electricity supply.

The semiconductor industry and its value chain remain key to the global AI computing ecosystem and Taiwan’s economic security. Sustaining this leadership depends not only on technological capability and talent, but also on a stable, cost-competitive, and resilient energy supply. Electricity demand driven by AI is projected to increase significantly by 2028, with growth rates expected to reach historical highs. This places increasing pressure on Taiwan’s power system and elevates energy security as a key factor for continued industry development.

The Committee recognizes the government’s efforts to enhance power system resilience. However, evolving global energy dynamics and geopolitical risks require a more forward-looking and coordinated policy approach. We recommend that the government establish a long-term energy strategy that strengthens supply resilience and provides greater predictability in electricity pricing, with priority given to the following areas:

1.1   Ensure a stable energy supply despite geopolitical risks. Taiwan’s power mix is increasingly dependent on imported liquefied natural gas (LNG), which has surpassed coal as the largest source of power generation. This growing reliance, combined with near-total import dependence, heightens exposure to supply disruptions, geopolitical stress, and price volatility. While existing regulations under the Natural Gas Industry Act require minimum stockholding levels for natural gas supply, these requirements are primarily defined through administrative measures and provide limited buffer capacity. By comparison, major LNG-importing economies such as Japan and South Korea maintain significantly higher storage capacity and diversified receiving infrastructure, enabling greater supply flexibility and resilience in the face of disruption

To strengthen energy resilience and support stable industrial operations, the Committee urges the government to:

a.)    Accelerate LNG infrastructure development. Capacity constraints at receiving terminals limit Taiwan’s ability to expand imports and build reserves. It is necessary to expedite the construction of LNG receiving terminals and related infrastructure, recognizing their strategic importance to energy security and supply continuity.

b.)    Strengthen legal requirements for LNG reserves. Drawing on the statutory framework under the Petroleum Administration Act, the government should amend Article 31 of the Natural Gas Industry Act to establish clear, enforceable, and statutory minimum LNG stockholding requirements, rather than leaving key parameters to be determined by subordinate regulations.

c.)    Enhance predictability in electricity pricing adjustments. Electricity price volatility creates planning challenges for industrial users. We recommend that the government provide advance notice of rate changes to allow sufficient time for operational and budget planning.

1.2   Build resilient and flexible energy policies. Renewable energy plays an increasingly important role in Taiwan’s energy system, not only in supporting decarbonization but also in strengthening energy resilience. As domestically generated sources, renewables can reduce reliance on imported fuels and improve supply stability under disruption scenarios.

The Committee recommends:

a.)    Reframe renewable energy policies with an eye to energy resilience. Renewable energy sources such as offshore wind and solar provide domestically generated power that is not dependent on imported fuels. In supply disruption scenarios, this characteristic enhances their role in sustaining critical economic activity. The government should more explicitly integrate energy security considerations into renewable energy targets and planning, alongside its carbon reduction objectives.

b.)    Strengthen cross-ministerial coordination and implementation planning. Energy transition policy requires alignment across grid development, land use, environmental review, and national security considerations. The government should establish a cross-ministerial coordination mechanism at the Executive Yuan level to define clear targets, timelines, and implementation pathways.

c.)    Consider market conditions and existing system constraints when setting renewable energy requirements. Current renewable energy development has not fully met demand, and green electricity remains costly for many enterprises. Regulatory requirements, including environmental review conditions and renewable energy procurement obligations, should be calibrated based on domestic supply availability, industry characteristics, and operational feasibility to avoid imposing disproportionate burdens on industry.

Suggestion 2: Enhance tax incentives to strengthen international talent recruitment and retention in the semiconductor industry.

Taiwan has made progress in attracting foreign professionals through initiatives such as the Employment Gold Card. However, current tax incentives under the Act for the Recruitment and Employment of Foreign Professionals remain limited in duration and competitiveness, constrain-ing long-term retention of high-skilled talent.

