In January, DeepSeek, a Chinese start-up, launched an AI chatbot that quickly became the most downloaded app in the U.S.
The model reportedly cost $6 million to train and outperforms some top U.S. models. This shocked Wall Street, causing Nvidia (NVDA) shares to drop 17%, losing $600 billion in market value.
Though Nvidia’s stock is still down by 9% since then, investors received good news from Amazon (AMZN) and Alphabet (GOOGL).
Some analysts are skeptical of DeepSeek’s $6 million training cost. Experts like Dan Ives from Wedbush Securities doubt the claim, stating it may not be accurate. SemiAnalysis reports that DeepSeek used next-gen Nvidia GPUs and spent $1.6 billion on training.
If training costs drop, it could increase demand for Nvidia GPUs. Cheaper AI training could lead more companies to develop AI products, driving demand for Nvidia’s hardware.
Amazon and Alphabet plan large investments in AI infrastructure. Amazon’s CEO, Andy Jassy, believes AI model costs will continue to fall, making it easier for companies to add AI to their apps. Amazon’s capital spending for 2025 is expected to exceed $100 billion, up from $83 billion in 2024.
Alphabet’s CEO, Sundar Pichai, said AI spending would rise as companies focus on running AI models rather than training them. Alphabet plans to spend $75 billion in 2025, up from $52 billion in 2024.
These moves show that Nvidia’s GPUs remain in high demand, even as AI training costs drop.
Despite recent stock drops, Wall Street predicts Nvidia will grow its earnings by 52% annually through 2026. Analysts still see Nvidia’s stock as a good investment at its current valuation.
Despite concerns over DeepSeek, efficient AI training methods will likely increase demand for Nvidia GPUs. Investments from Amazon and Alphabet support this outlook, offering a positive future for Nvidia.