Nvidia Corporation shares are trading lower following reports suggesting China’s DeepSeek is developing AI chips for inference, which reduces its reliance on the company. The potential shift in strategy by a major AI player poses a risk to Nvidia's dominance in the semiconductor market, as inference chips are critical for running trained models. Nvidia stock was down 1.62% at $192.39 at the time of publication.

DeepSeek’s Quiet Push Into Semiconductors

According to Reuters, DeepSeek is developing its own AI chip designed for inference — the stage of AI computing in which a trained model generates responses for users — rather than for training new models. The effort began approximately a year ago and remains at an early stage, with DeepSeek reaching out to external chip-design, foundry and memory companies. The company has also quietly increased hiring of chip-design engineers in recent months without posting public job listings.

If successful, the move would mark a major strategic shift for DeepSeek — widely regarded as China’s AI champion — and could reduce its reliance on both Nvidia and Huawei chips, which it has historically depended on to train and run its globally popular models.

The Broader Context

DeepSeek would be joining a growing list of AI companies seeking to reduce dependence on Nvidia by developing custom silicon. OpenAI last month unveiled Jalapeño, its first custom inference chip developed with Broadcom, while Anthropic has been weighing building its own chips, Reuters reported in April.

Nvidia Corp. on Monday rejected a report claiming its next-generation Kyber AI server platform has been delayed by a year, stating that its product roadmap remains intact. The denial follows assertions by semiconductor research firm SemiAnalysis that the Kyber system faced "massive delays" due to design-related manufacturing challenges, allegedly pushing its launch from 2027 to 2028. A company spokesperson told Yahoo Finance, "Our roadmap remains intact," directly contradicting the report's central claim without providing further elaboration.

The Kyber platform is expected to debut alongside Nvidia's Vera Rubin Ultra platform in the second half of 2027. The new server design is intended to double GPU density by arranging server racks vertically, allowing up to 144 GPUs per server compared with 72 in current-generation systems. Any delay to the platform could provide rivals, such as Advanced Micro Devices, Inc., additional time to narrow the gap in AI server hardware as demand for high-performance AI computing surges.

Analyst Reaction

Market observers offered differing interpretations of Nvidia's brief statement. CNBC's Jim Cramer wrote on X, "Nvidia says its roadmap intact. That, to me, means buy." Conversely, Futurum Group CEO Daniel Newman took a more measured view, noting that the phrase "roadmap is intact" leaves "some room for interpretation." He emphasized Nvidia's fiduciary responsibility to shareholders and the legal risks of misleading statements, suggesting the company likely considered its wording carefully.

Patrick Moorhead, CEO of Moor Insights & Strategy, also weighed in, describing the statement as generally encouraging and writing that it "sounds positive."

Market Performance

Nvidia shares closed 0.37% higher at $195.55 on Monday. The stock subsequently slipped 0.43% to $194.70 in after-hours trading. According to Benzinga Edge Rankings, Nvidia ranks in the 98th percentile for Growth, highlighting its strong long-term performance despite recent softening in short- and medium-term price trends.