Nvidia circuit board

Nvidia shares edge lower after report says DeepSeek is developing an in-house AI chip (NVDA)

Report points to DeepSeek’s custom chip ambitions

Nvidia (NASDAQ:NVDA) shares fell 1.6% in premarket trading on Tuesday after a Reuters report said Chinese artificial intelligence startup DeepSeek is designing its own semiconductor for AI computing, a move that could reduce its reliance on hardware supplied by Nvidia and Huawei.

According to sources familiar with the project, DeepSeek’s chip is being developed primarily for inference workloads, where trained AI models generate responses to user requests, rather than for training large language models.

Focus on inference reflects changing AI demand

The report noted that DeepSeek’s effort targets one of the fastest-growing areas of the AI market, as demand increasingly shifts from training models to deploying them in real-world applications.

Inference tasks can often be processed using specialised chips that are less expensive and consume less power than traditional graphics processing units (GPUs), making the segment increasingly attractive for AI developers.

Development remains at an early stage

Reuters reported that DeepSeek has been working on the initiative for around a year and has held discussions with chip design companies, semiconductor foundries and memory suppliers.

The company has also quietly expanded its semiconductor engineering team in recent months, recruiting chip designers without advertising vacancies on public job platforms.

Strategic move could reshape competition

If successful, the project would represent a major strategic shift for DeepSeek, which gained international attention after launching highly efficient AI models that challenged established industry players.

The report added that developing an in-house processor would also increase competitive pressure on Huawei while placing DeepSeek alongside a growing number of AI companies seeking to design proprietary chips instead of relying exclusively on Nvidia’s hardware.

Technical and regulatory hurdles remain

Despite the strategic appeal, developing a competitive AI processor typically requires significant investment and years of engineering work.

DeepSeek also faces additional challenges stemming from US export controls, which restrict Chinese companies’ access to leading-edge semiconductor manufacturing facilities and advanced high-bandwidth memory technologies essential for high-performance inference chips.

Nvidia stock price


Posted

in

,

by

Tags: