Editorial 1: ​​Tariff wars and a reshaping of AI’s global landscape

Context

Economic efficiency and innovation may suffer, but some countries could be vulnerable yet advantaged.

Introduction

Following the 2024 U.S. presidential election, the revival of major tariffs could trigger a deep restructuring of global tech supply chains critical to AI development. As dominant players adjust, countries like India may emerge in a precarious yet promising role — serving as a “third option” in the U.S.–China tech rivalry.

Impact of Tariffs on AI Infrastructure Costs

• Increased Cost of Critical Imports
  • Tariffs have significantly raised the prices of imported components essential for AI infrastructure.
  • In 2024, the U.S. imported nearly $486 billion worth of electronics.
    • Of this, around $200 billion was spent on data processing machines, primarily sourced from Mexico, Taiwan, China, and Vietnam—all affected by U.S. tariffs.
  • These rising costs risk making the U.S. the most expensive country to build AI infrastructure.
  • As a consequence, companies may relocate data centre projects to more cost-effective countries, including China—the very country many of the tariffs were meant to counter.

Tariff Evolution and Expansion

• Trump-Era Tariffs (2018–2020)
  • The initial wave led to increased prices for imported semiconductor components.
• Current Tariff Landscape (2025)
  • The current regime imposes tariffs of up to 27% on critical AI hardware components.
  • These include:
    • Specialised AI accelerators
    • Advanced logic chips
  • These components are fundamental to AI computation, making the tariffs especially disruptive to the U.S. tech sector.

Economics behind the scenes

1. Economic Rationale vs. Practical Challenges of Tariffs

• Theoretical Justification:
  • Tariffs are designed to stimulate domestic production through import substitution.
  • Example: U.S. semiconductor manufacturing capacity is projected to triple from 2022 to 2032 — the largest global growth rate in this sector.
• The Ricardian Reality:
  • Ricardian trade theory reminds us of comparative advantage, which continues to apply even under protectionism.
  • AI hardware production relies on globally dispersed technical capabilities, making it inefficient when global supply chains are disrupted.

2. Economic Costs of Protectionist Tariffs

• Losses in Efficiency and Innovation:
  • Tariffs cause:
    • Supply chain disruptions
    • Increased production costs
    • Investor uncertainty
  • These factors discourage innovation and long-term investment.
• Empirical Evidence:

• Deadweight Loss:
  • Tariffs may shield domestic firms from foreign competition, reducing innovation incentives.
  • Restricted access to advanced imported technologies creates inefficiencies, harming both producers and consumers.

3. AI-Specific Impacts: Infrastructure, Innovation, and Inequality

• AI Infrastructure Requirements:

• The rapid increase in AI chip demand necessitates massive infrastructure scaling.
  • Failure to meet these needs will undermine U.S. competitiveness in AI.
• Innovation Stratification:
  • Advanced, costly AI infrastructure becomes a barrier to entry and a key factor in innovation leadership.
  • This creates a stratification effect, where only a few players control major breakthroughs.
• Tariff-Driven Global Inequality:

• This complex dynamic alters innovation incentives and risks widening global disparities in AI capability.

India’s Strategic Opportunity Amid U.S.-China Tech Rivalry

• India is emerging as a strategic “third option” in the ongoing U.S.-China technological competition.
• IT export growth:
  • Indian IT exports have grown at 3.3% to 5.1% year-over-year recently.
  • AI and digital engineering are among the fastest-growing segments within India’s tech sector.
• The Indian government is actively supporting AI and semiconductor sectors through:
  • Significant AI programmes.
  • Billion-dollar semiconductor fab proposals.
  • Multinational R&D centres such as AMD’s $400 million design campus in Bengaluru.

India’s Comparative Advantages and Challenges

• Potential Risks:
  • Tariffs and supply chain issues may raise AI infrastructure costs, slowing India’s global AI ambitions.
• Potential Benefits:
  • India could gain if companies seek alternatives to China for manufacturing and data centre operations.

Economic Effects of Tariffs on AI Development

• Tariff policies have accelerated the “capital substitution effect”:
  • As hardware costs rise, firms shift focus to:
    • Algorithmic efficiency
    • Model compression techniques
    • Hardware optimization rather than simply scaling raw computational power.
• This creates price signals encouraging innovation in efficiency rather than just hardware expansion.

• Consequently, while tariffs may increase upfront infrastructure expenses, consumer-level AI applications may not experience immediate price rises.

Role of Regulatory and Economic Environments

• Tariffs interact with different regulatory frameworks to shape competitive dynamics:
  • Lenient data protection laws, widespread digital access, and abundant training data can offset hardware cost disadvantages.
• This interplay means that regulatory and economic factors may produce complex, non-linear effects on AI competitiveness, defying simplistic analysis.

Decentralised AI development

• Tariff changes have driven the development of specialised AI hardware designed for specific applications instead of general-purpose computing.
• This shift is characterized by the rise of application-specific integrated circuits (ASICs), marking a new architectural approach.
• To optimize data centre infrastructure for AI inference:
  • In 2023, about 30% of workload accelerators were custom ASICs.
  • By 2028, this share is expected to exceed 50%.

Conclusion

Ironically, efforts aimed at boosting domestic technological strength might unintentionally speed up the decentralisation of AI development. Historical parallels indicate that when technologies encounter market limitations, they tend to shift toward more distributed models. A relevant example is the transition from mainframes to personal computers during the 1980s, which illustrates this trend well.