AI Stocks Rebound as DeepSeek Threat Fades 20%

Jim Cramer, a prominent financial analyst, stated on Tuesday that the Chinese startup DeepSeek no longer poses a significant threat to the tech giants currently leading in the artificial intelligence space. He noted that investors who had previously abandoned stocks in companies such as Nvidia, AMD, Vertices, Microns, and Marvell Technologies are now returning to these investments.

DeepSeek had released an AI model in January that initially seemed as advanced as its competitors, requiring far less money and energy. This news caused a panic on Wall Street, as investors feared that Big Tech companies had overspent on building data centers and AI models. Consequently, many companies in the industry experienced a drop in their stock prices.

Cramer argued that the recovery of many AI stocks indicates that fears about DeepSeek’s dominance were short-lived. He suggested that the rally also refutes the idea that China had surpassed the U.S. in the AI arms race. “Now, looking back, with so much of tech bordering on new highs, it’s clear that these stocks never should’ve been sold in the first place. Because DeepSeek simply wasn’t that meaningful,” Cramer stated.

The American TV personality mentioned that Wall Street’s DeepSeek panic is a “microcosm” of the kind of situations that cause investors to make mistakes and unnecessarily write off certain stocks. Cramer believes many investors did not question DeepSeek’s assertions, even as some experts warned the data could be misleading.

An unnamed senior State Department official revealed that the Chinese AI startup sought to use Southeast Asian shell companies to obtain high-end Nvidia chips, including H100 chips, which cannot be shipped to China under U.S. rules. The official’s comment showed that DeepSeek’s fast-growing AI capabilities were exaggerated, as the company still relied heavily on U.S. technology.

The official also warned that the Hangzhou-based AI firm had provided user information to the Chinese government but had declined to comment on whether the U.S. would implement further export controls or sanctions against the Chinese AI startup. An Nvidia spokesperson confirmed that the company’s review indicates that the Chinese AI startup used lawfully acquired H800 products, not H100.

A study published by DeepSeek’s researchers revealed the firm used 2,048 Nvidia H800 chips to train its DeepSeek-V3 large language model (LLM). According to the report, DeepSeek-V3 required 2.788 million H800 GPU hours for its complete training, meaning the total training time was about 56.7 days. The Chinese AI company also claimed that the training cost for its AI model was only $58 million. Meta spent $500 million to train its Llama 3.1. A group of DeepSeek researchers revealed in a paper on January 22 that DeepSeek-R1’s training used the data from Alibaba’s Tongyi Qianwen (Qween) and Llama.

Commerce Secretary Howard Lutnich said the Chinese AI company could create “dirt cheap” AI models by purchasing many Nvidia chips and stealing data from Meta’s open platform. In October 2022, the Biden administration banned the export of Nvidia’s A100 and H100 chips to China. In October 2023, it also banned the exports of the A800 and H800 chips to China.