Nvidia stock remains under pressure on Tuesday: what's hurting the AI darling?

Nvidia stock NVDA fell on Tuesday after a report that Chinese artificial intelligence startup DeepSeek is developing its own inference chip added to investor concerns that major AI customers are increasingly looking to reduce their dependence on Nvidia's hardware.

Shares of the chipmaker dropped 1.9% to $191.82 in early trading. If the decline holds, the stock would close at its lowest level since April.

DeepSeek reportedly developing custom AI chip

The latest pressure on Nvidia followed a Reuters report that DeepSeek is developing its own artificial intelligence chip, citing people familiar with the matter.

According to the report, the processor is designed for inference—the stage of AI computing in which trained models generate responses to users—rather than for training new models.

Reuters reported that the effort remains in its early stages, with DeepSeek holding discussions with chip-design companies, foundries, and memory suppliers. The initiative reportedly began about a year ago.

If successful, the move would reduce DeepSeek's reliance on external suppliers, including Nvidia and China's Huawei Technologies.

The report said DeepSeek has used both Nvidia and Huawei chips to train and deploy its AI models.

DeepSeek previously said the foundation model behind its R1 reasoning model was trained using Nvidia's H800 processors, chips designed specifically for the Chinese market before US export restrictions barred their sale.

The company has since relied increasingly on Huawei hardware.

In April, DeepSeek released its V4 model adapted for Huawei's Ascend chips, while Huawei said its processors were used in part of the training of DeepSeek's V4-Flash model.

Export controls reshape China's AI chip market

DeepSeek's reported push into chip development comes as Chinese AI companies face continued restrictions on access to Nvidia's most advanced processors under US export controls.

The limitations have encouraged domestic technology companies to pursue alternative hardware solutions, while Beijing has continued encouraging the development of a domestic AI semiconductor ecosystem.

Huawei has emerged as one of the largest beneficiaries of those restrictions.

Huawei is now estimated to supply around half of China's estimated $50 billion domestic AI chip market.

However, that position is increasingly being challenged as companies, including Alibaba and Baidu, develop their own AI processors.

Part of a broader industry trend

Even if DeepSeek eventually deploys proprietary inference chips, the immediate business impact on Nvidia may be limited.

China has become a progressively smaller contributor to Nvidia's revenue following successive rounds of US export restrictions.

Although the development is unlikely to materially affect Nvidia's near-term financial results, investors may see it as another sign of mounting competitive pressure.

Large AI developers have increasingly sought greater control over their computing infrastructure by designing custom silicon tailored to their own workloads.

Major Nvidia customers, including Microsoft and Meta Platforms, have already been investing in internally developed AI chips as they seek to lower infrastructure costs associated with expanding data-center capacity.

The trend has also spread to leading AI model developers. Last month, OpenAI unveiled Jalapeño, its first custom inference chip developed with Broadcom, while Anthropic has reportedly been evaluating the development of its own processors.