DeepSeek R2 model with lower training costs may challenge Nvidia, potentially impacting market

From Nasdaq: 2025-05-02 00:51:00

In late January, Chinese AI research lab DeepSeek released its open-source and efficient AI model, DeepSeek-R1, causing a temporary 17% drop in Nvidia stock. The R1 model matched top models while being cost-effective. Now, news of the upcoming R2 model, with a hybrid Mixture-of-Experts architecture and lower training costs, raises questions about Nvidia’s future.

DeepSeek’s innovative models prioritize cost-efficient optimization techniques over hardware, leading to significantly lower costs compared to traditional large language models. DeepSeek-R1 demonstrated competitive performance with minimal computing power, offering API pricing at $0.55 per million input tokens and $2.19 per million output tokens, undercutting OpenAI. The upcoming R2 model, utilizing Huawei chips, could further challenge Nvidia’s dominance in the AI market.

As companies adopt DeepSeek’s cost-efficient model development approach, Nvidia faces potential cooling demand for AI computing power. While the use of Huawei chips by DeepSeek could diversify demand away from Nvidia, leaner and efficient models might also democratize AI access, expanding total demand. Trefis values Nvidia stock at $101 per share, slightly below the current market price, amidst evolving industry dynamics.



Read more at Nasdaq: Should DeepSeek R2 Worry Nvidia Investors?