Nvidia quantifies that new U.S. restrictions on H20 AI chip exports will result in $5.5 billion in lost annual revenue.

Nvidia (NASDAQ:NVDA) announced it will incur $5.5 billion in charges following the U.S. government's indefinite restriction on exports of its H20 AI chip to China, a critical market for the company's data center products. The H20 chip, which was tailored to comply with U.S. export limits, has served as Nvidia's most advanced model available in China and represents a key component in partnerships with tech giants including Tencent, Alibaba, and ByteDance. The charge accounts for inventory write-downs, purchase commitments, and other reserves related to the H20 chip.

The U.S. restrictions stem from concerns that H20 chips, despite reduced AI training capabilities, offer high-speed memory and chip connectivity—features potentially usable in supercomputers. Washington has banned chips linked to Chinese supercomputing efforts since 2022. The latest move follows a report by the Institute for Progress claiming that firms like Tencent and DeepSeek may already be deploying H20s in ways that violate export controls. Tencent denied the claims, asserting it maintains "full compliance" and stating it "has not built any supercomputers."

On April 9, Nvidia was informed by the U.S. government that H20 shipments to China would require a license, with those rules made indefinite by April 14. It remains unclear whether any licenses will be granted. The development comes one day after Nvidia revealed plans to build $500 billion worth of AI servers in the U.S. over the next four years with partners including TSMC, aligning with the Trump administration's focus on domestic semiconductor production.

Nvidia shares fell 6% in after-hours trading following the announcement, reflecting investor concerns over the impact of losing access to China's booming AI market. The company declined further comment, and the U.S. Department of Commerce has yet to respond.