From Oil to AI: India’s Next Import Vulnerability

The AI boom is reshaping global trade, but it is also exposing a new vulnerability for India: dependence on imported computing hardware. As data centres expand and AI investment accelerates, demand for servers, processors, graphics cards, memory and networking equipment is surging. The question is whether India can build its digital economy without creating a new strategic import dependence.

A recent note by the US Federal Reserve provides a useful starting point. It identifies HS-6 categories covering processing units, computing units and related parts as a proxy for AI-related trade. Global trade in these products exceeded $272 billion in the first half of 2025, up 65% from a year earlier. They still account for only about 2% of global merchandise trade, but their growth is far outpacing overall trade.

Asia is at the centre of this shift. Taiwan and South Korea are major beneficiaries because AI capital expenditure translates into semiconductor, memory and hardware production. Taiwan’s exports of AI-related products to the US alone were equivalent to roughly 14% of GDP in the second quarter of 2025, according to the Federal Reserve. South Korea is seeing a similar boom, with semiconductor exports rising 199.5% year-on-year to a record $44.8 billion in June 2026 as AI investment drove demand.

China is building its own AI infrastructure, while Malaysia has become a major data-centre hub. A new Asian investment cycle is taking shape, with some economies producing the hardware and others attracting the infrastructure that uses it.

India is entering this cycle largely from the other side of the trade ledger. Using the Federal Reserve’s definition, AI-related imports stayed broadly within $150 million-$500 million a month for almost two decades. That changed sharply in 2025. Monthly imports climbed to around $700 million-$1 billion before exceeding $2 billion in April 2026. In April and May, they rose 216.7% and 222.6% year-on-year respectively, well ahead of global growth.

Some caution is warranted. The Federal Reserve itself notes that these HS codes are an imperfect measure: they include some non-AI products and exclude other products that may be used in AI infrastructure. They should therefore be regarded as an indicator of the hardware-intensive AI investment cycle rather than a precise measure of AI imports.

But the broader trend is clear. Business Standard reported that India’s memory-chip imports reached $6.79 billion in 2025-26, up 53.4% from 2024-25. That matters because memory is increasingly critical to AI infrastructure, with chipmakers allocating more capacity to high-bandwidth and server memory.

India’s data-centre industry is expanding rapidly. Operational capacity rose from about 375 MW in 2020 to around 1,500 MW in 2025. The Ministry of Power estimates that data-centre electricity demand could reach 13.56 GW by 2031-32. As data centres multiply and AI computing requirements rise, demand for servers, processors, memory, networking equipment and related hardware is unlikely to fade.

New Dependence
For decades, India's dependence on imported crude oil has been one of the defining vulnerabilities of its external sector. The transition to a digital economy could create a different, but potentially equally important, form of import dependence. The critical input is no longer merely energy. It is computing capacity — and at the heart of that capacity are semiconductors. Chips will need to be replaced and upgraded every 5-7 years.

The government itself increasingly recognises this vulnerability. In August, the Ministry of Electronics and Information Technology acknowledged that India's AI strategy faces risks from limited domestic capabilities in semiconductor manufacturing, computing infrastructure, foundational models and advanced research. It also confirmed that India's current compute ecosystem relies on globally sourced GPUs.

India is responding. Twelve semiconductor manufacturing projects have been approved with committed investment of ₹1.64 trillion. Semicon 2.0, approved in July, carries an outlay of ₹1.3 trillion and aims to expand fabrication, advanced packaging, chip design and the wider semiconductor ecosystem. India has also joined the Pax Silica coalition, focused on strengthening the global “silicon stack” from critical minerals and semiconductor fabrication to advanced AI systems and deployment infrastructure.

These are important moves, but scale matters. Taiwan and South Korea convert AI investment into domestic production of critical hardware. China is seeking a more self-reliant AI stack. Malaysia is capturing data-centre investment and becoming a regional infrastructure hub. India risks capturing the downstream investment while importing much of the hardware needed to support it.

Strategic Nodes
The answer is not to make everything in India. Semiconductor supply chains are too complex, capital-intensive and specialised for that. Nor should every import be replaced with domestic production. The goal should be to build capabilities in strategically important parts of the supply chain while diversifying import sources.

The experience of renewables and electric vehicles is instructive. India rapidly expanded domestic deployment while remaining heavily dependent on imported solar modules, cells and batteries. Much of the investment multiplier therefore leaked abroad. Data centres could create the same problem on a much larger scale.

The opportunity is to move beyond being a large consumer of AI and become part of the production ecosystem it is creating. Domestic manufacturing of servers and data-centre equipment is a start. The larger prize lies upstream: semiconductor fabrication, advanced packaging, memory, networking equipment and other components where value is captured.

The AI boom is already redrawing global trade. India now has a choice: produce more of the critical inputs, or import them at scale and export the value that could have been captured at home. Chips could become the new crude.