Washington is drafting regulations that would require licenses for global exports of advanced AI chips. Shipments of up to 1,000 units would undergo a streamlined review, while larger orders would need pre-clearance and mega-deployments could involve host-government oversight. The proposal aims to give the U.S. broad gate-keeping authority without outright banning exports.

In fiscal 2026, about 70% of Nvidia’s $216 billion in sales were to U.S.-headquartered customers, with Taiwan accounting for roughly 20%. China-bound sales, including H200 GPUs, represent a key growth segment but are not separately quantified in earnings. Bernstein warns that stricter export requirements could slow sovereign adoption opportunities estimated at over $30 billion.

Nvidia has stopped production of H200 GPUs destined for China due to mounting regulatory pressure in both Washington and Beijing. The decision pauses a critical supply line for the Chinese AI market and could disrupt customer deployments. Nvidia will reallocate capacity to other regions while it assesses compliance and licensing options.

Global demand for AI computing infrastructure remains robust, driven by cloud providers and enterprise AI projects. However, more stringent export rules could delay large-scale rollouts in government and telecommunications sectors. Nvidia’s ability to navigate licensing hurdles will be key to sustaining its market leadership and revenue growth.