In This Story
Nvidia (NVDA-5.05%) addressed its tumbling stock on Monday by saying Chinese artificial intelligence startup DeepSeek will need more of its chips for new models.
The chipmaker’s shares were down by more than 17% toward the end of the trading day on Monday after being down most of the day amid a global sell-off of tech stocks sparked by the release of China’s DeepSeek-R1 reasoning models last week. Nvidia chief executive Jensen Huang saw his net worth plunge by $18 billion as a result of the stock rout.
“DeepSeek is an excellent AI advancement and a perfect example of Test Time Scaling,” an Nvidia spokesperson said in a statement shared with Quartz. “DeepSeek’s work illustrates how new models can be created using that technique, leveraging widely available models and compute that is fully export control compliant. Inference requires significant numbers of NVIDIA GPUs and high-performance networking.”
According to the Hangzhou-based startup, DeepSeek-R1 demonstrated a comparable performance to OpenAI’s reasoning models, o1-mini and o1, on several industry benchmarks — for a fraction of the cost and energy.
DeepSeek’s seemingly efficient and competitive models could challenge Nvidia’s business, which relies on major AI firms such as OpenAI, Meta (META+0.60%), and Google (GOOGL+0.33%) spending billions of dollars on its advanced chips.
In December, DeepSeek released its V3 model, which it said used a cluster of just under 2,050 graphics processing units (GPUs) from Nvidia for training — much less than the tens of thousands of chips U.S. firms are using to train similarly-sized models. Meta, for example, used 16,000 of Nvidia’s more powerful H100s to train its Llama 3 405B model.
Aside from prompting concerns about high AI chip spending, DeepSeek’s success also calls into question U.S. efforts to curb advanced chips from entering China.
The Chinese firm trained and developed DeepSeek-V3 with Nvidia’s H800 chips, according to a technical report, which have reduced capabilities compared to chips used by OpenAI, Meta, and other U.S. competitors. Nvidia is allowed to sell the less-powerful version of its H100 chips to Chinese firms under U.S. chip restrictions.