Trump's tariffs crushed traditional trade but AI-related commerce kept growing. The reason: AI hardware runs through US-allied supply chains. The winners are not obvious — and India is on the list.

The Tariff War Is Reshaping Global AI Trade — And the Winners Are Not Who You Think

When Donald Trump's second-term tariff regime went into full effect in early 2025, the conventional prediction was straightforward: higher barriers, slower trade, economic pain distributed roughly proportionally across all sectors. The conventional prediction was wrong — at least for AI.

New data published this week by Euronews, drawing on WTO and IMF trade statistics, shows that AI-related commerce has been the single largest exception to the global tariff shock. While traditional trade routes between major powers contracted sharply, AI-driven trade — hardware, cloud services, training data, inference infrastructure — continued to expand. The critical detail: much of this growth happened between geopolitically aligned economies, not the US-China corridor that the tariffs were designed to squeeze.

The mechanism is not complicated. AI infrastructure requires NVIDIA GPUs, which are manufactured primarily in Taiwan (TSMC) and assembled in supply chains that run through South Korea, Japan, and the Netherlands. These countries are US allies. The tariffs hit Chinese goods hard; they left the AI hardware supply chain largely intact. The result is that the AI buildout continued — accelerated, even — while simultaneously, the geopolitical reshaping of trade created new dependencies between the US and its Pacific allies that are now structural.

Bloomberg reported separately that China, facing restrictions on the most advanced NVIDIA chips, responded by accelerating domestic GPU development through companies like Huawei's Ascend division, and by doubling down on AI software and post-training infrastructure where the compute requirements are lower. This is not a story of China being left behind. It is a story of two parallel AI ecosystems diverging — one dependent on TSMC and NVIDIA, one increasingly self-contained — both growing rapidly.

What this means for Sam Altman's OpenAI, Dario Amodei's Anthropic, and Google DeepMind's Demis Hassabis is a strategic bifurcation that none of them fully anticipated. The global AI market is not one market anymore. It is several overlapping markets with different regulatory regimes, different compute supply chains, and increasingly different model architectures optimized for different hardware.

Harvard economists, speaking at an event streamed publicly this week, flagged the AI-tariff intersection as one of the most underanalyzed macro risks of 2026. The argument: policymakers are still thinking about AI as a software sector — intangible, borderless, hard to tax. The reality is that AI is profoundly physical. It runs on specific chips, built in specific fabs, cooled by specific infrastructure, powered by specific energy sources. Every one of those physical dependencies is now subject to trade policy. When tariffs reshape where chips get made and where data centers get built, they reshape where AI capability accumulates.

For India, this dynamic is an opening. India is not subject to the same restrictions as China on advanced chips. It has growing data center infrastructure, a massive English and multilingual developer base, and no existential chip supply chain vulnerability. The AI infrastructure investment flowing into Bengaluru and Hyderabad right now — Deccan AI's $25M round, Cartesia's new India operations, the NVIDIA data center investments announced last year — all make more sense when viewed against this tariff backdrop.

The winners of the tariff war in AI are not obvious from the headlines. They are the countries that stayed allied, stayed connected to the hardware supply chain, and built the human capital to use it. India is on that list. Europe — fragmented, slow on chips, dependent on American cloud — is increasingly not.

The geopolitics of AI is no longer a separate conversation from the geopolitics of trade. They are the same conversation now.

Deep Dive

See our previous coverage on India's AI rise: India's AI Startups Are Raising Real Money Now