Artificial intelligence (AI) stocks have driven gains in this bull market, and one company in particular has been at the head of the pack: Nvidia (NVDA 1.35%). The tech powerhouse is the world’s top designer of AI chips and has built an empire of related products and services — importantly, this portfolio has produced double- and triple-digit revenue growth well into the billions of dollars quarter after quarter.
And, once again, in the company’s latest earnings report, Nvidia showed it could keep up the momentum. The chip leader reported record quarterly and full-year revenue — and earnings roared past analysts’ estimates. All of this is great, but investors in recent times have wondered how long Nvidia’s showstopping growth actually may last. For example, when start-up DeepSeek recently announced it trained its model on low-cost chips, investors worried Nvidia’s biggest customers could follow — and that would lead to lower revenue for the chip designer down the road.
But investors may not have to worry about that because Nvidia chief Jensen Huang just delivered some fantastic news. Let’s check it out.
Image source: Getty Images.
Powering key AI tasks
First, though, here’s a bit of background on Nvidia’s path so far and how that’s translated into earnings growth. The company dominates the AI chip market, designing the fastest graphics processing units (GPUs) to power key tasks such as the training and inferencing of models. Nvidia also has expanded into areas such as software and networking, offering customers a full suite of products to help them along the AI development path.
As mentioned, this has lifted earnings over the past several years. As the world’s biggest tech companies rush to build up AI infrastructure and programs, they’ve loaded up on the most powerful tools — and these are found at Nvidia. The AI giant said in the recent earnings call that major cloud service providers represent about half of the company’s data center business, and revenue from these customers is two times higher than it was in the year-earlier period.
All of this has helped the momentum continue in the fourth quarter and fiscal 2025 full year. Nvidia reported record quarterly revenue of more than $39 billion, a 78% increase year-over-year. And annual revenue surged 114% to more than $130 billion — another record. As it’s done quarter after quarter, Nvidia’s earnings beat analysts’ estimates, with earnings per share coming in at 89 cents versus expectations of 84 cents and revenue surpassing the $38 billion forecast.
Huang’s big news
Now, let’s consider this fantastic news delivered by Huang — and this should dispel concerns about companies turning to cheaper training solutions. Nvidia in the quarter launched its new Blackwell architecture, a customizable platform featuring seven different chips, many networking options, and more. In this first quarter on the market, Blackwell brought in $11 billion in revenue and represented the company’s fastest production ramp ever.
But the even bigger news is this platform is well-suited to a particular task that should keep customers coming back to Nvidia: inferencing. And with a focus on “reasoning,” the inferencing of today and tomorrow powers the processing of information and the “thinking” step that allows the large language model (LLM) to offer answers to complex questions. The amount of compute needed for this long-thinking reasoning is 100 times more than the compute needed by LLMs for responses in the earlier days of AI.
“The vast majority of our compute today is actually inference, and Blackwell takes all of that to a new level,” Huang said during the company’s earnings call. “We designed Blackwell with the idea of reasoning models in mind.”
As a result, Blackwell offers as much as 25 times higher token throughput — tokens are bits of data processed — and 20 times lower cost versus its predecessor, Hopper.
Why customers will stick with Nvidia
The DeepSeek news late last month set off a decline in Nvidia stock as investors worried about a potential flight of customers to cheaper training options. The stock slipped nearly 8% from that time through Feb. 26.
But, considering tech giants’ commitments to AI spending and desire to win in AI over time, it’s likely they’ll stick with the market’s strongest tools — that means they probably won’t switch to low-cost chips for their projects. A few weeks ago, Nvidia customer Meta Platforms, for example, said the company will spend as much as $65 billion in infrastructure this year and will end the year with 1.3 billion GPUs.
And Huang once again emphasized the enormous demand for Blackwell in the quarter, calling it “extraordinary.”
On top of this, Nvidia’s dominance in inference — or the actual performance of LLMs — reinforces the idea that the company is not set to lose business. In fact, it’s likely heading for yet another wave of major growth ahead thanks to its strengths in this key area. And that’s why, especially at today’s valuation of 29x forward earnings estimates, Nvidia stock makes a solid 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. Adria Cimino has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Meta Platforms and Nvidia. The Motley Fool has a disclosure policy.
