Huawei steps in as Nvidia faces export restrictions, ready to launch AI chip 910C
China’s Huawei Technologies is ready to fill the gap left by the United States’ restrictions on Nvidia selling its AI chips to Chinese customers.
The move, which has been in the works for a while, comes after the Trump administration barred Nvidia from selling its custom H20 processor to customers in China.
The news sent Nvidia’s stock down 4.5% to close at 96.91 on Tuesday, and the company warned it could write off as much as $5.5 billion for inventory in its fiscal first quarter.
Now, according to a Reuters report, Huawei is ready to begin mass shipments of its new AI chip, the 910C, next month. The chip is designed to fill the void left by the banned H20 and has already seen some initial shipments.
The report says the 910C achieves performance on a par with the H20 by combining two of Huawei’s 910B processors into a single package. It’s not clear whether those shipments have gone out yet.
The Trump administration informed Nvidia earlier this month that it would require an export license to sell its custom H20 processor in China, prompting Nvidia to anticipate the write-off for inventory in its fiscal first quarter.
Nvidia’s chief rival, Advanced Micro Devices (AMD), is also feeling the heat from the trade restrictions. It faces a trade ban on its MI308 products and expects to incur a charge of up to $800 million for inventory related to AI chip sales to China.
In addition to the 910C, Huawei is planning to launch the Ascend 920 later this year—a chip said to match the performance of the Nvidia H20, with manufacturing reportedly handled by China’s SMIC.
However, Wedbush Securities analyst Matt Bryson raised concerns about SMIC’s ability to produce the chips and potential yield challenges. He noted that if those issues are overcome and U.S. restrictions remain in place, Huawei’s offerings could severely limit Nvidia’s opportunities in China.
Nvidia’s stock is down nearly 28% year-to-date amid ongoing trade concerns and the increasing demand for AI processors in data centers.
