One Year After Liberation Day: What Trump's Tariffs Actually Did to AI — The Honest Reckoning

One year ago today — April 2, 2025 — US President Donald Trump stood in the Rose Garden and announced sweeping 'reciprocal tariffs' against more than 50 countries. The market sold off immediately. AI stocks were among the largest losers. The conventional wisdom said the AI boom was threatened. One year later, global trade has grown faster than the world economy, both US imports and Chinese exports reached all-time highs, and the AI industry has continued to attract trillions in investment. But the tariff shock had real, lasting effects on AI infrastructure — effects that are still compounding today and will shape who leads the AI race over the next five years.

What the Tariffs Actually Did to AI Datacenter Costs

The most direct impact of tariffs on AI has been datacenter construction costs. US AI firms are currently spending hundreds of billions of dollars building new datacenters to train models. Tariffs increased the already enormous costs of that effort through several channels. Most Nvidia servers are assembled in Mexico and benefited from tariff exemptions under trade agreements — analysts called this a silver lining for the sector. However, much of the other datacenter infrastructure — cooling systems, power equipment, transformers — is heavily imported and faced full tariff exposure. The US faces a critical shortage of power transformers, with 55% of in-service units over 33 years old. US manufacturers cover only 20% of domestic demand. Tariffs on imports from Mexico and China directly increased the cost of building the electrical infrastructure that AI datacenters require.

The US-China AI Race: One Year On

• Tariffs escalated dramatically before a partial truce. The US imposed tariffs reaching 145% on Chinese goods; China responded with 125% on American goods. In May 2025, both sides reached a temporary agreement — US tariffs dropped to 30%, China reduced to 10%. The truce held through 2025 but remains fragile as of April 2026.
• China's export of rare earth materials emerged as the most effective counter-weapon. China restricted exports of six heavy rare earth elements essential for electric motors and advanced electronics. This leverage point forced more significant US policy adjustments than tariff retaliation alone.
• Chinese AI firms embraced open source and gained global developer trust. While US firms competed in the closed-model market, Chinese labs — particularly DeepSeek — released powerful open-weight models that could be downloaded and run locally. This earned them credibility in the global developer community that commercial US models could not match.
• The tariffs may have inadvertently helped China in the AI race by driving countries like Vietnam, Malaysia, and Cambodia closer to China economically. Taiwan's semiconductor exports to Southeast Asia hit record highs — some of which analysts believe were routed onward to Chinese data centers, potentially circumventing export controls.

The Global Trade Reality vs. The Political Narrative

McKinsey Global Institute's Geopolitics and the Geometry of Global Trade report, published in March 2026, reached a counterintuitive conclusion: the tariff war did not kill global trade. Global trade grew faster than the world economy in 2025 despite US tariff rates at their highest since World War II. Chinese exports reached an all-time high. The trade flows shifted — away from direct US-China trade and through third countries — but did not collapse. The official narrative of Liberation Day was about bringing manufacturing back to the US. The statistical reality is that US manufacturing has grown modestly while AI infrastructure costs increased and global supply chains reorganized around the US-China divide.