After the presidential election in the United States in 2024, the renewed implementation of sufficient tariffs can lead to a fundamental reorganization of global technology supply chains that Power Artificial Intelligence (AI) Development. While established players reconstituted, countries like India are finding themselves in a technical rivalry between the US and China as an indefinite, yet potentially profitable, in a position.
Tariffs have increased the cost of imported components that are important for AI infrastructure. In 2024, electronics imports in the US alone were approximately $ 486 billion, with data processing machine imports with a cost of about $ 200 billion, was largely sour from tariff affected countries such as Mexico, Taiwan, China and Vietnam. These tariffs risk the US to create the most expensive place in the world to build AI infrastructure, which irony for China is that companies driving to transfer data centers abroad.
The first wave of Trump tariff between 2018-20 resulted in the price increase for imported semiconductor components. The current tariff regime has expanded this up to 27% on important AI hardware components in 2025, especially affecting the component formed by special AI accelerator and advanced logic chips, computational foundation.
Economics behind curtains
Economic theory suggests that such tariff policies should stimulate domestic production through import replacement. Indeed, some reports states that the US will exceed its domestic semiconductor manufacturing capacity from 2022 to 2032, which is the largest estimated growth rate globally. However, the classical recordian business theory reminds us that comparative profit also remains operative under protectionist rule. The special nature of AI hardware production means that it has to deal with scattered technical abilities, when the global supply chains cause indispensable disabled when artificially fragmented.
This conservationist approach often comes at the cost of economic efficiency and innovation. Tariffs disrupt global supply chains, increase production costs, and create uncertainty that discourage investment. Eneristic studies suggest that a standard deviation increase in tariffs can reduce production growth by 0.4% in five years, and recent American tariffs can lead to 4% cumulative output gains. In terms of AI-where innovation cycles depend on rapid and state-of-the-art technology and access to global cooperation-such disruption can slow down technological progress and reduce productivity.
Tariffs can adopt domestic firms with competition, reduce their encouragement to innovation, and limit access to advanced imported technologies that are essential for AI advancement. It calls economists “deadweight loss”, where the amount of low trade produces economic disability that neither benefit producers or consumers.
Rapid expansion in demand for AI chip will require massive increase in data center power capacity, in 2024 potential 68 GW from 11 GW to 2027 and 327 GW by 2030. Failure to meet the needs of these infrastructure can reduce the competition of America in AI.
Research indicates that access to expensive, advanced computational infrastructure acts as a primary determinant of innovation capacity in AI, which has a stratification effect. In addition, tariff technology imposed by developed countries can reduce the transfer rates, temporarily changing innovation incentives, which, in turn, can slow down the overall speed of AI innovation. On the other hand, tariffs by developing countries can speed up technology transfer but can affect relative wages and innovation differently. This is a complex difference that can increase global inequalities in AI abilities.
Where India stands
This can create unexpected opportunities for India, which has deployed itself as a strategic “third option” in the US-China technical competition. The Indian IT export growth rate has increased from 3.3% to 5.1% to year-old year in recent years. The AI and Digital Engineering segments are among the fastest growing sectors within the technical field of India. The Government of India has launched important AI-related programs, and has increased semiconductor design, manufacturing and technology investment, which has several billion dollars in the $ 400 million design complex of AMD in semiconductor fib proposals and multinational research and development centers such as AMD in Bangalore.
India’s comparative advantage lies in low labor costs and special knowledge domains. India produces around 1.5 million engineering graduates annually, showing great qualifications for AI development.
India depends a lot on imported hardware components and international cooperations for this. Tariffs and supply chain disruption that AI increase the cost of infrastructure can slow down India’s global ambitions in AI. However, India can also benefit indirectly if companies seek China’s option for manufacturing and data center locations.
These tariff policies intensified the economic revival that has intensified that economists say “capital replacement effect”. As the cost of hardware increases, companies change rapidly towards adapting existing resources through algorithm proficiency, model compression techniques and hardware reforms instead of raw computational power. The tariff environment has effectively created these price signals. The cost of using the AI model is dramatically (about 40 times a year) falling. Therefore, while tariff upfront infrastructure can increase cost, consumer-level AI applications may not see immediate price increase.
Tariff structures novelly interact with specific regulatory environment to create competitive mobility. Lanent data protection regulation, comprehensive digital access, and data availability can partially offset the loss of hardware cost through more access to training data. Regulators and economic factors can defy simple analysis.
Decentralized ai development
Tariff changes have led to the development of special AI hardware designed for special applications rather than particularly general-objective calculations. This “application-specific integrated circuit” represents an architectural change. To optimize data centers infrastructure for AI estimate, more than 50% of the workload accelerator in 2028 can be custom asics by 2028, above 30% in 2023.
The irony is that policies can inadvertently accelerate decentralization of AI development with the aim of strengthening domestic technical capabilities. Historical similes suggest that technologies facing market deficiency often develop towards more distributed implementation. The mainframe-to-individual computer infection of the 1980s provides an educational parallel.
Arindam Goswami is a research analyst at a high-tech geophysical program in Taxila Institution, Bangalore.
Published – May 23, 2025 12:16 AM IST