2.1   Extend and stabilize tax incentives for foreign professionals. Current tax benefits are limited to five years and may discourage long-term career planning in Taiwan. To improve retention and support workforce stability, the government should consider extending the incentive period, drawing on international models such as Italy’s “5+5” framework, which allows an initial five-year tax benefit to be extended for an additional five years for eligible individuals who meet retention-related conditions (e.g., continued employment or family/relocation criteria).

2.2   Enhance the international competitiveness of tax treatment. After the incentive period ends, foreign professionals are subject to standard tax rates, which may reduce Taiwan’s attractiveness relative to other jurisdictions, such as Italy, the Netherlands, and Spain, which offer preferential tax regimes including income tax reductions, partial exemptions, or flat rates for qualifying foreign workers. The government should review and benchmark Taiwan’s tax regime against international practices to ensure competitive positioning in attracting global talent.

2.3   Expand the flexibility of talent-related tax mechanisms. Existing provisions under Article 19-1 of the Industrial Innovation Act provide for equity-based compensation but may not fully accommodate the compensation structures of high-skilled professionals, particularly given applicable caps and conditions. The government should enhance tax treatment for stock-based incentives, including a review of the appropriateness of existing caps as well as the determination of taxation bases and timing, so as to better realign the system with its intended role as a long-term incentive and talent retention tool, while enhancing overall flexibility and attractiveness.

Suggestion 3: Amend Article 10-2 of the Industrial Innovation Act to encourage sustained R&D investment in the semiconductor sector.

Sustained investment in research and development across the semiconductor value chain is essential to maintaining Taiwan’s global competitiveness and supply chain resilience. This includes design, manufacturing, packaging, and supply chains such as equipment and materials, all of which must continue to scale up to maintain Taiwan’s leading position.

Article 10-2 of the Statute for Industrial Innovation is intended to incentivize R&D and advanced equipment investment through tax deductions. However, eligibility criteria established under the implementing regulations may limit access for companies making substantial and sustained R&D investments. Current requirements specify that applicants must meet both a minimum R&D expenditure threshold and a corresponding R&D intensity ratio within the same tax year.

These dual thresholds often create unintended constraints. While companies may meet absolute R&D spending requirements, fluctuations in revenue can affect their ability to meet the R&D intensity ratio, introducing uncertainty in tax-incentive eligibility. As R&D investment is typically planned on a long-term basis, but revenue is subject to market and macroeconomic conditions, the current framework may reduce the predictability and effectiveness of the incentives.

We firmly believe that the policy intent of this tax incentive is to encourage companies to invest resources in researching cutting-edge technologies and to create tangible benefits. If the relevant eligibility requirements instead make it difficult for companies that have already invested and achieved positive results to meet the criteria, this is clearly not what was envisioned when the policy was originally formulated.

The Committee recommends:

1. Apply R&D expenditure and R&D intensity as alternative criteria. Companies should qualify for the incentive by meeting either the absolute R&D expenditure threshold or the R&D intensity requirement, rather than both, to avoid discouraging high-growth firms who have made significant investment in R&D.
2. Introduce flexible and sector-sensitive eligibility thresholds. R&D expenditure and intensity thresholds should be calibrated based on industry characteristics and business models, rather than applying a single fixed standard. This change would better reflect differences in revenue structure and R&D profiles across sectors.

Suggestion 4: Adopt a comprehensive AI strategy by advancing edge AI alongside cloud infrastructure.

Taiwan’s ambition to become a global leader in AI aligns with current technology trends. Achieving this goal will require a balanced approach that supports both centralized cloud infrastructure and distributed, device-level AI applications. While cloud-based training remains essential, long-term competitiveness will also depend on how AI inference is deployed across end-user devices and systems.

