China's DeepSeek, an AI developer at the center of China's push to compete globally in artificial intelligence, is reportedly developing its own chip to help run AI systems. Reuters reported on Tuesday, citing unnamed sources, that the chip is being designed for inference, the stage when trained AI models begin generating responses to user queries, rather than for building new models themselves.
The report appeared to weigh on Nvidia NASDAQ:NVDA, the leading AI chipmaker, whose shares fell 2.2% in early trading Tuesday. Investors may view the move as another sign that major AI developers are looking to tailor hardware more closely to their own systems, potentially reducing dependence on outside chip suppliers over time.
The broader custom-chip race also appears to be accelerating. OpenAI, an AI company developing advanced models, recently unveiled its first custom chip with Broadcom NASDAQ:AVGO, a chip and infrastructure software company, while Amazon.com NASDAQ:AMZN, the e-commerce and cloud computing company, is in talks to sell custom chips for other companies' data centers, and Alphabet NASDAQ:GOOGL, Google's parent company, has seen significant demand for its tensor processing units, or TPUs.