Recently, the artificial intelligence sector experienced significant disruption following the launch of an open-source AI model by the Chinese start-up DeepSeek. This model appears to rival top U.S. AI models while reportedly costing only $6 million to develop. Though there are doubts about the accuracy of these cost claims, the DeepSeek model seems to be genuinely competitive.

While affordable, high-performance AI models could benefit firms adopting artificial intelligence, they cast a shadow over Nvidia (NASDAQ: NVDA). The bullish outlook for Nvidia, which leads the market in powerful AI accelerators essential for training advanced AI systems, hinges on the belief that each new generation of AI will demand increasingly more computational resources.

The emergence of DeepSeek’s model raises many questions. This uncertainty contributed to a dramatic decline in Nvidia’s stock on Monday, erasing hundreds of billions of dollars in total market value.

Impending changes

The availability of an affordable AI model that can compete with the leading creations from OpenAI and Anthropic alone is not sufficient to derail Nvidia’s growth narrative. Lower-cost and efficient AI models might boost demand and lead to greater sales of AI accelerators in the future. Pat Gelsinger, former CEO of Intel, mentioned on X that “Computing follows the gas law. Making it significantly cheaper will broaden its market.” If affordable AI models can operate without extensive, high-end graphics processing units (GPUs), it may actually benefit Nvidia by increasing overall AI engagement.

However, another issue lurks ahead. Nvidia’s valuation, which stands in the trillions, is built on a crucial assumption: that AI models will continue to improve with increased computing resources. This notion was valid in the early days of AI but may no longer hold true.

AI разработчики have largely depleted the datasets used for training, and newer AI models are not showing significant performance advancements. Founders at Andreessen Horowitz indicated late last year that progress is slowing, suggesting AI models might be nearing their performance limits.

It’s vital to understand how large language models (LLMs) operate. These models operate by predicting the next token in a sequence — that’s essentially their function. They don’t engage in reasoning; it’s merely a semblance of it. Although new methodologies could lead to better performance, it stands to reason that there will be a cap on the abilities of this kind of AI model.

The combination of low-cost AI models and the potential plateau in capabilities could spell doom for Nvidia’s stocks. If AI models cease to make noteworthy improvements regardless of computational power and a top-tier model can be trained inexpensively on inferior hardware, that would signify the end of the game. Demand for Nvidia’s premium AI accelerators would likely plummet as the AI inflation bubble bursts.

Even in such a scenario, AI will retain its usefulness and transformative potential for organizations that utilize it effectively. However, Nvidia’s GPU revenue stream would likely dry up, and technology giants investing vast sums into AI data centers may find it impossible to recover those expenses.

High-stakes investment

Currently valued at over $3 trillion, Nvidia continues to sell its high-priced data center GPUs and generate impressive profits. However, this lofty evaluation rests upon two assumptions that are increasingly questionable: first, that AI models will necessitate continually growing computational power to function, and second, that there isn’t an upper limit to their capabilities.

DeepSeek’s affordable AI model signifies a significant chink in the armor of Nvidia’s growth narrative. It alone isn’t enough to topple the reigning AI giant. Yet, when combined with the tangible possibility that AI models may not significantly enhance their capabilities moving forward, the foundation for Nvidia’s elevated market value crumbles.

While the future remains uncertain and unpredictable for all, the volatility surrounding the AI sector renders Nvidia a highly risky investment.

Timothy Green has investments in Intel. The Motley Fool has stakes in and endorses Intel and Nvidia. The Motley Fool suggests shorting February 2025 $27 calls on Intel. The Motley Fool maintains a disclosure policy.

The Motley Fool partners with YSL News, providing financial news, insights, and commentary aimed at empowering people to take charge of their financial futures. Its content is created independently of YSL News.

Is it a good time to invest $1,000 in Nvidia?

From the Motley Fool: Before investing in Nvidia, consider the following:

The Motley Fool Stock Advisor team has pinpointed what they believe are the top 10 stocks to buy now, and Nvidia isn’t included. The ten selected stocks have the potential for substantial profits in the upcoming years.

For context, if you had invested $1,000 in Nvidia on April 15, 2005, when it was first recommended, your investment would now be worth $790,519!

Stock Advisor offers investors a straightforward pathway for success, featuring advice on portfolio creation, regular analyst updates, and two new stock recommendations every month. Since 2002, the Stock Advisor service has more than quadrupled the returns of the S&P 500.