How Trump’s planned chip tariffs could spare Big Tech while pressuring TSMC

According to officials familiar with the discussions, the commerce department is working on a system that would effectively shield major US tech companies such as Amazon, Google and Microsoft from a new wave of chip tariffs.

The mechanism would not be a blanket exemption. Instead, it would be tied directly to investment commitments made by Taiwan Semiconductor Manufacturing Company, better known as TSMC.

TSMC produces the bulk of the world’s most advanced chips, including those used in AI data centres. Much of its production remains in Taiwan, but the company has pledged to invest $165bn in expanding capacity in the US, including large fabrication plants in Arizona.

Under the proposed framework, the more TSMC invests and builds in the US, the more tariff exemptions it could earn. Those exemptions could then be allocated to its American customers, allowing them to import certain chips tariff-free.

Linking tariffs to investment

The plan builds on a broader US-Taiwan trade agreement under which Washington agreed to cut tariffs on Taiwanese imports to 15 per cent in exchange for a USD 250 billion investment commitment in US chip production.

Under outlines released by the commerce department, Taiwanese companies building new semiconductor plants in the US would be allowed to import chips tariff-free at levels tied to the planned output of those facilities. Companies with existing US plants would also receive scaled exemptions.

In practical terms, this means TSMC could bring in chips during construction or expansion phases without facing the full brunt of new tariffs. It could also extend the benefit of those exemptions to American customers, easing pressure on companies racing to build AI infrastructure.

However, administration officials have cautioned that the details remain in flux and have not yet been formally signed off by the president.

Balancing pressure and protection

The strategy reflects a broader pattern in Trump’s trade policy. Tariffs are used as leverage to drive domestic investment, but with carve-outs designed to avoid harming strategically important sectors.

So far, the administration has avoided imposing sweeping semiconductor tariffs that would directly hit the AI supply chains of US technology groups. Instead, it has targeted a narrower category of chips, including certain semiconductors imported into the US and then re-exported to China.

In January, the White House imposed 25 per cent tariffs on that limited category, affecting chips sold by companies such as AMD and Nvidia for re-export. But chips imported to build out domestic AI data centres have not been subject to those levies.

At the same time, the administration has signalled that broader tariffs could be introduced as part of a second phase of a national security investigation into the semiconductor sector. That second phase could include offset or rebate programmes for companies that invest in US production.

The strategic stakes

For Washington, the goal is clear: reduce dependence on overseas manufacturing, especially in Taiwan, while maintaining US dominance in AI.

For TSMC, the incentive is equally straightforward. Greater US investment could translate into preferential tariff treatment and stronger ties to key customers.

For Big Tech, the carve-out could prove crucial. The companies building the data centres that power generative AI systems depend on a steady flow of advanced semiconductors. Broad chip tariffs would raise costs and potentially slow expansion.

Whether this carefully structured approach succeeds will depend on execution. If the exemptions are seen as overly generous, critics may argue they undercut the tariff policy’s intent. If they are too restrictive, they risk disrupting the AI boom the administration is eager to support.

For now, the message from Washington is that tariffs remain central to its strategy, but they will be calibrated to protect the industries driving America’s technological edge.