Trump Administration Weighs Tariff Exemptions for Hyperscalers Tied to Taiwanese Chip Investments
February 10, 2026
The Trump administration is reportedly considering a targeted exemption from upcoming semiconductor tariffs for major US hyperscale cloud providers, a move that seeks to balance the push for domestic chip manufacturing with the immediate needs of the booming data center industry. This potential policy shift underscores the critical role of advanced semiconductors in powering artificial intelligence and cloud services, and the complex geopolitical and supply chain calculations shaping the sector.
According to a report by the Financial Times, the administration is planning to grant carve-outs from forthcoming chip tariffs to technology giants including Amazon, Google, and Microsoft. The exemptions would be directly linked to investment commitments from Taiwanese semiconductor companies, most notably TSMC. Under the proposed framework, these Taiwanese firms would be granted permission to allocate the tariff exemptions to their US-based customers, effectively shielding the hyperscalers from additional costs on imported chips essential for their data center operations. The discussions, which sources caution are still in an early phase with many details unresolved, represent an effort to reconcile the administration's "reshoring" ambitions with the industry's continued near-term reliance on global semiconductor supply chains. This follows a major trade agreement signed between the US and Taiwan in January 2026, which committed Taiwanese semiconductor and technology companies to invest at least $500 billion in US chip production capacity. The deal comprises $250 billion in direct corporate funding, backed by an additional $250 billion in credit guarantees from the Taiwanese government. As part of that January agreement, the US agreed to lower its tariff rate on goods from Taiwan to a maximum of 15 percent, down from 20 percent. It also established specific import allowances: Taiwanese companies planning new US chip fabrication plants can import up to 2.5 times the expected capacity of those facilities tariff-free, while firms with existing US fabs can import 1.5 times their capacity. The potential new carve-out for hyperscalers would build upon this structure, creating a more direct conduit for tariff relief aimed at the end-users fueling demand.
The broader context is a planned US tariff regime on semiconductors, proclaimed by the White House shortly after the US-Taiwan trade deal was announced. While the government stated that tariffs would not apply to chips imported to support the buildout of the US technology supply chain—including for data centers—the reported discussions indicate a more nuanced, incentive-based approach is being formulated. If implemented, this policy could significantly ease cost pressures on cloud providers amid an unprecedented AI-driven demand surge for computing power. It also reinforces the strategic interdependence between US technology firms and Taiwan's semiconductor industry, using trade policy to incentivize onshore investment while maintaining access to vital components. The outcome will have substantial implications for the cost structure of data center operators and the pace of AI infrastructure deployment across the United States.
Source: datacenterdynamics
