NVIDIA stock slides as AI-related capex faces questions post China’s DeepSeek launch
DeepSeek R1 is a large-language model that is seen as rival to ChatGPT and Meta while using a fraction of their budgets.
- Jan 27, 2025,
- Updated Jan 27, 2025 9:18 AM IST
Shares of NVIDIA Corporation fell over 3% on Friday as questions arise on the need for major capital expenditure on artificial intelligence after the release of China’s DeepSeek.
Nvidia stock fell 3.58% to a low of $141.88 in the previous session on Nasdaq against a close of $147.15 on January 24. Later, the stock closed 3.12% lower at $142.62.
DeepSeek R1 is a large-language model that is seen as rival to ChatGPT and Meta while using a fraction of their budgets.
DeepSeek is funded by Chinese quant fund High-Flyer. Reportedly, it had access to about 50,000 of Nvidia’s H100 AI GPUs, which are from the last generation of advanced AI chips.
DeepSeek claims R1 matches—and in some cases surpasses—ChatGPT in areas like mathematics and coding while being significantly more cost-effective.
Unlike previous Chinese AI models, which often followed a US-led blueprint, R1 is an innovative leap. It uses a hybrid architecture and a “chain of thought” reasoning method to break down complex problems step by step—similar to how GPT models operate but with a focus on greater efficiency.
Shares of NVIDIA Corporation fell over 3% on Friday as questions arise on the need for major capital expenditure on artificial intelligence after the release of China’s DeepSeek.
Nvidia stock fell 3.58% to a low of $141.88 in the previous session on Nasdaq against a close of $147.15 on January 24. Later, the stock closed 3.12% lower at $142.62.
DeepSeek R1 is a large-language model that is seen as rival to ChatGPT and Meta while using a fraction of their budgets.
DeepSeek is funded by Chinese quant fund High-Flyer. Reportedly, it had access to about 50,000 of Nvidia’s H100 AI GPUs, which are from the last generation of advanced AI chips.
DeepSeek claims R1 matches—and in some cases surpasses—ChatGPT in areas like mathematics and coding while being significantly more cost-effective.
Unlike previous Chinese AI models, which often followed a US-led blueprint, R1 is an innovative leap. It uses a hybrid architecture and a “chain of thought” reasoning method to break down complex problems step by step—similar to how GPT models operate but with a focus on greater efficiency.