Current national strategies have appropriately prioritized AI infrastructure and compute capacity, particularly through data center development. However, an overly centralized approach may introduce structural constraints, including increased energy demand and network bandwidth pressure. As AI applications expand, a growing share of value creation will occur at the edge, across devices and systems that interact directly with users, machines, and environments.

Edge AI enables inference to be performed closer to the data source, reducing reliance on network transmission while improving responsiveness, operational continuity, and data security. A hybrid architecture, where workloads are distributed across cloud and edge environments, is emerging as the global standard for scalable AI deployment.

To strengthen Taiwan’s AI ecosystem, the Committee recommends prioritizing the development and adoption of edge AI, alongside existing cloud strategies. This includes supporting deployment in key sectors such as robotics, automotive, PCs, Internet of Things, and industrial systems, as well as promoting adoption among public sector users and small and medium-sized enterprises (SMEs). A coordinated cloud-to-edge AI approach will enhance system resilience, reduce infrastructure strain, and support Taiwan’s competitiveness in next-generation AI applications.

4.1   Embed a distributed cloud-to-edge AI approach into national flagship programs. Taiwan can strengthen its position in AI-driven semiconductor development by integrating distributed AI architectures into national initiatives, including the Chip-based Industrial Innovation Program and the Ten AI Initiatives Promotion Plan. In addition to supporting large-scale infrastructure and model training, policy should accelerate the development and deployment of edge AI to enable distributed computing across cloud and device environments.

The government should introduce targeted incentives to support R&D, prototyping, and commercialization of distributed AI applications across key industries. Existing programs under the Industrial Innovation Act and digital development initiatives should be leveraged to encourage the adoption of edge AI solutions, particularly among enterprises and SMEs.

In parallel, incentive mechanisms, including grants, vouchers, and tax measures, should recognize AI-capable end-user devices, such as PCs and workstations, as part of the national AI infrastructure. Policy design should also account for the growing importance of AI inference workloads, ensuring that infrastructure planning extends beyond centralized training capacity.

A coordinated cloud-to-edge AI strategy will improve system efficiency, reduce energy and network burdens, and support the development of an integrated AI ecosystem spanning chips, devices, infrastructure, and applications.

4.2   Improve cross-ministerial coordination to ensure more effective and focused policy implementation. AI development requires coordinated policy, infrastructure, and resource allocation across multiple domains. Without central alignment, fragmented programs and overlapping investments may reduce policy effectiveness and slow adoption. A cross-ministerial coordination mechanism is therefore necessary to integrate computing resources, data governance, and application development, while supporting industry collaboration and scalable deployment.

The Committee recommends designating an Executive Yuan-level coordinating authority to oversee AI policy across ministries. This entity should be responsible for setting strategic priorities, aligning budgets, defining performance metrics, and facilitating public–private partnerships to improve implementation efficiency.

The Committee welcomes the enactment of the AI Basic Act and the planned establishment of a National AI Strategy Special Committee. Once operational, this body should play a central role in coordinating national AI strategy and accelerating deployment, including supporting the expansion of AI inference capabilities alongside existing infrastructure development.

4.3   Strengthen international cooperation on supply chain security and trusted AI technology ecosystems. To deepen U.S.-Taiwan cooperation on supply chain security, the Ministry of Economic Affairs should build on the outcomes of the U.S.-Taiwan Economic Prosperity Partnership Dialogue to identify priority areas for collaboration, including AI, semiconductors, drones, and robotics. We also recommend that Taiwan engage with emerging U.S.-led initiatives on AI technology ecosystems to expand opportunities for industrial cooperation.

Policy efforts should support collaboration across key areas, including supply chain diversification through overseas manufacturing partnerships, development of high-quality traditional Chinese-language datasets for AI training, and joint efforts to promote trusted and secure AI technologies in global markets.

We urge the government to engage in regular, structured consultation with industry to inform program design and implementation, strengthen coordination, and ensure that initiatives effectively support cooperation in supply chain resilience and AI ecosystem development between Taiwan and the United States.