Regarding this AI stock, a case can be made for each, but there is one right answer.
Following its massive run over the past five years, perhaps no stock captures as much investor attention as Nvidia (NVDA 3.96%). The company has become the exemplar for the artificial intelligence (AI) infrastructure buildout, and no company has benefited from AI as much as Nvidia.
Given that, it is unsurprising that many investors wonder if Nvidia stock is a buy, sell, or hold. Let’s look at the cases for each.
The buy case for Nvidia
The buy case for Nvidia centers around the AI infrastructure buildout still being in its early days despite the massive demand the company has already seen for its graphics processing units (GPUs). And there is solid evidence to believe this is the case.
As large language models (LLMs) become more sophisticated, they need more and more computing power — which is provided by GPUs — to be trained. This can be seen in newer LLMs not only needing more GPUs for training than their predecessors, but exponentially more.
For example, xAI’s Grok-3 AI model is set to train on 100,000 GPUs compared to Grok-2 being trained on 20,000 GPUs, while Meta Platforms has said its Llama 4 LLM would likely need 10 times the compute power as the Llama 3, which was trained on 16,000 GPUs.
Meanwhile, capital expenditure (capex) on AI projects continues to rise, and the CEOs of both Alphabet and Meta have said that they think the biggest risk is underinvesting in AI. Meanwhile, Oracle CTO and Chairman Larry Ellison said on his company’s most recent earnings call that he doesn’t see spending on AI training slowing down over the next five to 10 years. As such, the comments from Nvidia’s customers suggest that we are still in the early innings for the AI infrastructure buildout.
Nvidia, meanwhile, has carved itself out as the dominant player in the GPU space due to its CUDA software being the de facto program with which developers have learned to program GPUs. At the same time, it has been at the forefront of innovation in the space and is looking to sell chips as well as entire GPU servers. The company has also accelerated its development cycle on the next generation of chips from once every two years to once a year, which should help it maintain its technological lead and pricing power.
And as the cherry on top, Nvidia’s stock is relatively cheap, trading at a forward price-to-earnings ratio (P/E) of only about 29 based on next year’s analyst estimate, and a price/earnings-to-growth ratio (PEG) under 0.8. A PEG under 1 is generally considered undervalued, and growth stocks generally have PEGs well above 1.
NVDA PE Ratio (Forward 1y) data by YCharts
The sell case for Nvidia
The sell case for Nvidia is that demand for GPUs and AI infrastructure is elevated. Once the initial euphoria is over, its sales could come crashing down from current levels. There is less computing power needed for AI inference compared to AI training. If AI models reach the point of being deemed good enough, then there could be a spending shift going more toward AI inference than training, and fewer GPUs would be needed.
And while companies currently spend a tremendous amount of money building out infrastructure to handle AI applications, these companies need to be able to make money on their investments. Nvidia can’t be the only AI winner; that needs to extend to its customers and its customers’ customers as well. Otherwise, spending on AI will dramatically slow.
So far, hyper scalers such as Microsoft and Alphabet have seen nice boosts in their cloud computing segments, although software and hardware companies have been a bit of a mixed bag. Many software companies at the forefront of AI have seen modest AI sales uplifts, such as Adobe and Microsoft. Yet it hasn’t been dramatic, while there have been some initial worries over iPhone 16 sales, which were supposed to give Apple a lift due to the model’s integration of AI features.
Image source: Getty Images.
The hold case for Nvidia
If you already own Nvidia, there is a good case to be made to just hold the stock. In all likelihood, unless you just bought the stock recently, you probably have some nice gains that are likely to continue growing.
Nvidia still has some potential upside ahead given its reasonable valuation and the AI boom looking like it could be in the early innings. In that case, while it is best to manage your position and make sure it hasn’t become too large of your overall portfolio, investors likely would be well served to continue to hold the stock.
The verdict
While there is certainly a risk that spending on AI infrastructure will slow, the comments and actions from Nvidia’s customers suggest that this does not appear likely to happen soon. As such, given its valuation and the growth prospects in front of the company, I would consider Nvidia stock a buy.
Randi Zuckerberg, a former director of market development and spokeswoman for Facebook and sister to Meta Platforms CEO Mark Zuckerberg, is a member of The Motley Fool’s board of directors. Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool’s board of directors. Geoffrey Seiler has positions in Alphabet. The Motley Fool has positions in and recommends Adobe, Alphabet, Apple, Meta Platforms, Microsoft, Nvidia, and Oracle. The Motley Fool recommends the following options: long January 2026 $395 calls on Microsoft and short January 2026 $405 calls on Microsoft. The Motley Fool has a disclosure policy.