Nvidia Shares Fall Nearly 2% After Report Says China's DeepSeek Is Developing Its Own AI Inference Chip

Shares of Nvidia fell Tuesday, trading at $191.70, down $3.85, or 1.97 percent, after a Reuters report said Chinese artificial intelligence company DeepSeek is developing its own semiconductor for AI computing, a move that could reduce the startup's reliance on chips from both Nvidia and Huawei.

Note: This article is intended to provide factual context and does not constitute financial advice. Readers should consult a licensed financial advisor before making investment decisions.

According to the Reuters report, citing three sources familiar with the matter, DeepSeek's chip is being designed specifically for inference, the stage of AI computing in which an already-trained model generates responses to user queries, rather than for training new models from scratch. That focus places the project squarely within what has become the fastest-growing segment of AI computing demand, as more of the industry's overall workload shifts from building large language models to running them for end users. Inference tasks can often be handled by specialized, lower-cost chips that consume less power than the general-purpose graphics processing units Nvidia has built its business around.

The effort reportedly began roughly a year ago and remains in its early stages. According to the Reuters report, DeepSeek has been in discussions with outside chip-design firms, semiconductor foundries and memory suppliers as part of the initiative, while also quietly increasing its hiring of chip-design engineers in recent months without posting public job listings for those specific roles.

DeepSeek's move would mark a significant strategic shift for a company that has been held up in China as a national AI success story since it drew rapid global attention last year with the release of two highly efficient AI models. The base model for DeepSeek's R1 reasoning system was trained using Nvidia's H800 chip, a China-specific variant that the U.S. government subsequently banned from export in late 2023. Since then, DeepSeek has increasingly shifted its computing workloads toward Huawei's Ascend chip lineup, releasing a version of its V4 model earlier this year that was specifically optimized for Huawei's hardware.

Should DeepSeek successfully develop its own competitive inference chip, the move would add to the competitive pressure already facing both Nvidia and Huawei within China's AI chip market, according to the Reuters report. It would also place DeepSeek alongside a growing list of AI developers worldwide that have moved to design their own custom silicon rather than relying solely on Nvidia's hardware. OpenAI unveiled its first custom inference chip, developed in partnership with Broadcom and named Jalapeño, last month, while Anthropic has separately been exploring the development of its own chips, according to earlier Reuters reporting from April.

DeepSeek faces significant obstacles specific to its position as a Chinese company, however. U.S. export restrictions bar Chinese chip designers from using the most advanced semiconductor manufacturing foundries located overseas, while separate American export controls have limited China's access to high-bandwidth memory, a component considered essential to building competitive inference chips. DeepSeek founder Liang Wenfeng acknowledged in a 2024 interview that these export restrictions presented genuine obstacles for the company's broader ambitions.

The chip development news comes as DeepSeek has also been pursuing its first external funding round. According to media reports cited by TradingKey, the company completed a Series A funding round in mid-June, raising approximately 51 billion yuan, or roughly $7.5 billion, at a post-money valuation of around 400 billion yuan. The funding round reportedly employed an unusual transaction structure, with external investor funds directed into a limited partnership managed by Liang rather than invested directly into the DeepSeek entity itself, and with those investors receiving no voting rights or board representation in exchange for their capital.

DeepSeek has moved quickly to put that capital to use. The company posted a large recruitment call in late June covering 33 positions across seven categories, including algorithms, research and development, and data engineering, stating it was working to at least double the size of all its departments. The company has also begun building out its own computing infrastructure, posting its first data center-related job listings in Ulanqab, Inner Mongolia, earlier this year, a shift that analysts described as moving DeepSeek away from a lighter-asset research and development model toward a more capital-intensive approach involving self-built computing clusters.

Nvidia's decline Tuesday came amid a broader pullback across chip and technology stocks tied to a separate development: blowout second-quarter earnings guidance from Samsung Electronics that nonetheless triggered profit-taking across the memory chip sector. Micron Technology fell roughly 4.7 percent to 7 percent during the session, while Western Digital and SanDisk also posted significant declines, according to multiple market reports. Nvidia's own decline was compounded by a separate report indicating that some U.S. companies have been exploring cheaper Chinese AI models as alternatives to offerings from major American technology firms, adding further pressure to sentiment around Nvidia and other companies central to the AI infrastructure buildout.

Nvidia's stock has underperformed some of its chip-sector peers so far in 2026, according to a technical analysis published by BeInCrypto. The stock lost the $200 level on June 23 and has not reclaimed it since, with the analysis identifying $189 as a key support level within the stock's broader trading channel. According to the same analysis, strong AI spending guidance from major cloud computing companies later in July, or a significant new agreement involving China, could help push Nvidia shares back above $200, while continued erosion of confidence in the broader AI trade could see the stock fall further below its current trading range.

Nvidia remains up approximately 4.9 percent year-to-date as of Monday's close, according to data cited by Moomoo, even as the stock has faced repeated bouts of volatility throughout 2026 tied to shifting sentiment around AI infrastructure spending, export policy developments affecting China, and competitive dynamics within the broader semiconductor industry. With DeepSeek's chip development effort still in its early stages and facing significant regulatory and technical hurdles, analysts have generally cautioned that any meaningful competitive impact on Nvidia's business would likely take years to materialize, even as Tuesday's report added to a broader narrative of AI companies worldwide seeking greater independence from Nvidia's dominant position in AI computing hardware.