The White House has imposed a 25% tariff on certain sales of AI chips from Nvidia and AMD to China to directly capture a share of the revenue generated from strategic technologies.
Assumed Transactional Agreement by the White House
The United States has reached a new milestone in its industrial and trade policy. The White House has announced the implementation of tariffs equivalent to 25% of the value of sales of certain artificial intelligence chips from Nvidia and AMD to China. This unprecedented mechanism is designed to allow the federal government to directly capture a portion of the revenue generated by the export of strategic technologies.
We’re going to make 25% on the sale of these chips, roughly. So we’re allowing them to do it, but the United States gets 25% of the chips in monetary terms. And I think that’s a very good deal.
Behind this decision lies a behind-the-scenes negotiated compromise. In December, the U.S. administration agreed to ease its export restrictions by allowing Nvidia to supply its H200 processors to Chinese customers, which were previously blocked. In return, Washington demanded to recover 25% of the sales value.
The new tariffs announced this week serve to give a legal basis to this arrangement. According to several industry leaders, this mechanism aims to secure the agreement against potential legal challenges by integrating it into a national security investigation already open on semiconductors.
The American president fully embraces this approach. For the White House, it’s a “good deal”: companies can export, but the government directly collects a substantial share of the revenue.
Targeted Chips, an Evolving Scope
The tariffs notably cover Nvidia’s H200 processors and AMD’s MI325X. The levy applies when these chips are first imported into the United States before being shipped to international customers, including China. However, components intended to strengthen AI infrastructure on U.S. soil are not subject to this tax.
This distinction reflects Washington’s strategy: encouraging domestic production and use while monetizing access to foreign markets. But the presidential text also hints at a threat. A second phase of the investigation could result in much broader tariffs affecting all semiconductors and their derivatives.
A Precedent with Far-Reaching Consequences for the Tech Industry
This measure marks a break. Until now, the U.S. mainly used export controls as a geopolitical tool. Now, they also use it as a direct source of revenue. For Nvidia and AMD, the financial impact is immediate: a 25% cut on sales already subject to strong political constraints.
Nvidia, however, publicly praised the decision, stating that it strikes a balance between economic interests and national security. American companies know that an outright refusal to export to China would have much heavier consequences on their results, in a context where this market remains crucial despite tensions.
Chinese Uncertainty in the Background
However, there’s no guarantee that these chips will actually reach their end customers. Chinese authorities are sending signals in favor of technological self-sufficiency and urging local companies to prioritize domestic alternatives. Recent instructions have even temporarily blocked certain customs procedures.
Beyond the cases of Nvidia and AMD, this decision illustrates a broader evolution: the transformation of technological value chains into tools of economic and budgetary policy. For markets, the message is clear. Access to American technology now comes at a price, set directly by the state.
