Published 5 hours ago • loading... • Updated 1 hour ago
China's DeepSeek developing its own AI chip: Reuters
The early-stage effort could help DeepSeek lower chip costs as Chinese firms face tighter U.S. export controls and memory supply limits.
Chinese AI startup DeepSeek is developing its own inference chip, according to three people familiar with the matter, aiming to reduce reliance on Nvidia and Huawei chips used to train and run its models.
Tapping inference demand, the company targets the fastest-growing segment of AI computing, where specialized chips can be cheaper and less power-hungry than general-purpose GPUs running trained models.
The Hangzhou-based company has quietly increased hiring of chip-design engineers in recent months without public job postings and is holding discussions with external foundry and memory partners.
Following global trends, DeepSeek joins developers like Anthropic seeking hardware control, while simultaneously pursuing a maiden funding round valuing the company at between $52 billion and $59 billion.
Washington's export controls present significant hurdles, barring Chinese designers from accessing advanced foundries and cutting access to high-bandwidth memory critical for competitive inference chip manufacturing.
(Beijing=Yonhap News) Correspondent Jung Sung-jo = Chinese artificial intelligence (AI) startup DeepSeek is developing its own AI chip, and if this move bears fruit, Nvidia...