Recently, the AI sector was jolted by the launch of a new open-source AI model from the Chinese startup DeepSeek. This new model, which reportedly cost only $6 million to develop, is capable of matching the performance of top-tier AI models from leading U.S. tech firms. Though there are doubts about the accuracy of the reported training costs, the capabilities of the DeepSeek model seem to be legitimate.
While the emergence of affordable, potent AI models is beneficial for companies looking to implement AI technologies, it poses a significant threat to Nvidia (NASDAQ: NVDA). Nvidia has long dominated the market for high-performance AI accelerators, which are essential for training sophisticated AI models. The company’s growth strategy is based on the expectation that newer AI models will continuously require more computational power.
The uncertainty introduced by DeepSeek’s innovation led to a dramatic drop in Nvidia’s stock value this Monday, erasing billions from the company’s market capitalization in a single day.
Further Developments Awaited
However, the existence of an economical AI model that rivals those developed by Open AI and Anthropic doesn’t alone threaten Nvidia’s ongoing success story. If these cost-effective and efficient AI models lead to increased adoption, they could actually boost the demand for AI accelerators in the future. As Pat Gelsinger, the former CEO of Intel, remarked on X, “Computing obeys the gas law. Making it dramatically cheaper will expand the market for it.” If these AI models can operate without needing large clusters of advanced GPUs, it might not necessarily spell trouble for Nvidia if the overall market for AI expands significantly.
Yet, another significant challenge is emerging. Nvidia’s valuation, which stretches into trillions, hinges on another critical assumption: that AI models will continue to grow in capability as more computing resources are devoted to them. This assumption held true in the early stages of AI development, but signs are showing that it might not hold much longer.
AI firms are beginning to max out the data available for training AI models, and the performance improvements in new AI models are plateauing. The founders of Andreessen Horowitz highlighted last year that enhancement rates are slowing, suggesting that AI models may be reaching a performance ceiling.
It’s crucial to understand that large language models (LLMs) simply predict the next token in a sequence of tokens—they don’t actually ‘think’. Despite this, the performance of these models has been impressive, but there could be a limit to how much they can improve. The combination of affordable AI models and this potential capability ceiling could spell disaster for Nvidia. If AI models cease to improve despite increased computing power, and if a high-quality model can be trained on less expensive hardware, the demand for Nvidia’s costly AI accelerators could plummet as the AI market cools down.
Even in such a scenario, AI will remain a valuable and transformative technology for businesses that effectively leverage it. However, Nvidia’s revenue stream from GPU sales could be severely impacted, and the massive investments tech giants are making in AI data centers might not yield expected returns.
A Volatile Investment
Despite its current valuation exceeding $3 trillion and its significant sales of data center GPUs, Nvidia’s market price is based on precarious assumptions. These include the beliefs that AI models will always need more power and that there is no limit to their performance enhancements.
The emergence of DeepSeek’s inexpensive AI model introduces a substantial risk to Nvidia’s growth narrative. Alone, it may not be enough to topple the AI giant, but combined with the real possibility that AI capabilities are nearing their peak, it undermines the justification for Nvidia’s astronomical market valuation.
I cannot foresee the future, and neither can you, but the growing uncertainties in AI development make Nvidia a highly risky investment at this time.
Timothy Green holds a position in Intel. The Motley Fool owns shares of and recommends Intel and Nvidia. The Motley Fool recommends the following options: short February 2025 $27 calls on Intel. The Motley Fool has a disclosure policy.
Is Investing $1,000 in Nvidia Right Now a Good Idea?
Advice from the Motley Fool: Before you decide to invest in Nvidia, here’s something to consider:
The Motley Fool Stock Advisor team has recently identified what they believe are the ten best stocks for investors to buy right now… and Nvidia isn’t one of them. The selected stocks are expected to generate impressive returns in the upcoming years.
Reflect on when Nvidia was recommended on April 15, 2005… if you had invested $1,000 based on that recommendation, your investment would now be worth $790,519!*
Stock Advisor offers a straightforward plan for success, including portfolio-building advice, regular updates from analysts, and two fresh stock picks each month. Since 2002, the Stock Advisor service has more than quadrupled the S&P 500’s returns.
Learn more »
Passionate about analyzing economic markets, Alice M. Carter joined THE NORTHERN FORUM with a mission: to make financial concepts accessible to everyone. With over 10 years of experience in economic journalism, she specializes in global economic trends and US financial policies. She firmly believes that a better understanding of the economy is the key to a more informed future.
